ISP PQ Tool User Guide

Overview

This document mainly introduces SpacemiT image tuning, including

• The tuning tool
• Calibration plugins
• Image analysis tool (VRF Viewer)
• Platform debugging support

Abbreviations

Name	Description
ISP	Image Signal Process
VRF / vrf	RAW image with information at the end
BLC	Black Level Correction
LSC	Lens Shading Correction
AWB	Auto White Balance
AEC	Auto Exposure Control
AF	Auto Focus
OTP	One Time Programmable
AEM	Auto Exposure Monitor
AFM	Auto Focus Monitor
CCM	Color Correction Matrix
CT	Color Temperature
BPC	Bad Pixel Correction
CAC	Color Aberration Correction
LTM	Local Tone Mapping
PDC	Phase Detection Compensation
PDF	Phase Detection Correction
PDAF	Phase Detection Auto Focus
SE	Special Effect
EIS	Electronic Image Stabilization
CDAF	Contrast Detection Auto Focus
FV	Focus Value
SAD	Sum of absolute difference
ROI	Region of Interest
NR	Noise Reduction
EE	Edge Enhancement
HDR	High Dynamic Range
Qn	Accuracy, 2n is double

Tuning Tool Overview

Tuning Tool Architecture

PC-Side Tuning Tool Installation

The tuning software is a portable compressed package — no installation is required. Simply extract the file named AsrIspToolVX.X.X.X.rar to use it. The tool can be downloaded from the following link: https://archive.spacemit.com/tools/isp-tunning/

After extraction, the following files are included:

Debugging Environment Setup

Hardware and Software Requirements

• Hardware Environment

  • Desktop or laptop computer
  • 1GHz or faster processor
  • 1GB RAM (32-bit) / 2GB RAM (64-bit)
  • At least 10GB of available hard disk space
  • Screen resolution of 1920 × 1080 or higher
  • USB port
  • Terminal device integrated with ASR ISP

• Software Environment

  • Windows 7 64-bit or later version of the operating system

Device Connection

AsrIspTool connects to the terminal device via USB and communicates with the device through ADB.

Note. Before connecting, the device must first start the tuning server thread, i.e., start the camera.

Basic Operations of the Tuning Tool

Main Interface of the Tuning Tool

Double-click AsrIspTool.exe to launch the tuning tool. The main interface is shown below:

• Menu: Functional Menu Area

  • Open: Open parameter file.
  • Save: Save parameter file.
  • Save As: Save parameter file under a new name.
  • IP Address: Reserved.
  • ADB (SN): Connect to the terminal device via ADB; supports manual input of ADB serial number.
  • Connect: Connect to the terminal device.
  • View: Switch between single-window, horizontal split, and vertical split display modes.
  • Format: Toggle between decimal and hexadecimal display.
  • Display: Switch between matrix editing, row editing, and column editing modes.
  • Plugins: Plugins.
  • Frequency: Adjust parameter refresh rate.
  • Capture: Capture VRF data (vrf).
  • Register: ISP register read/write tool.
  • I2C: I2C read/write tool.
  • Push: Reserved.
  • Transfer: Reserved.
  • VRF: Image viewer tool.
  • DNG: Reserved.

• Module list & Filter list: Module and filter list

• Parameter list: List of parameter

• Log: Log area

Basic Online Operations

Connecting to the Terminal Device

After launching the tool, select ADB (SN) and click Connect. Once the connection is successful, the tool will automatically read the parameters of all current modules and periodically refresh the read-only parameters every 500 ms (this interval can be adjusted via the Frequency option in the menu). If multiple terminal devices are connected to the PC, you can specify the serial number manually.

To enable periodic refreshing of read-write parameters as well, check the AutoUpdate option in the upper-right corner (Note. Parameters cannot be modified while this option is checked).

To perform a one-time read of all parameters, click the Read button in the upper-right corner.

Note. The ADB connection method is only applicable for projects using the Android system. We primarily use TCP network connections to tune the development board.

Parameter Type Description

• Adjustable Parameters

  • Parameters that can be checked, for example, m_bAutoCalculateAEMWindow.
  • Editable parameters, for example, m_nPreEndingPercentage.
  • Editable array parameters, for example, m_pSubROIPermil. For two-dimensional arrays, you can switch between matrix, row, and column editing modes.

• Read-Only (Gray)

  • Read-only parameters, for example, m_nAdjacentLumaSAD.

• Special Notes

  • After modifying parameters in plugins or the parameter list, the changed content will be highlighted in red. Hovering the mouse over it will display the original value.

Real-Time Parameter Modification

1. Expand the module list for the module you want to debug.
2. Click the desired module in the module list area.
3. Adjust the parameter value using the slider or by directly editing the value in the parameter list area. The changes take effect immediately.

Capturing VRF Images

1. Click the Capture button in the menu area.
2. Select RAW and set the save path.
3. Click Start Capturing to generate raw images with the .vrf file extension.

Register Read/Write

1. Click the Register button in the menu area.
2. Set the Address (register address).
3. Set the Value (8bit) (register value).
   • Read: Read the register.
   • Write: Write to the register.
4. Set the Value (32bit) (register value).
   • Read: Read the register.
   • Write: Write to the register.

I2C Read/Write

1. Click the I2C button in the menu area.
2. Set the Device ID (I2C device number).
3. Set the Device Address (slave device address).
4. Set the Address Bytes (register address byte width).
5. Set the Register Address (register address).
6. Set the Value Bytes (register value byte width).
7. Set the Value (register value).
   • Read: Read the register.
   • Batch Read: Batch read registers by importing a file.
   • Write: Write to the register.
   • Batch Write: Batch write registers by importing a file.

Batch Register Read/Write File Format
The file format should be {Address, Value}. When performing batch register operations, click Batch Read or Batch Write to import the reg_batch.txt file. The results will be displayed in the red log area, and a file with the same name ending in _read.txt will be generated for later review.

Example of Batch Register Read/Write File Format

Saving Parameters

1. Click the Save button in the menu area.
2. Choose the path and set a file name.
3. Click Save to generate the parameter file.

Opening a Local Parameter File

Click the Open button in the menu area, or simply drag the parameter file into the corresponding module in the tool. This operation will directly write the parameters to the hardware.

Basic Offline Operations

Open a Local Parameter File

• Click the Open button in the menu area, or drag the parameter file into the tool directly.

Modify Parameters

1. Expand the module list for the module you want to debug.
2. Click the desired module in the module list area.
3. Adjust parameter values using the slider or by directly editing the values in the parameter list.
4. If it is a one-dimensional vector, click the waveform button in the parameter editing interface to enter curve editing mode.

Calibration Plugins

• Click the Plugins drop-down menu in the menu area to select a plugin.

Save Parameters

• Click the Save button in the menu area, enter a file name, and the parameters will be saved to a local file

ISP Plugins

This section introduces calibration and tuning for BLC, LSC, AWB, CCM, Curve, Noise, PDC, and PDAF, as well as auxiliary debugging tools like General Information and Raw Preprocessor.

Calibration plugins support both online（connecting device) and offline（importing paqrameter files) modes. The plugin can only be opened when the corresponding Filter parameters are enabled.

BLC Calibration and Tuning

VRF Image Requirements for BLC Calibration

Capture VRF data in a completely dark environment or with the lens fully covered.

BLC Calibration Steps

The BLC calibration interface is shown below:

1. In the BLC plugin, click Load to import the VRF image.
2. Select the Pipe ID (optional if not a single pipeline).
3. Select the Channel ID.
4. Click Calibrate. The calibration result will be displayed in the Calibrated Result section. If the result is unsatisfactory, you can also manually edit the Result.
5. Click Update to apply the parameters to the parameter list. If the result is unsatisfactory, click Cancel to recalibrate.

Notes on BLC Calibration

• The Calibrated Result panel displays the values of 4 channels in both 10-bit and 8-bit formats. When the parameters are saved to a file, they are mapped to 12-bit values.
• Channel ID: Indicates the BLC parameters corresponding to a specific gain level of 2ᵅ. BLC can be adjusted dynamically with gain, ranging from 1x to 2048x gain, for a total of 12 levels (see the Gain–BlackValue diagram below). The final level, manual, takes effect only when manual mode is enabled; in this mode, BLC does not adjust with gain.

Gain – BlackValue Diagram

Notes on BLC Tuning

BLC parameters are located in CDigitalGainFirmwareFilter.

• If BLC should not change with gain, set m_bManualMode to 1. In this case, the BLC value is taken from m_pGlobalBlackValueManual.
• If BLC should change with gain, set m_bManualMode to 0. In this case, the BLC value is taken from m_pGlobalBlackValue.

LSC Calibration and Tuning

VRF Image Requirements for LSC Calibration

Capture several uniformly illuminated images using a diffuse cover over the lens in a lightbox environment (D65, CWF, or A light) or any environment with shading.

LSC Calibration Steps

The LSC calibration interface is shown below:

1. In the LSC plugin, click Load to import the VRF image.
2. Select the Pipe ID (optional if not a single pipeline).
3. Select the Channel ID.
4. Adjust the compensation ratio using Current Percentage. It is recommended to start with 100%; you can later fine-tune the compensation intensity using strength.
5. Click Calibrate. The simulated correction result will be displayed in Calibrated Image.
6. Click Update to apply the parameters to the parameter list. If the result is not satisfactory, click Cancel to recalibrate.

Notes on LSC Calibration

• Channel ID:
  • 0: Low color temperature compensation table
  • 1: Medium color temperature compensation table
  • 2: High color temperature compensation table
  • manual: Effective when manual mode is enabled; in this mode, LSC does not adjust with color temperature changes.

Notes on LSC Tuning

LSC can be adjusted according to CT or CorrelatedCT (see the CT-LSCProfile diagram below).

• CT Definition: 256 × AWB_RGain / AWB_BGain (can be obtained via CT info / 4 in the AWB plugin).
• CorrelatedCT Definition: Correlated color temperature, representing how closely the light emitted by a source matches the blackbody radiation at a certain color temperature.

LSC parameters are located in CLSCFirmwareFilter.

• If LSC needs to change with color temperature, set an appropriate m_pCTIndex to select the shading table for different color temperatures.
• Recommended to use: m_nCorrelationCT (read from CCTCalculatorFilter).

Note. LSC interpolation can be based on either the CT result calculated by AWBFilter (read CT in AWB plugin) or the CCT result calculated by CCTCalculatorFilter (read m_nCorrelationCT in WbFirmwareFilter)

CCM and CCT Calibration and Tuning

VRF Image Requirements for CCM Calibration

Capture an image of a 24-color chart in a lightbox environment. The color chart should be as centered and aligned as possible, occupying about 1/9 of the frame. D65, CWF, and A light sources are required.

CCM Calibration Steps

The CCM calibration interface is shown below:

1. In the CCM plugin, click Load to import the VRF image. The VRF image should be compensated for LSC and PDF (if PD pixels exist) using the Raw Preprocessor plugin.
2. Select the entire color chart in the image by drawing a box, ensuring all 24 ROIs fall within the color patches. If the image is misaligned or heavily distorted, click Start, check the ROIs you want to adjust individually, and then manually drag the ROIs.
3. Set the desired saturation for calibration.
4. Click Calibrate. The calibration simulation result will be displayed in Calibrated Result.
5. Select the Pipe ID (optional if not a single pipeline).
6. Select the Channel ID.
7. Click Update to apply the parameters to the parameter list. If the result is not satisfactory, you can adjust the saturation of individual blocks in the Saturation Table and then recalibrate.

CCT Calibration Steps

The CCM calibration interface is shown below:

1. The CCT calibration can be performed simultaneously with CCM calibration. CCT requires only A and D65 light sources.
2. After calibrating CCM with A light, select profile 2850K and click UpdateCTMatrix.
3. After calibrating CCM with D65 light, select profile 6500K and click UpdateCTMatrix.
4. The results will automatically update in CCTCalculatorFilter under m_pCTMatrix_low / m_pCTMatrix_high.

Notes on CCM Calibration

• Use Internal Curve – Calibration: No need to check.
• Use Internal Curve – Render: No need to check.
• Use AGTM: Check this option.
• Target: Keep as D50。
• View Environment: Keep as D50。
• Calibrated Result: Displays the simulated result after calibration.
• CCM Result: Lists the calibrated color matrix. This can also be manually edited here. Click Set to apply it to the hardware。
• Channel ID:
  • 0: Low color temperature CCM parameters
  • 1: Medium color temperature CCM parameters
  • 2: High color temperature CCM parameters
  • manual: Effective when manual mode is enabled; CCM does not adjust with color temperature changes.
• Make DNG Profile: reserved
• UpdateCTMatrix: Update the CCT matrix
• SaveImage: Save the rendered image

Notes on CCM Tuning

CCM can be adjusted based on color temperature (see the CCM–Color Temperature Control Curve below).

CCM parameters are located in CColorMatrixFirmwareFilter.

• If CCM needs to change with color temperature, set an appropriate m_pCTIndex to select the color matrix for different color temperatures.
• Recommended to use: CorrelatedCT.

Note. CCM interpolation can be based on either the CT result calculated by AWBFilter (read CT in the AWB plugin) or the CCT result calculated by CCTCalculatorFilter (read m_nCorrelationCT in WbFirmwareFilter).

AWB Calibration and Tuning

VRF Image Requirements for AWB White Point Calibration

No additional images are needed for AWB calibration; it can be performed after completing CCT calibration.

AWB White Point Calibration Steps

The AWB calibration interface is shown below:

1. Open the AWB plugin.
2. Click Optimize. The calibration parameters will be automatically updated in the parameter interface.

AWB Brightness Calibration

After AE tuning is completed, brightness calibration of the module is required to obtain the Lux value needed by AWB. The calibration steps are as follows:

1. Place the camera in the lightbox and capture the lightbox wall with the light source set to D65.
2. Measure the lightbox illuminance using a color temperature illuminance meter and enter the value into m_nCalibSceneLux in the AECFilter.
3. Read m_nExpIndexLong from AECFilter and enter it into m_nCalibExposureIndex in AECFilter.
4. Read m_nLumQ16 from AECFilter and enter it into m_nCalibSceneLum in AECFilter.

Parameter Name	Description	Recommended to Tune	Special Notes
m_nCalibExposureIndex	Exposure index for brightness calibration	Yes
m_nCalibSceneLum	Scene brightness for brightness calibration	Yes
m_nCalibSceneLux	Actual illuminance corresponding to calibration scene	Yes
m_nSceneLux	AWB debug parameter, current scene illuminance calculated by AE	-	Read-Only

AWB Debug

Block Debug Information

When connected to the device, open the AWB plugin. The position of each block will be displayed as a blue dot on the coordinate axis, and the AWB statistics image will be shown as a thumbnail.

You can select a region on the statistics chart (by default, points from all regions are displayed). After selection, only the points within the selected blocks will be shown.

White Points within ROI

Click Show ROI to view the blocks contained within different ROIs. The white blocks are those participating in the white balance calculation, meaning the blocks that fall inside the ROI area. The image below shows the specific affiliation of 32 x 24 blocks to their respective ROIs

Block Weight

Checking auto update will periodically refresh the lux value and statistics chart, and calculate the block weights in real-time.

By adjusting the Weight Percentage slider, you can control the ratio between the actual scene and the blocks participating in the white balance calculation on the debug image.

• At 0%, the image shows the actual scene.
• At 100%, it shows the weights of blocks participating in the white balance calculation (weights are referenced in the heatmap).

If set to 100%, the screen may appear completely black, indicating that all blocks have zero weight under the current lux.

Setting Weight Percentage to 100% displays the block weights as a heatmap. You can hover the mouse over any block to see its weight displayed on the right side of the heatmap (debug info is also available in AWB Frameinfo).In the image below, the mouse is selecting block[12][2], which has a weight of 16.

White Balance Gain Position

In the RGB gain panel module, check enabled, and enter RGB gain values in Q12 precision (from debug info CurrentResult). The corresponding white balance gain will be displayed as a blue square in the chromaticity (color temperature) coordinate system. You can zoom in by dragging the mouse from the top-left to the bottom-right, and zoom out by dragging from the bottom-right to the top-left.

The current white balance gain is shown as a red square in the chromaticity coordinate system.

AWB Debug Description

• Calibration Panel

  • Calibrate Panel
    • Visible: When checked, this ROI will be shown in the chromaticity coordinate system.
    • Enable: When checked, this ROI is enabled for calibration.
  • Calibrate Files Panel
    • Percentage: Proportion of the VRF image to be used for calibration. For example, 20% means the central 20% area of the image will be used, as the center is less affected by shading.
    • Optimize: Performs automatic calibration.
    • Load config: Reserved.
    • Save config: Reserved.
    • Load: Import VRF file.
    • Calibrate: Reserved.
    • Update: Update parameters to the parameter list.
    • Cancel: Cancel parameter updates.
    • ShowROI: Show the white points within each ROI.

• Debugging Panel

  • Control Panel

    • Pipe ID: Current pipeline ID (selectable when not using a single pipe).
    • Auto Update: Automatically updates the current brightness and statistics window when online (when checked, parameters in the plugin cannot be modified).
    • Manual Lux: Fixes the current brightness.
    • Weight Percentage: Debug parameter to adjust the display of white balance statistic blocks and their weights.
      • 0% shows the statistic blocks;
      • 100% shows a heatmap of block weights.
    • RGB Gain Debug-applied gain, corresponding to the red point in the chromaticity coordinate system.
    • Luminance Boundary: Brightness range (8-bit) of pixels participating in AWB statistics.
    • Valid Number: Minimum number of valid blocks required for white balance calculation, range [0,768]。
    • Correct Limit Panel: Block range limits. Blocks that exceed the limits and fall within the ROI will be remapped in the XY direction.

  • Green Shift Panel

    • Shift Max Weight: Shift weight; multiplied by exposure shift to get the final shift weight. Maximum is 32, which fully biases toward the outdoor gain.
    • Green Number Threshold: Threshold: The number of blocks falling into the "G region"; if within this range, green shift is activated.
    • Outdoor RGB Gain: Target gain for green shift.。
    • Shift Weight: Weight adjusted based on exposure.

  • Luminance Panel

    • Lux: Brightness index.
    • Weight: The weight of the corresponding ROI at the current brightness level.
    • Min / Max: Luminance range (8-bit).

• Debug Panel

  • RGB Gain Panel

    • enabled: Used to display the white balance gain position in the color temperature coordinate system.
    • RGB: White balance gain.

  • Gain Panel: Debug information. Hovering the mouse over the color temperature coordinate system will display the corresponding debug info.

    • X Y: The XY coordinates of the point on the color temperature coordinate system.。
    • CT: The correlated color temperature (CT) value at the point, which can be used as a reference for LSC and CCM interpolation.
    • RGB: The white balance gain at the point on the color temperature coordinate system.
    • CCT、Tint: The correlated color temperature and tint at the point, also usable for LSC and CCM interpolation.
    • CCT curve: Displays the CCT curve on the color temperature coordinate system.
    • Vaild only: Displays only the valid statistic blocks on the color temperature coordinate system.
    • Applied Gain: The white balance gain point of the current scene, shown as a red dot on the color temperature coordinate system.。
    • Block、Weight: When hovering the mouse over the statistics image, the corresponding block position and weight are displayed.

Curve Debugging

Curve Debugging Steps

The Curve debugging interface is shown below:

1. Open the Curve plugin.
2. Select Pipe ID (applicable if not using a single pipeline).
3. Select Channel ID.
4. Move the mouse to the point on the curve you want to adjust, then left-click and drag it to the desired position.
5. Click Update to apply the parameters to the parameter list. If the result is unsatisfactory, click Cancel to redo the calibration.

Curve Debugging Description

• Channel ID:
  • BacklightCurveManual: Controls the background brightness. It is recommended to keep the curve unchanged and only adjust the strength.
  • ContrastCurveManual: Controls the contrast. It is recommended to keep the curve unchanged and only adjust the strength.
  • GTMCurve0: Curve used when the gain equals m_pGainIndex[0] (Q4 precision).
  • GTMCurve1: Curve used when the gain equals m_pGainIndex[1] (Q4 precision).
  • GTMCurve2: Curve used when the gain equals m_pGainIndex[2] (Q4 precision).

Note. When m_nCurveSelectOption is set to 0, the curve is interpolated based on the current gain (see diagram below: Curve-Gain Control Curve Diagram).

Curve parameters are located in CCurveFirmwareFilter.

• The curve can vary with gain. Set appropriate m_pGainIndex values to specify different curves for different gains.

Noise Calibration and Debugging

RAW Image Requirements for Noise Calibration

In a laboratory environment, use a 24-color chart for shooting. Ensure that the color chart is as straight/aligned as possible, centered in the frame, and occupies approximately 1/9 of the image.

Adjust the lighting brightness accordingly, and sequentially capture images of the color chart under the following gain settings: 1×, 2×, 4×, 8×, 16×, 32×, 64×, 128×, 256×, 512×, 1024×, and 2048× gain.

Noise Calibration Steps

The calibration interface is shown in the figure below:

1. In the Noise plugin, click Load to import the RAW image. Use the Raw Preprocessor plugin to compensate for LSC and PDF (if there are PD pixels).
2. In the image, select the bottom 6 color patches on the color chart. Ensure that all 6 ROIs fall entirely within the color blocks. If the image is tilted or has noticeable distortion, click Start, check the ROIs you want to manually adjust, and drag them into the correct position.
3. Set the desired Denoise Strength.
4. Click Calibrate. The calibrated noise levels will appear under Noise Result.
5. Select the Pipe ID.
6. Select the Channel ID.
7. Click Update to apply the parameters to the parameter list. If the result is not satisfactory, click Cancel to re-calibrate

Noise Calibration Description

• Noise Result: Displays the calibrated noise level.
• Channel ID:
  • 0: Denoise parameters at 1× gain.
  • 1: Denoise parameters at 2× gain.
  • And so on, up to 11: Denoise parameters at 2048× gain.
  • manual: Denoise parameters used in manual mode.

PDAF Calibration

PDAF Calibration VRF Image Requirements

Capture a checkerboard pattern in a laboratory environment, with the object distance at 2 meters, and the checkerboard parallel to the sensor. Control the lighting brightness so that the gain is as close to 1× as possible. Capture images from the motor moving from the minimum to the maximum valid position (divide the entire scan area into 30 segments, resulting in 31 positions), for a total of 31 images. (VRF file naming convention: position.vrf; PD raw files naming convention: position_L.raw, position_R.raw).

PDAF Calibration Steps

The PDAF calibration interface is shown below:

1. In the PDC plugin, click Load to select the folder containing VRF files (if importing already extracted PD raw files, you will also need to input the raw width and height).
2. Click Calibrate, which will display a position – shift map corresponding to the image divided into 5x5 blocks.
3. Select the Pipe ID (applicable if not a single pipeline).
4. Click Update, and the m_pPDShiftPositionLUT parameter in CAFFilter will be updated.
5. If the result is unsatisfactory, click Cancel to recalibrate.

PDC Calibration

PDC is used to compensate the brightness of PD pixels or shadow pixels to normal brightness for use by the PDAF autofocus algorithm. Given the known PD points and shadow distribution (via m_pPixelMask and m_pPixelTypeMask), PDC parameters m_pRatioBMap are calibrated using images containing the PD pixel distribution.

PDC Calibration VRF Image Requirements

In a lightbox environment D65, shoot the lightbox wall using frosted glass.

PDC Calibration Steps

1. In the PDC plugin, click Analyze. The plugin will check whether the settings for m_pPixelMask and m_pPixelTypeMask are reasonable. If not, these two parameters need to be adjusted.
2. After the Analyze process confirms the settings are reasonable, the Load button becomes enabled. For QuadBayer PD, you can select the compensation mode (channel 0-1 complementary or channel 2-3 complementary; if the number of PD points in the four channels is equal, four-channel complementary compensation is also available).
3. After successfully Loading the image, the right panel will display small images composed of extracted PD points for the corresponding channels. Press the Calibrate button to calculate the m_pRatioBMap based on the image, and the compensated PD points using the new m_pRatioBMap will be displayed on the right.
4. The Update button updates the m_pRatioBMap parameter in PDC. If the result is unsatisfactory, click Cancel to recalibrate.
5. Select the Pipe ID (if multiple pipelines are used).
6. Click Update to apply the parameters to the parameter list. If the result is unsatisfactory, click Cancel to recalibrate.

PDC Calibration Explanation

Explanation of m_pPixelMask and m_pPixelTypeMask

1. These two parameters calibrate the distribution of PD and shadow pixels in the image, with a 32x32 periodic pattern.

• If m_pPixelMask = 1, the current pixel is a PD pixel, and m_pPixelTypeMask indicates one of four directions of pixel shadowing.
• If m_pPixelMask = 0 and m_pPixelTypeMask > 0, the current pixel is a shadow pixel.

2. These two parameters are usually provided by the sensor manufacturer. If not available, manual calibration can be performed by capturing RAW images.

Raw Preprocessor Plugin

Raw Preprocessor Plugin Description

The Raw Preprocessor plugin is used for raw data preprocessing. It supports PD pixel correction, LSC compensation, and unpacking VRF files (ASR RAW packed format).

Using the Raw Preprocessor Plugin

The Raw Preprocessor interface is shown below

1. Set the input and output VRF files in the Raw Preprocessor plugin.
2. Select the corresponding pipe and LSC channel.
3. Choose the desired preprocessing functions: PDF, LSC, and Unpack.
4. Click Preprocess.
5. Batch preprocess supports importing folders for batch processing of VRF files.

General Information Plugin

The General Information plugin is used to connect to the device and display some debug information in real-time.

General Information Display

By default, the following information is configured for debugging engineers’ reference:

General Information Extension

Click Setting to open the information editing page as shown below. You can freely edit the information you want to monitor. Each line represents one display item. For format details, refer to the Expression Manual.

ISP Tuning

CTopFirmwareFilter Debug Explanation

CTopFirmwareFilter is used to configure ISP Top information.

TOP Parameters

Parameter Name	Description	Recommended to Tune	Special Notes
m_nBayerPattern	Bayer pattern mode: 0: RGGB; 1: GRBG; 2: GBRG; 3: BGGR; 4: Monochrome mode	According to hardware settings
m_bAELinkZoom	AE window linked to zoom	User setting
m_bAFLinkZoom	AF linked to zoom	User setting
m_bAWBLinkZoom	AWB linked to zoom	User setting
m_nPreviewZoomRatio	Preview zoom ratio, Q8 format	User setting
m_bPreviewLowPowerMode	Preview low power mode	User setting
m_nAEProcessPosition	AE processing timing 0: eof; 1: sof	User setting
m_nAEProcessFrameNum	AE processing frequency: EOF every frame EOF every two frames SOF every frame SOF every three frames	User setting
m_bHighQualityPreviewZoomEnable	Reserved

CAEMFirmwareFilter Parameter Description

The CAEMFirmwareFilter module is used to configure the Auto Exposure (AE) statistics module.

AEM Enable and Parameters

Parameter Name	Description	Recommended to Tune	Special Note
m_bEnable	AEM Enable: 0: Disable AE statistics module; 1: Enable AE statistics module	No
m_nAEStatMode	AE statistics mode: 0: Statistics without white balance 1: Statistics with white balance	No
m_bZSLHDRCapture	Zero Shutter Lag HDR capture enable 0: Disable 1: Enable zero shutter lag HDR capture	User setting
m_nInitialExpTime	vInitial exposure time	Yes
m_nInitialAnaGain	Initial analog gain	Yes
m_nInitialSnsTotalGain	Initial sensor total gain	Yes
m_nInitialTotalGain	Initial total gain	Yes
m_nStableTolerance	AE stability tolerance percentage: If the difference between current exposure and previous exposure is less than previous exposure * m_nStableTolerance%, AE StableFlag is issued for use by other modules like LTM	User setting
m_nStableToleranceExternal	AE stability tolerance percentage for external systems	User setting
m_bAutoCalculateAEMWindow	AE statistics window calculation method: 0: Configured by hardware; 1: Controlled by firmware	No
m_nPreEndingPercentage	Percentage of rows (relative to image height) excluded from AE statistics	No
m_bDRCGainSyncOption	DDRC gain synchronization: 0: Sync every frame; 1: Sync after AE stable	No
m_pSceneChangeSADThr	SAD threshold for scene change detection	User setting
m_pSubROIPermil	Start and end coordinates of 6 sub-statistics modules relative to image width and height in permille; modifiable by application (e.g., face metering or focus metering linkage)	User setting
m_nSubROIScaleFactor	Sub-window scaling percentage factor	User setting
m_nFaceLumaOption	Face luminance statistics method: 0: Hardware statistics (pixel) ; 1: Software statistics (block)	No
m_bMotionDetectEnable	Motion detection enable: 0: Disable 1: Enable	User setting
m_bMotionDetectExt	Motion detection method: 0: Use internal AEM statistics; 1: Use external gyro sensor	User setting
m_nMotionStrengthExt	External motion strength control, effective when motion detection method is external gyro	User setting
m_nSADIntervalFrame	Interval frame count for SAD calculation, effective when motion detection method is internal AEM statistics	No
m_nMotionThreshold	SAD threshold to judge motion, effective when motion detection method is internal AEM statistics	No
m_nMotionDetectFrame	Number of consecutive frames with SAD exceeding threshold to consider as motion scene	No
m_nFaceDetFrameID	Frame ID where face was detected	-	Read-only
m_nAdjacentLumaSAD	Current SAD	-	Read-only
m_pMainRoiCoordinate	Main window coordinates	-	Read-only
m_pSubRoiCoordinate	Sub-window coordinates	-	Read-only

CDigitalGainFirmwareFilter Parameter Description

The CDigitalGainFirmwareFilter module is used to configure digital gain and black level.

Digital Gain Enable and Parameters

Parameter Name	Description	Recommended to Tune	Special Note
m_bEnable	Digital gain enable: 0: Disable digital gain; 1: Enable digital gain	User setting
m_nISPGlobalOffsetValue12bit	0: Subtract black level in stretch; 1: Subtract black level fully in digital gain; 2-511: Offset added after black level subtraction at 12-bit (this offset is subtracted in stretch)	User setting
m_bManualMode	Manual mode enable 0: Auto mode 1: Enable manual mode; black level parameters do not change with gain and use manual parameters; for debug use
m_pGlobalBlackValueManual	Manual mode parameters, function same as auto mode
m_pGlobalBlackValueManualCapture	Same as above, effective during capture
m_pGlobalBlackValue	Black levels for R/GR/GB/B channels (see Gain-BlackValue illustration)	Calibration result	Can vary with gain
m_pGlobalBlackValueCapture	Same as above, effective during capture	Calibration result	Can vary with gain
m_pWBGoldenSignature	White balance golden module signature	-	Read-only
m_pWBCurrentSignature	White balance current module signature	-	Read-only

The BlackValue diagram is as follows:

CWBGainFirmwareFilter Parameter Description

The CWBGainFirmwareFilter module is used for Auto White Balance (AWB) gain.

WB Gain Enable

Parameter Name	Description	Recommended to Tune	Special Note
m_bEnable	WB gain enable: 0: Disable white balance gain; 1: Enable white balance gain	No

CStretchFirmwareFilter Parameter Description

The CStretchFirmwareFilter module is used to compensate for pixel undersaturation after black level subtraction.

Stretch Enable

Parameter Name	Description	Recommended to Tune	Special Note
m_bEnable	Stretch enable: always on, used to compensate for black level subtraction and pixel undersaturation	No

CColorMatrixFirmwareFilter Parameter Description

The CColorMatrixFirmwareFilter (CCM) module is used for color correction.

CCM Enable

Parameter Name	Description	Recommended to Tune	Special Note
m_bEnable	CCM enable: 0: Disable the color correction matrix; 1: Enable the color correction matrix	No

CCM Parameters and Tuning

Parameter Name	Description	Recommended Tuning	Special Notes
m_bUseCorrelatedCT	Interpolation basis option: 0: Use AWB's CT; 1: Use m_nCorrelationCT from WbFirmwareFilter (CCT matrix needs calibration)	User-defined
m_pColorTemperatureIndex	Color temperature segmentation control points. (See example below: cmc-color temperature control curve) If the color temperature is in the range [0, Index[0]], it is considered a low color temperature range, and CMC0 is used; In [Index[0], Index[1]], an interpolated matrix between CMC0 and CMC1 is used; In [Index[1], Index[2]], it is considered a medium color temperature range, and CMC1 is used; In [Index[2], Index[3]], an interpolated matrix between CMC1 and CMC2 is used; In [Index[3], 8192], it is considered a high color temperature range, and CMC2 is used;	Yes
m_pCMC0	Low color temperature correction matrix, calibrated by the CCM plugin. R'G'B' to RGB, Q12 precision.	Calibration result parameter	Callable based on color temperature
m_pCMC1	Medium color temperature correction matrix, calibrated by the CCM plugin. R'G'B' to RGB, Q12 precision.	Calibration result parameter	Callable based on color temperature
m_pCMC2	High color temperature correction matrix, calibrated by the CCM plugin. R'G'B' to RGB, Q12 precision.	Calibration result parameter	Callable based on color temperature

cmc-Color Temperature Control Curve (see figure below)

CCM Color Fringe Suppression Function and Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_bColorFringleRemoveEnable	Color fringe suppression enable: 0: Disable 1: Enable	User-defined
m_nColorFringRemovalStrength	Color fringe suppression strength: The larger the value, the stronger the suppression effect	Yes

Note. Final CFR_Ratio = HueRatio*EdgeRatio>>HighFreqTransShiftNum

• Hue Control Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_nHueTransShiftNum	Hue transition zone offset coefficient: (See example below: HueTrans-HueRatio control curve) Hue values falling in the range [(ColorFringeHueRange[0]-(1<<ShiftNum)，ColorFringeHueRange[0]] will be smoothed; Hue values in the range [ColorFringeHueRange[1]，(ColorFringeHueRange[1]+(1<<ShiftNum)] will also be smoothed;	Yes
m_pColorFringeHueRange	Hue range for color fringe suppression (see example below: HueTrans-HueRatio control curve). HueRange[0] must be less than HueRange[1]	Yes

ColorFringeHueRange[0],[1] is used to define the Hue range for color fringe suppression;

HueTransShiftNum is used to set the smoothing transition zone:

• Freq Control Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_nHighFreqThreshold	Lower frequency threshold for color fringe suppression (see example in HighFreqTrans-EdgeRatio curve). The larger the value, the fewer edges enter the color fringe suppression area	Yes
m_nHighFreqTransShiftNum	High-frequency transition band offset coefficient (see example in HighFreqTrans-EdgeRatio curve). Frequencies falling into the range [HighFreqThreshold, HighFreqThreshold +(1<<HighFreqTransShiftNum)] are smoothed	Yes

HighFreqTrans-EdgeRatio curve shown below

CCM Manual Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_bManualMode	Manual mode enable 0: Automatic mode; 1: Enable manual mode; in this case, the color correction matrix parameters do not change with color temperature, manual parameters are used; for debugging	-	Debug parameter
m_pCMCManual	Manual mode parameters, same function as automatic	-	Debug parameter

CCM Other Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_bDisgardHFEnable	Discard high-frequency information enable: 0: Add high-frequency information; 1: Discard high-frequency information	User-defined
m_pCMCSaturationList	Saturation control		Can vary with Gain

CBPCFirmwareFilter Tuning Description

CBPCFirmwareFilter (BPC) module is used for bad pixel correction.

BPC Enable

Parameter Name	Description	Recommended Tuning	Special Notes
m_bEnable	BPC enable 0: Disable bad pixel correction; 1: Enable bad pixel correction	User-defined

BPC Dynamic Control Parameters

BPC strength can be dynamically adjusted according to gain and brightness.

Parameter Name	Description	Recommended Tuning	Special Notes
m_pBpcGainIndex	Gain index, it is recommended to keep the default value	No	Gain control node
m_pSegG	Brightness index, the interval between adjacent levels must be a power of 2, it is recommended to keep the default value	No	Lum control node

• The gain control parameter is m_pBpcGainIndex, with twelve groups from 0 to 11; 16 corresponds to 1x gain. When the gain is between two nodes, the parameter is the interpolation result of the two node parameters.

• The brightness control parameter is m_pSegG, with nine groups from 0 to 8, where the 8th group is fixed at 255 and cannot be changed, corresponding to VRF data pixel value (mapped to 8 bits). When brightness is between two nodes, the parameter is the interpolation result of the two node parameters. The interval between adjacent levels must be a power of 2, it is recommended to keep the default value.

• The strength control parameter can be dynamically adjusted with changes in gain and brightness.

Parameter Name	Description	Recommended Tuning	Special Notes
m_pCrossChnStrength	Cross-channel strength; the larger the value, the more other channel information is referenced, but it is also more susceptible to bad pixels from other channels	Yes	Can vary with Gain
m_pSlopeG	G channel control curve parameter; the larger the value, the greater the tolerance and the weaker the bad pixel correction ability	Yes	Can vary with Gain and Lum
m_pInterceptG	G channel control curve parameter; the larger the value, the greater the tolerance and the weaker the bad pixel correction ability	Yes	Can vary with Gain and Lum
m_pSlopeRB	RB channel control curve parameter; the larger the value, the greater the tolerance and the weaker the bad pixel correction ability	Yes	Can vary with Gain and Lum
m_pInterceptRB	RB channel control curve parameter; the larger the value, the greater the tolerance and the weaker the bad pixel correction ability	Yes	Can vary with Gain and Lum

Taking m_pSlopeG as an example:

• Column represents the Gain level, corresponding one-to-one with m_pBpcGainIndex.

• Row represents the Lum level, corresponding one-to-one with m_pSegG.

• Parameter interpolation with Lum change explanation

Notes:

• The above values changing with Lum include Slope and Intercept
• Final tolerance is jointly determined by Slope, Intercept, and Ratio. Tolerance = (Lum*Current_Slope + Current_Intercept)* Ratio. The greater the tolerance, the weaker the bad pixel correction

Explanation: Current_Slope and Current_Intercept are interpolated based on Lum and gain changes. Explanation: Ratio is divided into Dead/SpikeRatio and RB/G, totaling 4 cases.

BPC Functional Modules and Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_nMinThrEn	Dark pixel detection channel selection Bit 0: Enable reference to cross-channel information; Bit 1: Enable reference to GrGb channel information; Bit 2: Enable reference to the same channel information.	No
m_nMaxThrEn	Bright pixel detection channel selection Bit 0: Enable reference to cross-channel information; Bit 1: Enable reference to GrGb channel information; Bit 2: Enable reference to the same channel information.	No
m_nNearThr	Lower brightness threshold for using cross-channel information; cross-channel information is used only when brightness is above this threshold.	No
m_bDeadEnable	Dark pixel correction enable	User-defined
m_nDeadRatioG	G channel dark pixel coefficient; the larger the value, the greater the tolerance for G channel dark pixels and the weaker the bad pixel correction ability	Yes
m_nDeadRatioRB	RB channel dark pixel coefficient; the larger the value, the greater the tolerance for RB channel dark pixels and the weaker the bad pixel correction ability	Number
m_bSpikeEnable	Bright pixel correction enable	User-defined
m_nSpikeRatioG	G channel bright pixel coefficient; the larger the value, the greater the tolerance for G channel bright pixels and the weaker the bad pixel correction ability	Yes
m_nSpikeRatioRB	RB channel bright pixel coefficient; the larger the value, the greater the tolerance for RB channel bright pixels and the weaker the bad pixel correction ability	Yes
m_bSameChnNum	Same channel pre-correction enable; 1: enable same channel pre-correction, which can exclude interference from bad pixels in the same channel	User-defined
m_nDeltaThr	Same channel pre-correction threshold; it is recommended to keep the default value	No
m_nRingGRatio	Same channel pre-correction threshold; it is recommended to keep the default value	No
m_nRingMeanRatio	Same channel pre-correction threshold; it is recommended to keep the default value	No
m_bCornerDetEn	Corner detection enable; can protect corners	User-defined
m_pSlopeCorner	Corner control curve parameter; the larger the value, the fewer corners are protected; unrelated to tolerance	Yes	Can vary with Gain and Lum
m_pInterceptCorner	Corner control curve parameter; the larger the value, the fewer corners are protected; unrelated to tolerance	Yes	Can vary with Gain and Lum
m_bEdgeDetEn	Edge detection enable; can protect edges	User-defined
m_nEdgeTimes	Edge determination threshold; the smaller the value, the more edges are protected	Yes
m_bGrGbNum	GrGb channel pre-correction enable; 1: enable GrGb channel pre-correction, which can exclude interference from bad pixels in GrGb channel	User-defined
m_bAroundDetEn	Bright block detection enable; can protect pixels with sudden brightness changes	User-defined
m_bBlockDetEn	2x2 bad block detection enable; can exclude interference from 2x2 bad blocks	User-defined

---

BPC Manual Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_bManualMode	Manual mode enable 1: Enable manual mode; in this case, BPC parameters do not change with gain and manual parameters that are used; for debugging	-	Debug parameter
m_nCrossChnStrengthManual	Manual mode parameter, functionally consistent with automatic mode	-	Debug parameter
m_pSlopeGManual	Manual mode parameter, functionally consistent with automatic mode	-	Debug parameter
m_pInterceptGManual	Manual mode parameter, functionally consistent with automatic mode	-	Debug parameter
m_pSlopeRBManual	Manual mode parameter, functionally consistent with automatic mode	-	Debug parameter
m_pInterceptRBManual	Manual mode parameter, functionally consistent with automatic mode	-	Debug parameter
m_pSlopeCornerManual	Manual mode parameter, functionally consistent with automatic mode	-	Debug parameter
m_pInterceptCornerManual	Manual mode parameter, functionally consistent with automatic mode	-	Debug parameter

CLSCFirmwareFilter Parameter Description

CLSCFirmwareFilter module is used for lens shading correction.

LSC Enable

Parameter Name	Description	Recommended Tuning	Special Notes
m_bEnable	LSC enable 0: Disable lens shading correction; 1: Enable lens shading correction	User-defined
m_bUseOTP	LSC OTP enable	User-defined

LSC Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_bAutoScale	AutoScale enable: 0: Disable automatic scaling parameter calculation; 1: Enable automatic scaling parameter calculation	No
m_bEnhanceEnable	Enhance enable: Enable when the current module’s expected compensation multiple exceeds 4x 0: Disable shading enhancement; 1: Enable shading enhancement	User-defined
m_nProfileSelectOption	Shading compensation table selection: 0: Automatically select based on color temperature; 1: Use LSC Profile[0]; 2: Use LSC Profile[1]; 3: Use LSC Profile[2]	No
m_nFOVCropRatioH	Horizontal cropping ratio	Yes	Determined by binning size
m_nFOVCropRatioV	Vertical cropping ratio	Yes	Determined by binning size
m_pLSCStrength	Shading compensation strength (see Gain-strength illustration) 64 represents 1x; 32 represents 1/2x; 16 represents 1/4x; Other values follow similarly	Yes	Can be adjusted based on gain
m_bUseCorrelatedCT	Interpolation basis option: 0: Use AWB’s CT 1: Use m_nCorrelationCT in WbFirmwareFilter (CCT matrix needs calibration)	User-defined
m_pCTIndex	Color temperature segment control for LSC profile (see LSC-color temperature control curve below): Effective when m_nProfileSelectOption is set to 0. When color temperature is in [0, CTIndex[0]], considered low color temperature range, use compensation table of LSCProfile[0]; When in [CTIndex[0], CTIndex[1]], use interpolation of LSCProfile[0] and LSCProfile[1]; When in [CTIndex[1], CTIndex[2]], considered medium color temperature range, use compensation table of LSCProfile[1]; When in [CTIndex[2], CTIndex[3]], use interpolation of LSCProfile[1] and LSCProfile[2]; When in [CTIndex[3], 8192], considered high color temperature range, use compensation table of LSCProfile[2];	Yes
m_pLSCProfile	LSC compensation table, calibrated by LSC plugin	Calibration result parameter	Can be called based on color temperature

LSC-color temperature control curve below

Gain-strength illustration below

Adaptive Color Shading Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_bLSCCSCEnable	Adaptive color shading correction enable (CSC)	User-defined
m_bAdjustCSCTblMinEnable	Adjust CSC table based on minimum R/G (B/G) values	No
m_nDifThr	Dif threshold after vector median filtering; if dif is greater than this value, the block is excluded from CSC calculation	No
m_nDifThrMinPerc	Dif threshold after vector median filtering; if dif is less than DifThrMinPerc*DifThr, angle is not considered in CSC calculation	No
m_nAngleThr	Angle threshold after vector median filtering; if angle is greater than this value, the block is excluded from CSC calculation	No
m_nDifThrVMF	Dif threshold before and after vector median filtering; if dif is greater than this value, the block is excluded from CSC calculation	No
m_nAngleThrVMF	Angle threshold before and after vector median filtering; if angle is greater than this value, the block is excluded from CSC calculation	No
m_nGradThrMin	Minimum effective gradient after vector median filtering	No
m_nGradThrMax	Maximum effective gradient after vector median filtering	No
m_nGradMaxError	Maximum tolerance for error between CSC estimated gradient and real gradient statistics	Yes
m_nGradThrConv	Threshold for gradient difference between two CSC calculations; if difference is less than this value, CSC is not updated	Yes
m_nGradMax	Maximum CSC compensation strength	Yes
m_nTblAlpha	Convergence speed; the larger the value, the faster the convergence	Yes
m_nCSCGlobalStrength	CSC global strength	Yes
m_nEffPNumAll	Effective block threshold for entire image	No
m_nEffPNumHalf	Effective block threshold for half image	No
m_nEffPNumQuarter	Effective block threshold for quarter image	No
m_pEffNumRing	Effective block thresholds for three ROIs: ROI0 is center 6x4 ROI1 is center 12x8 (excluding ROI0) ROI2 is 12x12 (excluding ROI0, ROI1)	No
m_pCSCCTIndex	CSC color temperature control points; CSC invalid when CT > CSCCTIndex[1]	Yes
m_pCSCLuxIndex	CSC brightness control points; CSC invalid when Lux > CSCLuxIndex[1]	Yes

LSC Manual and frameinfo

Parameter Name	Description	Recommended Tuning	Special Notes
m_bManualMode	Manual mode enable: 1: Enable manual mode; in this case LSC parameters do not change with color temperature and manual parameters are used; for debugging		Debug parameter
m_nLSCStrengthManual	Manual mode parameter, functionally consistent with automatic mode		Debug parameter
m_pLSCProfileManual	Manual mode parameter, functionally consistent with automatic mode		Debug parameter
m_nRGPolyCoefRO	Current CSC compensation R ratio		Read only
m_nBGPolyCoefRO	Current CSC compensation B ratio		Read only
m_pRGRatio	R/G ratio corresponding to 16x12 statistical blocks		Read only
m_pBGRatio	B/G ratio corresponding to 16x12 statistical blocks		Read only
m_pOTPProfileInternal	OTP shading table		Read only

CDemosaicFirmwareFilter Parameter Description

CDemosaicFirmwareFilter (Demosaic) module is used for Bayer interpolation.

Demosaic Subfunction Enable

Parameter Name	Description	Recommended Tuning	Special Notes
m_bIfEdgeGenerate	High-frequency information generation enable: can be used for inline sharpen (inline sharpen function requires cooperation with cmc module) 0: Disable; 1: Enable	User-defined
m_bIfGbGrRebalance	GbGr difference elimination enable 0: Disable; 1: Enable	No
m_bIfDNS	Inline denoise enable, recommended to disable 0: Disable; 1: Enable	No

Demosaic Dynamic Control Parameters

Demosaic parameters can be dynamically adjusted with gain.

Gain control nodes N range from 0 to 11, twelve groups in total. The gain at node N is 2^N times, i.e., node 0 corresponds to 1x gain; node 11 corresponds to 2048x gain.

Parameter Name	Description	Recommended Tuning	Special Notes
m_nInterpOffset	Noise tolerance in four directions	No	Can change with Gain
m_nInterpOffsetHV	Noise tolerance in horizontal and vertical directions	No	Can change with Gain
m_nNoiseSTD	Noise tolerance standard deviation	No	Can change with Gain
m_nLowpassGLevel	Interpolation frequency control parameter; smaller values favor 4-direction interpolation results, larger values favor directionless low-pass interpolation results	Yes	Can change with Gain
m_nGbGrThr	Threshold corresponding to GbGr difference elimination function; smaller values weaken the function, larger values strengthen it	Yes	Can change with Gain
m_nSharpenStrength	High-frequency information amplification factor; larger values mean stronger sharpening	Yes	Can change with Gain
m_nShpThreshold	Threshold for soft threshold processing of high-frequency information; recommended to keep default value	No	Can change with Gain
m_nDenoiseThreshold	Soft threshold for inline denoise function; larger values mean stronger denoise; recommended to keep default value	No	Can change with Gain
m_nNoiseAddbackLevel	Noise feedback strength for inline denoise; larger values weaken denoise; recommended to keep default value	No	Can change with Gain
m_pDenoiseLumaStrength	Luma-based scaling coefficient for denoise strength; luma levels in 8-bit [8,16,32]; for luma above 64, coefficient is 32, no scaling	No	Can change with Gain
m_nChromaNoiseThreshold	Threshold for chroma noise removal; larger values mean stronger removal; recommended to keep default value	No	Can change with Gain
m_pUSMFilter	USMFilter = conv([1 2 1], [usm2 usm1 usm0 64-2*(usm0+usm1+usm2) usm0 usm1 usm2]); recommended to keep default value	No	Can change with Gain

Taking m_nSharpenStrength as an example:

Column represents the gain level

• Column[0] corresponds to the value at 1x gain
• Column[11] corresponds to the value at 2048x gain

Gain – Sharpen illustrative chart as below

Demosaic Other Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_pHFFragShiftIndex	Segment information for high-frequency segmented gain processing; recommended to keep default value	No
m_pHFFragGainIndex	Gain information for high-frequency segmented gain processing; recommended to keep default value	No

Demosaic Manual Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_bManualMode	Manual mode enable 0: Automatic mode; 1: Manual mode; in this case demosaic parameters do not change with gain, manual parameters are used; for debug	-	Debug parameter
Parameters ending with Manual	Manual mode parameters, functionally consistent with automatic mode	-	Debug parameter

CRawDenoiseFirmwareFilter Parameter Description

CRawDenoiseFirmwareFilter (RawDenoise) module is used for RAW domain denoise.

RawDenoise Enable

Parameter Name	Description	Recommended Change	Special Notes
m_bEnable	RAW denoise module enable switch 0: Disable RAW domain denoise; 1: Enable RAW domain denoise	Based on user settings

RawDenoise Dynamic Control Parameters

RawDenoise parameters can be dynamically adjusted with gain.

Gain control nodes N range from 0 to 11, twelve groups in total. The gain at node N is 2^N times, i.e., node 0 corresponds to 1x gain; node 11 corresponds to 2048x gain. (See Gain-Denoise_strength illustrative chart)

Parameter Name	Description	Recommended Change	Special Notes
m_pMaxSpacialDenoiseThreGain	Maximum corner denoise strength, Q8 precision; the larger the value, the stronger the denoise strength achievable at corners	Yes	Can change with Gain
m_pSigma	Denoise strength threshold; the larger the value, the stronger the denoise strength	Yes	Can change with Gain
m_pGns	G channel denoise strength; the larger the value, the stronger the denoise strength	Yes	Can change with Gain
m_pRbns	RB channel denoise strength; the larger the value, the stronger the denoise strength	Yes	Can change with Gain
m_pL0	Denoise strength scaling coefficient at 1.5% brightness, Q5 precision; the larger the value, the stronger the denoise strength at this brightness	Yes	Can change with Gain
m_pL1	Denoise strength scaling coefficient at 7.8% brightness, Q5 precision; the larger the value, the stronger the denoise strength at this brightness	Yes	Can change with Gain
m_pL2	Denoise strength scaling coefficient at 20% brightness, Q5 precision; the larger the value, the stronger the denoise strength at this brightness	Yes	Can change with Gain
m_pL3	Denoise strength scaling coefficient at 45% brightness, Q5 precision; the larger the value, the stronger the denoise strength at this brightness	Yes	Can change with Gain

m_pL0 - m_pL3 correspond to denoise strength at different brightness levels, as shown in the figure below:

Taking m_pSigma as an example:

Column represents the Gain level:

• Column[0] corresponds to the parameter at 1x gain;
• Column[11] corresponds to the parameter at 2048x gain;

Gain - Denoise_strength illustrative chart is shown below

RawDenoise Functional Modules and Parameters

Parameter Name	Description	Recommended Change	Special Notes
m_bMergeEnable	2x2->1x1 conversion method when calculating denoise weight 0: Take lower-left corner 1: Take average of 2x2	No
m_bLocalizedEnable	Enable denoise strength variation with local brightness 1: Disable 2: Enable	No
m_bSpacialEnable	Enable edge denoise strength enhancement: 0: Disable 1: Enable	Based on user settings
m_bSpacialAddbackEnable	Enable edge denoise add-back: 0: Disable 1: Enable	Based on user settings
m_nSpacialOffCenterPercentage	Edge denoise enhancement area control parameter; denoise strength enhancement starts from Centerpercentage*R (see R – CenterPercent illustrative chart below)	Yes
m_pMaxSpacialDenoiseThreGain	Maximum edge denoise enhancement threshold, the maximum denoise strength achievable at the farthest distance (see Distance – RadialGain illustrative chart below)	Yes

R - CenterPercent illustrative chart below

Distance - RadialGain illustrative chart below

RawDenoise debug Parameters

Parameter Name	Description	Recommended Change	Special Notes
m_bManualMode	Manual mode enable 0: Auto mode; 1: Enable manual mode; raw denoise parameters do not change with gain and manual parameters that are used for debug
m_nSigmaManual	Manual mode parameter, same function as auto mode
m_nGnsManual	Manual mode parameter, same function as auto mode
m_nRbnsManual	Manual mode parameter, same function as auto mode
m_nL0Manual	Manual mode parameter, same function as auto mode
m_nL1Manual	Manual mode parameter, same function as auto mode
m_nL2Manual	Manual mode parameter, same function as auto mode
m_nL3Manual	Manual mode parameter, same function as auto mode

CAFMFirmwareFilter Parameter Description

CAFMFirmwareFilter module is used for the Auto Focus Measurement (AFM) statistics module.

AFM Enable

Parameter Name	Description	Recommended Tuning	Special Notes
m_bEnable	AFM enable 0: Disable auto focus statistics module; 1: Enable auto focus statistics module	User setting

AFM Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_nAFStatMode	AF statistics module mode: 0: After white balance; 1: Without white balance	No
m_nWinStartXPermil	AFM horizontal start coordinate in permillage	User setting
m_nWinStartYPermil	AFM vertical start coordinate in permillage	User setting
m_nWinEndXPermil	AFM horizontal end coordinate in permillage	User setting
m_nWinEndYPermil	AFM vertical end coordinate in permillage	User setting
m_nMinWidthPermil	AFM minimum width in permillage	User setting
m_nMinHeightPermil	AFM minimum height in permillage	User setting
m_bConfigDone	FW control parameter, set to 1 when AF window configuration is done	No
m_pFVList	Focus values of each AF window	-	Read-only
m_nFVAvg	Average focus value	-	Read-only
m_nWinStartX	AFM horizontal start coordinate	-	Read-only
m_nWinStartY	AFM vertical start coordinate	-	Read-only
m_nWinWidth	AFM width	-	Read-only
m_nWinHeight	AFM height	-	Read-only

CPDCFirmwareFilter Parameter Description

CPDCFirmwareFilter module is used to compensate PD pixels or shadow pixels to normal brightness for PDAF algorithm usage.

PDC Enable

Parameter Name	Description	Recommended Tuning	Special Notes
m_bEnable	PDC enable: 0: Disable PDC module; 1: Enable PDC module	User setting

PDC Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_bOut	PD dump enable 0: Disable dumping all PD points inside the window; 1: Enable dumping all PD points inside the window	No
m_nHOffset	Horizontal offset of ISP processed image relative to sensor output image width (see window diagram below for m_nHOft)	Yes
m_nVOffset	Vertical offset of ISP processed image relative to sensor output image height (see window diagram below for m_nVOft)	Yes
m_bLRAdjust	Direction control: 0: Disable mirror; 1: Enable mirror	Yes
m_bTBAdjust	Direction control: 0: Disable flip; 1: Enable flip	Yes
m_nFullWidth	Sensor output image width	Yes
m_nFullHeight	Sensor output image height	Yes
m_nWindowMode	Window mode: 0: Auto-calculate PD dump region; 1: Calculate PD dump region via m_nWinStartXPermil, m_nWinStartYPermil, m_nWinEndXPermil, and EndYPermil	No
m_nWindowScaleFactor	Scale factor of PDC statistics window relative to AFM statistics window	User setting
m_nMinWidthPermil	Minimum width of PDC statistics window (permille)	User setting
m_nMinHeightPermil	Minimum height of PDC statistics window (permille)	User setting
m_pPDFirstX	Horizontal offset of PD region relative to sensor output image width (see window diagram below)	Yes
m_pPDFirstY	Vertical offset of PD region relative to sensor output image height (see window diagram below)	Yes
m_pRatioA	Global adjustment ratio for four channels	Yes
m_pPixelMask	Distribution of PD points within a 32x32 area	Yes
m_pPixelTypeMask	Distribution of PD point types within a 32x32 area; PD types divided into shadow up, down, left, right	Yes
m_pRatioBMap	Compensation coefficients for four-channel PD points	Yes
m_bSoftCompEnable	Software compensation switch for PD brightness difference: 0: Use hardware PDC compensation; 1: Software compensation (sensor extracts PD pixels)
m_nWinStartXPermil	Horizontal coordinate (permille) of PD dump region top-left corner	-	Read-only
m_nWinStartYPermil	Vertical coordinate (permille) of PD dump region top-left corner	-	Read-only
m_nWinEndXPermil	Horizontal coordinate (permille) of PD dump region bottom-right corner	-	Read-only
m_nWinEndYPermil	Vertical coordinate (permille) of PD dump region bottom-right corner	-	Read-only
m_nWinStartX	Horizontal start coordinate of PDC statistics window	-	Read-only
m_nWinStartY	Vertical start coordinate of PDC statistics window	-	Read-only
m_nWinWidth	Width of PDC statistics window	-	Read-only
m_nWinHeight	Height of PDC statistics window	-	Read-only

Window diagram shown below:

CPDFFirmwareFilter Parameter Description

The CPDFFirmwareFilter module is used to correct PD pixels to normal pixel values.

PDF Enable

Parameter Name	Description	Recommended Tuning	Special Notes
m_bEnable	PDF enable 0: Disable phase detection pixel correction; 1: Enable phase detection pixel correction	User setting

PDF Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_nExtPRm	PD pixel outlier correction enable 0: Disable PD pixel outlier correction; 1: Enable PD pixel outlier correction	User setting
m_nWA	Center block weight; 8 corresponds to 50% weight. Usually, the center point is a compensated PD point, so weight is generally set smaller.	No
m_nWB	Diagonal block weight; 8 corresponds to 50% weight, usually set to 50%.	No
m_nFullWidth	Sensor output image width	Yes
m_nFullHeight	Sensor output image height	Yes
m_nHOffset	Horizontal offset of ISP processed image relative to sensor output image width (see window diagram m_nHOft)	Yes
m_nVOffset	Vertical offset of ISP processed image relative to sensor output image height (see window diagram m_nVOft)	Yes
m_bLRAdjust	Direction control: 0: Disable horizontal mirror 1: Enable horizontal mirror;	Yes
m_bTBAdjust	Direction control: 0: Disable vertical flip; 1: Enable vertical flip	Yes
m_bRefRB	- 0: Green channel correction does not reference RB channel; 1: Green channel correction references RB channel	No
m_bRefCnr	- 0: Green channel correction does not reference corner info; 1: Green channel correction references corner info	No
m_nRefNoiseL	Current noise level used to determine edge direction	No
m_nExtPThre	Threshold for PD pixel outliers	No
m_nExtPSft	Soft threshold for PD pixel outliers	No
m_nExtPOpt	Selection of PD pixel outlier correction	No
m_pPDFirstX	Horizontal offset of PD region relative to sensor output image width (see window diagram)	Yes
m_pPDFirstY	Vertical offset of PD region relative to sensor output image height (see window diagram)	Yes
m_pPixelMask	Distribution of PD points within 32x32 area	Yes
m_pPDResult	Four-direction PD shift and confidence		Read-only

CPDAFFirmwareFilter Parameters Description

CPDAFFirmwareFilter module is used for Phase Detection Auto Focus (PDAF).

PDAF Lookup Table

Parameter Name	Description	Recommended Tuning	Special Notes
m_bMirrorShift	Output phase difference inversion	User setting
m_bShiftLutEn	Lookup table enable	User setting
m_pShiftLut	Output phase difference lookup table	No

PDAF Error Control

Parameter Name	Description	Recommended Tuning	Special Notes
m_nErrorDistWeight	Weight of error source (see the correlation fitting curve below) 0: shape weight = 1; 128: shape and distance weights are equal; 256: distance weight = 1;	No
m_nErrorDistCoef	Adjustment coefficient for distance (correlation fitting curve distance)	No
m_nErrorShpCoef	Adjustment coefficient for shape (correlation fitting curve shape)	No

Correlation fitting curve illustration:

PDAF Dynamic Control Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_pLumThre	Brightness threshold at different gains (see LumThre-gain control curve below). When current brightness is below the threshold, confidence decreases.	Yes	Adjust according to gain
m_pGainThre	Basis for dividing different gain intervals, used to control LumThre and SwingThre according to gain	No
m_pSwingThre	Amplitude threshold at different gains (see SwingThre-gain control curve below). When current amplitude is below this threshold, confidence decreases.	Yes	Adjust according to gain

LumThre-Gain control curve:

SwingThre-Gain control curve:

PDAF Confidence Control Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_bConfAdjust	Confidence adjustment switch: 1: Increase confidence when current amplitude is above this threshold; 0: Confidence remains unchanged when current amplitude is above this threshold	No
m_nConfOff	Confidence offset, used to adjust the final confidence value	No
m_nConfLimit	Maximum confidence (see Error-Confidence conversion curve below)	No
m_nErrorThre1	Error threshold for confidence conversion (see Error-Confidence conversion curve below)	Yes
m_nErrorThre2	Confidence threshold for error conversion (see Error-Confidence conversion curve below)	Yes
m_nSearchRange	PD shift search range. The higher the PD pixel density, the larger this value. Usually 0 for shield pixel density 3 for Dual PD	Yes

Error-Confidence conversion curve shown below:

CWbFirmwareFilter Parameter Description

The CWbFirmwareFilter module is used for white balance.

WB Enable

Parameter Name	Description	Recommended Tuning	Special Notes
m_bEnable	WB enable: 0: Disable white balance statistics module; 1: Enable white balance statistics module	No

WB Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_bSyncWB	HDR mode white balance synchronization method: 0: AWB calculated separately; 1: Use long exposure as AWB calculation source	No
m_bAutoWindow	Window adjustment method: 0: Fixed window size; 1: Automatically calculate window size based on zoom factor	No
m_nMode	White balance mode selection: 0: auto mode; 1: custom; 2: D75; 3: D65; 4: D50; 5: CWF 6: TL84; 7: A 8: H 9: lock	No
m_nInitMode	White balance initialization mode: 0: custom; 1: D75; 2: D65; 3: D50; 4: CWF; 5: TL84; 6: A; 7: H;	Yes
m_pManualGain	Manual WB gain: indices 0-7 correspond to custom / D75 / D65 / D50 / CWF / TL84 / A / H	Yes
m_nAWBStableRange	Threshold for AWB stable state: smaller value means harder to judge AWB as stable	Yes
m_nAWBStableFrameNum	Number of frames reference for AWB stability: stable frames exceeding this value mean AWB stable	Yes
m_nAWBUnStableRange	Threshold for AWB unstable state: smaller value means easier to judge AWB as unstable	Yes
m_nAWBUnStableFrameNum	Number of frames reference for AWB instability: unstable frames exceeding this value mean AWB unstable	Yes
m_nAWBStep1	AWB relative step length for convergence: larger value means larger adjustment step, faster convergence but less accuracy	Yes
m_nAWBStep2	AWB absolute step length for convergence: larger value means larger step, faster convergence but less accuracy	Yes
m_nLowThr	Lower brightness threshold for AWB statistics: higher value excludes more dark regions from AWB stats	Yes
m_nHighThr	Upper brightness threshold for AWB statistics: lower value excludes more bright regions from AWB stats	Yes
m_nCorrelationCT	Current correlated color temperature	-	Read-only
m_nTint	Current tint value	-	Read-only
m_bAWBStableFlag	Current AWB status	-	Read-only
m_nDistance	Absolute sum of difference between applied white balance gain and target gain	-	Read-only

CCTCalculatorFilter Parameter Description

CCTCalculatorFilter module is used to calculate the true color temperature (CCT).

CCTCalculator Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_nIterateNumber	Number of iterations for CT calculation	No
m_nColorTemperatureLow	Color temperature corresponding to the low color temperature matrix
m_nColorTemperatureHigh	Color temperature corresponding to the high color temperature matrix
m_pCTMatrixLow	Low color temperature matrix		Calibration result
m_pCTMatrixHigh	High color temperature matrix		Calibration result

CRGB2YUVFirmwareFilter Parameter Description

CRGB2YUVFirmwareFilter module is used for RGB to YUV conversion.

RGB2YUV Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_nOutputColorSpace	Output color space selection: 0: Rec.601; 1: Rec.709	User setting
m_nGlobalSaturation	Global saturation coefficient, Q7 precision	Yes
m_pSaturationCP	Saturation control parameters with gain (see example saturation CP-gain curve)	Yes	Changes with gain

RGB2YUV Manual Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_bManualMode	Manual mode enable: 0: Auto mode, saturation varies with gain; 1: Manual mode, saturation fixed to manual value		Debug parameter
m_nSaturationManual	Manual saturation coefficient, Q7 precision		Debug parameter

sat_CP-gain control curve is shown below:

CSpecialEffectFirmwareFilter Parameter Description

CSpecialEffectFirmwareFilter module is used for special effect tuning.

SE Enable

Parameter Name	Description	Recommended Tuning	Special Notes
m_bEnable	Special effect enable: 0: Disable special effect 1: Enable special effect with 6 control regions; regions 0-5 have decreasing priority	User setting

SE Parameters

Parameters are divided into zones 0-5, a total of 6 groups. Each group has the same meaning and corresponds to 6 control zones with decreasing priority from zone 0 to zone 5.

Parameter Name	Description	Recommended Tuning	Special Notes
m_bZoneEb_0	Zone 0 special effect enable	User setting
m_pRyTable_0	Brightness rotation parameter, calibration parameter	Calibration result
m_nSyTable_0	Brightness translation parameter, calibration parameter	Calibration result
m_pRuvTable_0	UV rotation parameter, calibration parameter	Calibration result
m_pSuvTable_0	UV translation parameter, calibration parameter	Calibration result
m_pMargin_0	Transition margin band: [Lum_min - Margin_0[0], Lum_max + Margin_0[1]] smooth range; [Hue_min - Margin_0[2], Hue_max + Margin_0[3]] smooth range; [Sat_min - Margin_0[4], Sat_max + Margin_0[5]] smooth range	Yes
m_pYTable_0	Target adjustment area Lum range	Yes
m_pHTable_0	Target adjustment area Hue range	Yes
m_pSTable_0	Target adjustment area Saturation range	Yes

SE Dynamic Control Parameters

GainWeight_0-5 share one GainLut, used to set different intensities for different brightness ranges.

Parameter Name	Description	Recommended Tuning	Special Notes
m_nGainLut	GainWeight segment points, corresponding actual scene value = exposure_time(us) * total_gain(Q8) >> 8
m_pGainWeight_0	Intensity of special effect (see GainWeight-GainLut control curve below). Larger values mean stronger special effect	Yes	Can vary according to GainLut

SE Manual Parameters

Parameters are divided into zones 0-5, total 6 groups, each with the same meaning.

Parameter Name	Description	Recommended Tuning	Special Notes
m_bManualMode_0	Zone 0 manual mode enable	-	Debug parameter
m_nManualGainWeight_0	Special effect intensity in manual mode	-	Debug parameter

CCurveFirmwareFilter Parameter Description

CCurveFirmwareFilter module is used for gamma curve.

Curve Enable

Parameter Name	Description	Recommended Tuning	Special Notes
m_bEnable	Curve enable: 0: Disable gamma curve; 1: Enable gamma curve	User setting

Curve Control Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_bEbGtmAfterLinearcurve	GTM changes with linear curve enable: 0: gtm1 curve unchanged; 1: gtm1 curve changes with linear curve	No
m_nCurveSelectOption	Curve usage selection: 0: Auto calculate based on gain; 1: Use GTMcurve0; 2: Use GTMcurve1; 3: Use GTMcurve2; 4: Use A-Log curve;	No
m_nBacklightStrengthManual	Low-light area brightness adjustment parameter: the larger the value, the more brightness is provided to the low-light area; 0 means no enhancement, and BacklightCurveManual does not affect the final curve	User
m_nContrastStrengthManual	Contrast adjustment parameter: the larger the value, the higher the contrast; 128 means no contrast adjustment and ContrastsCurveManual does not affect the final curve	User setting
m_nBrightnessStrengthManual	Brightness adjustment parameter: the larger the value, the brighter the overall image; 4096 represents 1x brightness	User setting
m_pBacklightCurveManual	Curve used to adjust low-light area	No
m_pContrastsCurveManual	Curve used to adjust contrast	No
m_pGainIndex	Gain segment control points (see Curve-Gain control curve illustration below) Color temperature segment control points. Gain in [128, GainIndex[0]] uses GTMCurve0 curve; Gain in [GainIndex[0], GainIndex[1]] interpolates between GTMCurve0 and GTMCurve1; Gain at GainIndex[1] uses GTMCurve1; Gain in [GainIndex[1], GainIndex[2]] interpolates between GTMCurve1 and GTMCurve2; Gain in [GainIndex[2], 2048] uses GTMCurve2 curve.	Yes
m_pGTMCurve0	Curve 0	Yes	Called based on Gain
m_pGTMCurve1	Curve 1	Yes	Called based on Gain
m_pGTMCurve2	Curve 2	Yes	Called based on Gain

Curve-Gain control curve illustration below:

CLTMFirmwareFilter Parameter Description

CLTMFirmwareFilter module is used for Local Tone Mapping (LTM).

LTM Enable

Parameter Name	Description	Recommended Tuning	Special Notes
m_bEnable	LTM enable: 0: Disable local tone mapping; 1: Enable local tone mapping;	User setting
m_bHistEnable	LTM histogram enable: 0: Disable LTM histogram; 1: Enable LTM histogram	No
m_nOffsetY	Horizontal image offset, determined by input image size	User setting
m_nOffsetX	Vertical image offset, determined by input image size	User setting

LTM Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_nLtmStrength	LTM strength; the larger the value, the stronger the LTM effect	Yes
m_nCurveAlpha	Convergence speed for blending the current frame's curve with the historically saved curve; smaller values converge faster	Yes
m_nSlopeThr	If Drc DrcDark < Thr, equivalent to GTM; if Drc DrcDark >= Thr, each block calculates LTM strength individually	No
m_nPhicBeta	LTM curve attenuation control point; smaller values cause earlier curve attenuation, reducing brightening of mid-high luminance blocks (see PhicBeta diagram)	No
m_nBlendCurveFP	First control point for tone mapping curve based on DRCGain and DrcGainDark; BlendCurveFP <= BlendCurveSP. Before FP, tone mapping strength depends on DRCGain * DrcGainDark. Larger BlendCurveFP means a larger brightening range in dark areas.	No
m_nBlendCurveSP	Second control point; after SP, tone mapping strength depends on DRCGain; smooth transition between FP and SP. Smaller BlendCurveSP means smaller brightening range.	No
m_nSubDarkPercThrLow	Sub-dark area brightening adjustment parameter; smaller values mean stronger brightening in sub-dark areas	No
m_nSubDarkPercThrHigh	Sub-dark area brightening adjustment parameter; smaller values mean stronger brightening; SubDarkPercThrLow < SubDarkPercThrHigh	No
m_nSubDarkAdjMeanThr	Sub-dark area brightening adjustment; combined with SubDarkPercThrLow, SubDarkPercThrHigh, controls LTM strength in sub-dark areas based on percentage of sub-dark points in a block. If current block mean > SubDarkAdjMeanThr and sub-dark percentage > SubDarkPercThrLow, LTM strength increases. Maximum at SubDarkPercThrHigh.	No
m_nDarkPercThrLow	Dark area brightening adjustment parameter; smaller values mean stronger LTM in dark areas	No
m_nDarkPercThrHigh	Dark area brightening adjustment parameter; smaller values mean stronger LTM; DarkPercThrLow < DarkPercThrHigh	No
m_nPhCDarkMaxExtraRatio	Dark area brightening adjustment parameter; combined with DarkPercThrLow and DarkPercThrHigh, controls LTM strength in darkest areas based on point percentage; larger values mean stronger LTM	No
m_pDstAlphaGainIndex	LTM strength gain control nodes	No
m_pDstAlphaIndex	Controls LTM strength based on DstAlphaGainIndex; larger values mean stronger LTM	Yes

PhicBeta diagram below:

CUVDenoiseFirmwareFilter Parameter Description

CUVDenoiseFirmwareFilter module is used for removing color noise.

UVDenoise Enable

Parameter Name	Description	Recommended Tuning	Special Notes
m_bEnable	UVDenoise enable	No	Recommended to use CPP denoise

---

CEELiteFirmwareFilter Parameter Description

CEELiteFirmwareFilter module is used for controlling edge enhancement.

EE Enable

Parameter Name	Description	Recommended Tuning	Special Notes
m_bEnable	EE enable	User setting

CEELite Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_pCurveLumaP	Positive edge luminance-based adaptive enhancement; larger values increase edge strength	Yes
m_pCurveLumaN	Negative edge luminance-based adaptive enhancement; larger values increase edge strength	Yes
m_pCurveFreqTxP	Positive edge frequency-based adaptive enhancement; larger values increase texture detail	Yes
m_pCurveFreqTxN	Negative edge frequency-based adaptive enhancement; larger values increase texture detail	Yes
m_nNodeFreq	Frequency adjustment nodes
m_pGlobal_SharpenStrengthP	Positive edge global sharpening; larger values increase edge strength	Yes
m_pGlobal_SharpenStrengthN	Negative edge global sharpening; larger values increase edge strength	Yes
m_pGlobal_SharpenStrengthPCapture	Capture mode positive edge global sharpening; larger values increase edge strength	Yes
m_pGlobal_SharpenStrengthNCapture	Capture mode negative edge global sharpening; larger values increase edge strength	Yes
m_pLPF1	USMFilter filter parameters	No
m_pLPF2	USMFilter filter parameters; calculated as conv([(flt1[0]-flt2[0]), (flt1[1]-flt2[1]), 2*(flt2[0]+flt2[1] - flt1[0]-flt1[1]), (flt1[1]-flt2[1]), (flt1[0]-flt2[0])])	No
m_pLPF1Capture	Capture mode USMFilter filter parameters	No
m_pLPF2Capture	Capture mode USMFilter filter parameters; same calculation as m_pLPF2	No
m_pFreqExpLevel1	Low-frequency range frequency segmentation control; larger values mean denser segmentation	No
m_pFreqExpLevel2	High-frequency range frequency segmentation control; larger values mean denser segmentation	No
m_pFreqOffsetTx	Frequency soft threshold; larger values mean stronger noise resistance	Yes
m_pTxClipP	Texture region positive edge sharpening clipping; larger values increase sharpening	Yes
m_pTxClipN	Texture region negative edge sharpening clipping; larger values increase sharpening	Yes
m_pTxThrdP	Texture region positive edge sharpening threshold; smaller values increase sharpening	Yes
m_pTxThrdN	Texture region negative edge sharpening threshold; smaller values increase sharpening	Yes
m_pClipPos	Positive edge sharpening clipping; larger values increase sharpening	Yes
m_pClipNeg	Negative edge sharpening clipping; larger values increase sharpening	Yes
m_pHCStrength	Edge halo strength control; smaller values mean stronger halo suppression	Yes
m_pCoffW	Halo region determination parameters	No
m_pGainIndex	Gain control nodes (Q4 format)	No

EE Manual Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_bManualMode	Manual mode enable	-	Debug parameter
Manual suffix parameters	Manual mode parameters, same as automatic mode	-	Debug parameter

CBitDepthCompressionFirmwareFilter Description

The CBitDepthCompressionFirmwareFilter module is used for bit depth compression.

Dithering Enable

Parameter Name	Description	Recommended Tuning	Special Notes
m_bEnableDithering	Dithering enable: 0: Disable dithering; 1: Enable dithering	No

CFormatterFirmwareFilter Description

The CFormatterFirmwareFilter module is used for output format control.

Formatter Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_nOutputFormat	Output format: 0: NV12; 1: P010; 2: RGB888	No
m_nSwingOption	- 0: full swing; 1: studio swing	No
m_bConvertDitheringEnable	- 0: rgb2yuv dithering disable; 1: rgb2yuv dithering enable	No
m_bCompressDitheringEnable	- 0: full swing to studio swing dithering disable; 1: full swing to studio swing dithering enable	No

CEISFirmwareFilter Description

The CEISFirmwareFilter module is used for electronic image stabilization (EIS).

EIS Enable

Parameter Name	Description	Recommended Tuning	Special Notes
m_bEnable	EIS enable: 0: Disable EIS; 1: Enable EIS	User setting

EIS Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_bCalcEnable	Firmware calculation enable 0: Disable; 1: Enable	User setting
m_bFilterEnable	Motion smoothing filter enable 0: Disable motion smoothing filter (usually off when device/scenario motion is minimal) ; 1: Enable motion smoothing filter	Yes
m_nRangeX	Detection range of movement in X direction (unit: pixels)	No
m_nRangeY	Detection range of movement in Y direction (unit: pixels)	No
m_nMarginX	Offset X of LDC crop window	No
m_nMarginY	Offset Y of LDC crop window	No
m_pFilterWeight	Motion smoothing filter weights	Yes
m_pPeakConfLevel	Confidence levels	Yes
m_pPeakErrorThre	Error tolerance thresholds corresponding to confidence levels	Yes
m_bReverseDirectionX	Horizontal direction reverse control	Yes
m_bReverseDirectionY	Vertical direction reverse control	Yes
m_nCenterStatRatio	Center statistics ratio, 256 equals 100%	Yes

CAECFilter Parameters Description

CAECFilter (AEC) module is used for Auto Exposure Control.

AE Basic Parameters

Parameter Name	Description	Recommended Tuning	Special Notes
m_bManualAEEnable	AEC mode selection: 0: Auto mode; 1: Manual mode, exposure value uses manual parameters; 2: Lock mode, exposure value locked, no AEC adjustment; 3: Manual exposure index mode, exposure locked to m_pExpIndexManual	User setting
m_pMinExpTime	Minimum exposure time (parameters with “short” annotations are reserved; applies to AE parameters here)	Yes
m_pMaxExpTime	Maximum exposure time	Yes
m_pMinAnaGain	Minimum analog gain	Yes
m_pMaxAnaGain	Maximum analog gain	Yes
m_pMinSnsDGain	Minimum sensor digital gain	Yes
m_pMaxSnsDGain	Maximum sensor digital gain	Yes
m_pMinTotalGain	Minimum total gain	Yes
m_pMaxTotalGain	Maximum total gain	Yes
m_pRouteNode50Hz	50Hz exposure table, first column: exposure time, second: total gain, third: precise gain control (reserved)	Yes
m_pRouteNode60Hz	60Hz exposure table, first column: exposure time, second: total gain, third: precise gain control (reserved)	Yes
m_pMeteringMatrix	16x12 metering weight matrix	User setting
m_bLumaCalcMode	Luma calculation method 0: RGB max; 1: Y average;	User setting
m_bQuickResponseEnable	AE quick response enable; when enabled, AE adjusts immediately	User setting

AE Target Brightness Control

Parameter Name	Description	Recommended Tuning	Special Notes
m_bAdvancedAECEnable	Target brightness control parameter: 0: Use m_pTargetRange[1] as target brightness; 1: Dynamically calculate target brightness in range [m_pTargetRange[0], m_pTargetRange[1]]	User setting
m_nCompensation	Target brightness compensation factor, 100 means 1x	Yes
m_nSensitivityRatio	Sensitivity ratio between capture and preview, 256 means 1x	User setting
m_pSatRefBin	Saturation reference bin; Target constrained to histogram pixels ratio in range [m_pSatRefBin[0], 255] falling within [m_pSatRefPerThr[0][0]/10000%, m_pSatRefPerThr[0][1]/10000%]	Yes
m_pSatRefPerThr	Saturation ratio upper and lower limits	Yes
m_pExpIndexThre	Exposure threshold, corresponds to 4 groups of m_pLumaBlockWeight, used to select which group participates in weight calculation (see Exp_index-luma_weight diagram)	User setting
m_pLumaBlockWeight	Four groups of luminance block weight tables, each group: Weight[i][0]: low luminance Weight[i][1]: mid luminance Weight[i][2]: high luminance (see Exp_index-luma_weight and Luma-Weight diagrams)	User setting
m_pLumaBlockThre	Luminance block threshold, controls weight curve of luminance blocks (see Luma-Weight diagram)	User setting
m_pLuma7ZoneWeight	Weight table for statistics window and 6 sub-ROIs	User setting
m_pTargetRange	Target brightness range	Yes
m_pTargetDeclineRatio	Target brightness adjustment by gain; gain nodes are powers of 2	Yes	Can vary with Gain

---

Exp_index – luma_weight diagram:

Luma – weight diagram:

AE Mode Control

Parameter Name	Description	Recommended Tuning	Special Notes
m_bLumaSelectOption	0: Luminance calculation does NOT reference 6 sub-ROIs in AEM; 1: Luminance calculation references 6 sub-ROIs in AEM
m_nStrategyMode	AE adjustment strategy selection: 0: Auto mode; 1: Highlight priority (ensure proper exposure in bright areas) 2: Low light priority (ensure proper exposure in dark areas)
m_bAntiFlickerEnable	Anti-flicker enable: 0: Disabled, exposure time not limited by flicker; 1: Enabled, exposure time limited to integer multiples of flicker period
m_nAntiFlickerFreq	Anti-flicker frequency selection: 0: 50Hz; 1: 60Hz

AE Convergence Control

Parameter Name	Description	Recommended Tuning	Notes
m_nTolerance	Target brightness error tolerance. AE will not adjust when the measured brightness is within [target±m_nTolerance]	Yes
m_nStableRange	Stable brightness range: when AE converges, if the difference between the measured brightness (mean) and the target brightness exceeds m_nStableRange, analog gain is adjusted to compensate	Yes
m_nSmoothness	AE convergence smoothness; smaller values result in smoother convergence	Yes
m_pStableRefFrame	Number of stable reference frames; if the count of stable frames exceeds this, AE considers the scene stable	Yes
m_pUnStableRefFrame	Number of unstable reference frames; if the count of unstable frames exceeds this, AE considers the scene unstable	Yes
m_nDualTargetBlendWeight	Blend ratio between the previous target brightness and the current target brightness; 8 means 1:1 blend; 16 means fully use current target	Yes
m_pFastStep	Large step adjustment discount ratio. When m_pFastStep = 256 (and not limited by MaxFastRatio), it means a single step adjustment to the target (see step-target diagram below)	Yes
m_pFastStepRangePer	Large step threshold; large steps used when condition	current_luma – target_luma	> m_nFastStepRangePer * target_luma is met
m_pFastStepMinRange	Lower limit of the large step threshold	Yes
m_pSlowStep	Small step adjustment discount ratio (same as m_pFastStep, see step-target diagram below)	Yes
m_pSlowStepRangePer	Small step threshold; used when m_nSlowStepRangePer * target <	current_luma – target	< m_nFastStepRangePer * target
m_pSlowStepMinRange	Lower limit of the small step threshold	Yes
m_pFineStep	Fine step adjustment discount ratio (same as m_pFastStep, see step-target diagram below)	Yes
m_pMaxFastRatio	Maximum large step adjustment ratio (Q8): When increasing exposure, max large step ratio is m_pMaxFastRatio; When decreasing exposure, max large step ratio is 65536/m_nMaxFastRatio;	Yes
m_pMaxSlowRatio	Maximum small step adjustment ratio (Q8): When increasing exposure, small step ratio limits range is [256 + m_nMaxFineRatio, 256 + m_nMaxSlowRatio]; When decreasing exposure, range is [256 - m_nMaxSlowRatio, 256 - m_nMaxFineRatio];	Yes
m_pMaxFineRatio	Maximum fine step adjustment ratio (Q8)	No
m_pMinFineRatio	Minimum fine step adjustment ratio When increasing exposure, small step ratio limits range is [256 + m_nMinFineRatio, 256 + m_nMaxFineRatio]; When decreasing exposure, range is [256 - m_nMaxFineRatio, 256 - m_nMinFineRatio];	No
m_pAjustSplitFrameNum	Number of steps required for a single adjustment (1st column reserved)	No
m_pSingleStepAdjustLumaThr	Single adjustment brightness threshold (1st column reserved). Together with m_pAjustSplitFrameNum determines max luma per single adjustment: Single_luma_max = max((target - mean)/AjustSplitFrameNum, SingleStepAdjustLumaThr) When increasing exposure, if adjusted luma exceeds Single_luma_max, then adjustRatio= 256 + Single_luma_max * 256 / mean Similarly for decreasing exposure.	No
m_pFaceAjustSplitFrameNum	Same meaning as m_pAjustSplitFrameNum, but only effective in face AE mode	No
m_pFaceSingleStepAdjustLumaThr	Same meaning as m_pSingleStepAdjustLumaThr, but only effective in face AE mode	No

Step–Target Illustration:

Luma–Step Illustration:

Automatic Dynamic Range Compensation Gain Calculation

Parameter Name	Description	Recommended for Debugging	Notes
m_nDarkRefBin	Upper limit reference value of pixel brightness in dark areas; the brightness range for dark area statistics will not exceed this value	Yes
m_bLumaPredict	Luma prediction for DRCgain calculation: 0: No correction during DRCgain calculation (when frame statistics are applied to current frame’s LTM) ; 1: Correction applied during DRCgain calculation (when frame statistics are applied to next frame’s LTM)
m_nDarkPerThrL	Lower threshold of dark area pixel percentage. When the percentage of dark pixels is below this value, DRCGainDark is set to 1 (see illustration: pixelNumPercent–DRCGainDark)
m_nDarkPerThrH	Upper threshold of dark area pixel percentage. When the percentage of dark pixels exceeds this value, all dark pixels are used to calculate DRCGainDark (see illustration: pixelNumPercent–DRCGainDark)
m_nDarkTarget	Target brightness value for dark areas. The larger the value, the higher the resulting DRCGainDark
m_nMaxDRCGain	Maximum DRC gain	Yes
m_nMaxDRCGainDark	Maximum DRC gain for dark areas	Yes

PixelNumPercent–DRCGainDark Illustration:

Lux Calibration

Parameter Name	Description	Recommended for Debugging	Notes
m_nCalibExposureIndex	Exposure index for brightness calibration	Yes
m_nCalibSceneLum	Scene brightness for calibration	Yes
m_nCalibSceneLux	Actual illumination (lux) of the calibration scene	Yes

AEC Manual Parameters

Parameter Name	Description	Recommended for Debugging	Notes
m_pExpTimeManual	Manual exposure time, effective when AEC is in manual mode
m_pAnaGainManual	Manual analog gain, effective when AEC is in manual mode
m_pSnsDGainManual	Manual sensor digital gain, effective when AEC is in manual mode
m_pTotalGainManual	Manual total gain, effective when AEC is in manual mode
m_pExpIndexManual	Manual exposure index, effective when AEC is in manual mode

AEC Frame Info

Parameter Name	Description	Recommended for Debugging	Notes
m_nFrameCounter	AE frame counter	-	Read-only
m_pExpTimeLong	Exposure time (us)	-	Read-only
m_pAnaGainLong	Analog gain (Q8)	-	Read-only
m_pSnsDGainLong	Sensor digital gain (Q12)	-	Read-only
m_pIspDGainLong	ISP digital gain (Q12)	-	Read-only
m_pTotalGainLong	Total gain (Q8)	-	Read-only
m_pExpIndexLong	Exposure index (Exposure Time * Total Gain / 256)	-	Read-only
m_pTotalGainDBLong	Total gain (dB)	-	Read-only
m_pExpTimeShort	Reserved	-	Read-only
m_pAnaGainShort	Reserved	-	Read-only
m_pSnsDGainShort	Reserved	-	Read-only
m_pIspDGainShort	Reserved	-	Read-only
m_pTotalGainShort	Reserved	-	Read-only
m_pExpIndexShort	Reserved	-	Read-only
m_pTotalGainDBShort	Reserved	-	Read-only
m_nCurHdrRatio	Reserved	-	Read-only
m_nHistPixelNum	Number of pixels used in histogram statistics	-	Read-only
m_pMeanLuma	AE calculated brightness [0]: long [1]: Reserved	-	Read-only
m_pTargetLuma	AE target brightness [0]: long [1]: Reserved	-	Read-only
m_pLuma7Zone	AE main and 6 sub-window luma	-	Read-only
m_pAERouteNum	AE exposure node [0]: long [1]: Reserved	-	Read-only
m_pRouteNodeLong	Long frame exposure table	-	Read-only
m_pADNodeLong	Sensor analog and digital gain allocation table	-	Read-only
m_pLumaMatrixLong	Luma thumbnail	-	Read-only
m_pHistLong	Histogram	-	Read-only
m_pRouteNodeShort	Reserved	-	Read-only
m_pADNodeShort	Reserved	-	Read-only
m_pLumaMatrixShort	Reserved	-	Read-only
m_pHistShort	Reserved	-	Read-only
m_pRefSaturateNum	Reference overexposed pixel count [0]: long [1]: Reserved	-	Read-only
m_pCurSaturateNum	Current overexposed pixel count [0]: long [1]: Reserved	-	Read-only
m_pEstSaturateNum	Estimated overexposed pixel count when brightness reaches upper limit [0]: long [1]: Reserved	-	Read-only
m_pAdjustRatio	AE adjustment ratio	-	Read-only
m_pStableFlag	AE stability flag [0]: long [1]: Reserved	-	Read-only
m_pStableFlagBuf	AE frame sequence stability flag [0]: long [1]: Reserved	-	Read-only
m_pUnStableFlagBuf	AE frame sequence instability flag [0]: long [1]: Reserved	-	Read-only
m_nDRCGain	DRCGain result of AE calculation for current frame	-	Read-only
m_nDRCGainDark	DRCGainDark result of AE calculation for current frame	-	Read-only

CAFFilter Parameter Description

The CAFFilter module is used for automatic focus control.

AF Parameters

Parameter Name	Description	Recommended for Debugging	Notes
m_nAFMode	Focus mode: 0: SAF (Single Auto Focus); 1: CAF (Continuous Auto Focus)	User configurable
m_bHybridAFEnable	Enable hybrid AF mode (CDAF + PDAF): 0: Disabled; 1: Enabled	User configurable
m_bAFTrigger	AF trigger; must be enabled when using CAF	User configurable
m_bMotorManualTrigger	Manual focus trigger, used together with ManualMotorPosition	User configurable
m_nManualMotorPosition	Manual motor position	User configurable
m_nFrameRate	Current sensor frame rate, used to calculate AF skip frames; should be updated by software	User configurable
m_nMaxSkipFrame	Maximum number of skipped frames	No

Motor Property Parameters

Parameter Name	Description	Recommended for Debugging	Notes
m_nMinMotorPosition	Minimum motor position	Yes
m_nMaxMotorPosition	Maximum motor position	Yes
m_nMotorResponseTime	Response time per motor step (ms); used to calculate AF skipped frames — the higher the value, the more frames are skipped	Yes
m_nHyperFocalMotorPosition	Hyperfocal position	Yes

Motor Movement Control Parameters

Parameter Name	Description	Recommended for Debugging	Notes
m_nFirstStepDirectionJudgeRatio	First step direction judgment ratio (%). When AF algorithm is running in CDAF mode, this sets the percentage used to determine the first step direction: If the motor's initial position is within the first X% of its full range, move from min to max position; If it's within the last (100%-X)%, move from max to min position.	User Setting
m_bPreMoveMode	Pre-move mode. When AF algorithm is running in CDAF mode, defines motor behavior for first step: 0: Stay at initial position; 1: Move to min or max motor position based on direction judgment percentage.	User Setting
m_bPreMoveToInfMode	Pre-move to infinity mode. When AF algorithm is running in CDAF mode, defines motor behavior for first step: 0: Stay at initial position; 1: Move to minimum motor position.	User Setting
m_bStartFromTrueCurrent	Start from actual current position. When AF algorithm is running in CDAF mode, defines motor behavior for first step: 0: Move to closest macro position from initial position; 1: Stay at initial position.	User Setting
m_nBackTimeRatio	Frame skipping ratio during motor movement. Higher value means more frames are spent per motor step. Works in conjunction with m_nBackTimeDivisor.	User Setting
m_nBackTimeDivisor	Motor step frame skipping control divisor. Smaller value means more frames are spent per step. Works in conjunction with m_nBackTimeRatio.	User Setting
m_nPreMoveTimeRatio	Frame skipping ratio in pre-move mode.	User Setting
m_nFailMotorPositionOption	Final motor position after focus failure: 0: Auto-selected by algorithm (requires correct hyperfocal position); 1: Best sharpness; 2: Infinity; 3: Hyperfocal; 4: Macro; 5: Current position	User Setting

Motor Step Control Parameters

Parameter Name	Description	Recommended for Debugging	Notes
m_nMinMacroReverseStepNum	Minimum steps motor must move when reversing direction from max to min position upon detecting FV drop	Yes
m_nMinReverseStepNum	Minimum steps motor must move when reversing direction from min to max position upon detecting FV drop	Yes
m_nFineStepSearchNum	Number of fine step searches	Yes
m_nCoarseStep	Coarse step size in CDAF	Yes
m_nFineStep	Fine step size in CDAF	Yes

Parameter Name	Description	Recommended for Debugging	Notes
m_nSoftwareMotorCtrlMode	Target position control mode (used for first-step pre-move, manual move, fine step start, focus failure handling, etc.): 0: Soft landing mode; 1: Fixed step mode; 2: Direct mode	User Setting	See Figure 1
m_nMinSafePosition	Minimum safe position when motor moves from max to min position in soft landing mode	User Setting	See Figure 2
m_nMinStepRatio	Minimum step ratio in soft landing mode (same as above)	User Setting	See Figure 2
m_nMaxSafeStep	Maximum safe step in soft landing mode (same as above)	User Setting	See Figure 2
m_nMinSafePositionMacro	Minimum safe position when motor moves from min to max position in soft landing mode	User Setting	See Figure 2
m_nMinStepRatioMacro	Minimum step ratio in soft landing mode (same as above)	User Setting	See Figure 2
m_nMaxSafeStepMacro	Maximum safe step in soft landing mode (same as above)	User Setting	See Figure 2

Figure 1

Description per condition group:

• Graphic: Visual demonstration
• Output: One output scenario

Figure 2

Parameter Name	Description	Recommended for Debugging	Notes
m_bAdaptiveStep	Adaptive step mode 0: Disabled; 1: Enabled	User Setting
m_nVCMStep1Ratio	Motor step ratio 1, divides motor range into segments in adaptive step mode	User Setting	See Figure 3
m_nVCMStep2Ratio	Motor step ratio 2, divides motor range into segments in adaptive step mode	User Setting	See Figure 3
m_nVCMCoarseStep1	Motor coarse step 1, actual step computed based on current motor position segment in adaptive mode	User Setting	See Figure 3
m_nVCMCoarseStep2	Motor coarse step 2, actual step computed based on current motor position segment in adaptive mode	User Setting	See Figure 3
m_nVCMFineStep1	Motor fine step 1, actual step computed based on current motor position segment in adaptive mode	User Setting	See Figure 3
m_nVCMFineStep2	Motor fine step 2, actual step computed based on current motor position segment in adaptive mode	User Setting	See Figure 3

• MotorMoveStep indicates the real-time step length, changing according to motor movement direction and whether in coarse or fine step mode; it can be positive, negative, large, or small.

Figure 3

Description per condition group:

• Graphic: Visual demonstration
• Output: One output scenario

Focus Value (FV) Judgment Parameters

Parameter Name	Description	Recommended for Debugging	Notes
m_pFVDropPercentage	FV drop percentage threshold for pre-move mode. Calculated separately for 6 types of FV based on gain.	No
m_pCurrentStartFVDropPercentage	FV drop percentage threshold for non-pre-move mode. Calculated separately for 6 types of FV based on gain.	No
m_pFVFailPercentage	In pre-move mode, max-min FV difference percentage to judge FV curve invalidity. If difference is below this threshold, FV curve is too flat and considered invalid. Applied separately for 6 FV types based on gain. First condition to judge focus failure.	No
m_pCurrentStartFVFailPercentage	Same as above but for non-pre-move mode. First condition to judge focus failure.	No
m_pLastStepChangePercentage	Percentage threshold for judging FV curve as invalid when FV continuously rises or falls. Calculated separately for 6 FV types based on gain.	No
m_pWindowWeightMatrix	5x5 window weights for FV statistics calculation.	No

PDAF Control Parameters

Parameter Name	Description	Recommended for Debugging	Notes
m_bPDReverseDirectionFlag	PDShift step direction reverse flag 0: No reverse; 1: Reverse	Yes
m_nPDDirectConfThrRatio	PD threshold percentage; when confidence exceeds m_nPDDirectConfThrRatio%, directly apply PD step calculated	No
m_nPDTryStepRatio	Step ratio tried first in hybrid AF mode relative to CDAF coarse step	No
m_nPDCoarseStep	PDAF coarse step	Yes
m_nPDFineStep	PDAF fine step	Yes
m_nPDStepDiscountRatio	Discount ratio applied to PDShift calculated step	No
m_pPDConfThr	PD confidence threshold	Yes
m_pPDFVIncreaseRatio	FV increase ratio used in hybrid AF mode to judge if FV rises after PD step, determining PD info validity	No
m_pPDFVDropPercentage	FV drop percentage threshold used in PDAF mode	No
m_pPDShiftPositionLUT_0_0 ~ m_pPDShiftPositionLUT_4_4	5x5 window PD Shift – step lookup table	PDAF plugin calibration

Continuous Autofocus (CAF) Control

Parameter Name	Description	Recommended for Debugging	Notes
m_bCAFHold	Hold current state in continuous autofocus mode	User setting
m_bCAFForce	Force trigger single autofocus once in continuous autofocus mode	User setting
m_nCAFHoldPDShiftThr	PD shift threshold to judge scene stability in CAF mode; higher value means easier to judge stable	Yes
m_nCAFHoldPDConfThr	PD confidence threshold to judge scene stability in CAF mode; lower value means easier to judge stable	Yes
m_nPreFocusMode	Prefocus mode enable in CAF: 0: No prefocus; 1: Semi-prefocus (try focus once based on PD shift)	User setting
m_nPFPDStepDiscountRatio	Prefocus step discount ratio	Yes
m_nLumaCalcOpt	Luma calculation method 0: Calculate each channel separately; 1: Calculate RGB max value; 2: Calculate weighted Y average	No
m_bReFocusEnable	Refocus enable in CAF mode	User setting
m_pRefocusLumaSADThr	SAD threshold to decide whether to trigger SAF when CAF triggers SAF	Yes
m_nFlatSceneVarThr	Variance threshold of thumbnail to judge flat scene in CAF mode	Yes
m_nFlatSceneLumaThr	Luma threshold to judge flat scene in CAF mode	Yes
m_bStableJudgeOpt	Condition for scene stability judgment in CAF: 0: FV or luma satisfies condition; 1: Both FV and luma satisfy condition	User setting
m_nRefStableFrameNum	Number of frames judged stable in reference state during CAF	User setting
m_nDetStableFrameNum	Number of frames judged stable in detection state during CAF	User setting
m_pStableExpIndexPercentage	Percentage of exposure index change to judge stability in CAF	Yes
m_pStableFVSADPercentage	SAD percentage of FV to judge stability in CAF	Yes
m_pStableLumaSADThr	SAD threshold of luma to judge stability in CAF	Yes
m_bChangeJudgeOpt	Condition for scene change judgment in CAF: 0: FV or luma satisfies condition; 1: Both FV and luma satisfy condition	No
m_nChangeFrameNum	Number of frames to judge scene change in CAF	Yes
m_nChangeStatAreaPercentage	Threshold percentage of AF statistic window area change to judge scene change in CAF; effective in face mode	No
m_nChangeStatCenterPercentage	Threshold percentage of AF statistic window center position change to judge scene change in CAF; effective in face mode	No
m_pChangeExpIndexPercentage	Exposure index change percentage to judge scene change in CAF	Yes
m_pChangeFVSADPercentage	SAD percentage of FV to judge change in CAF	Yes
m_pChangeLumaSADThr	SAD threshold of luma to judge change in CAF	Yes
m_pChangePDShiftThr	PDShift percentage threshold to judge change in CAF	No

Foreground Tracking Mode

Parameter Name	Description	Recommended for Debugging	Notes
m_bForeGroundTrackInCoarse	Foreground tracking enable during coarse step adjustment: 0: Disable foreground tracking; 1: Enable foreground tracking	User setting
m_bForeGroundTrackInFine	Foreground tracking enable during fine step adjustment: 0: Disable foreground tracking; 1: Enable foreground tracking	User setting
m_nForeGroundTrackWindow	Foreground tracking window size: 3: 3x3; 4: 4x4; 5: 5x5	User setting
m_bForeGroundTrack	Foreground tracking mode enable: 0: Disable foreground tracking; 1: Enable foreground tracking	User setting
m_pFVBlkDropPercentage	FV value drop percentage threshold when foreground tracking mode is enabled	No
m_pFVBlkIncreasePercentage	FV value increase percentage threshold when foreground tracking mode is enabled	No
m_pFineStepStartPosSelectConf	Confidence threshold for start position when beginning fine step in foreground tracking mode	No
m_pFineStepEndPosSelectConf	Confidence threshold for final position after fine step in foreground tracking mode	No

AF Frameinfo

Parameter Name	Description	Recommended for Debugging	Notes
m_nGain	Current system gain, Q4 format	-	Read-only
m_nPDShift	PD offset	-	Read-only
m_nPDConf	PD confidence	-	Read-only
m_nCurrentPosition	Current motor position	-	Read-only
m_nFSMStatus	AF state machine status	-	Read-only
m_nFocusStatFlag	Focus status flag 0: Initial; 1: Success; 2: Failure	-	Read-only
m_nFocusFrameCost	Number of frames taken for a single focus	-	Read-only
m_nMotorMoveStep	Real-time motor step	-	Read-only
m_bCoarseFineStepFlag	Real-time coarse/fine step flag 0: Fine step; 1: Coarse step	-	Read-only
m_nMotorMoveSkipFrameNum	Number of frames skipped per motor move	-	Read-only
m_bStartFromMinFlag	Real-time motor movement direction 0: From max to min motor position; 1: From min to max	-	Read-only
m_bStartFromCurrentFlag	Motor initial movement from current position flag 0: From pre-move position; 1: From initial position	-	Read-only
m_bSoftwareMotorCtrlFlag	Flag for software control during movement from current to target position	-	Read-only
m_nPositionScanNum	Count of motor movement direction scans	-	Read-only
m_nPositionAdjustNum	Count of motor moved positions	-	Read-only
m_pRTFVList	Real-time FV (Focus Value) list	-	Read-only
m_pMinFVList	Minimum FV values during single focus process	-	Read-only
m_pMaxFVList	Maximum FV values during single focus process	-	Read-only
m_pMaxFVPositionList	Positions of maximum FV values during single focus process	-	Read-only
m_pFVUnchangeFlag	FV curve invalidation flag 0: FV curve valid; 1: FV curve invalid	-	Read-only
m_pFailCondition	Status of three local focus fail conditions 0: Success; 1: Fail	-	Read-only
m_pFSMPosSingleFocus	Frame number - motor position - state machine info during single focus	-	Read-only
m_pPosFvSingleFocus	Frame number - 6 types of FV values during single focus	-	Read-only
m_pRefFV	Reference FV value in continuous focus mode	-	Read-only
m_pSeqFV	Real-time FV sequence in continuous focus mode	-	Read-only
m_pRefLumaR	Reference R channel thumbnail in continuous focus mode	-	Read-only
m_pRefLumaG	Reference G channel thumbnail in continuous focus mode	-	Read-only
m_pRefLumaB	Reference B channel thumbnail in continuous focus mode	-	Read-only
m_pSeqLumaR	Real-time R channel thumbnail in continuous focus mode	-	Read-only
m_pSeqLumaG	Real-time G channel thumbnail in continuous focus mode	-	Read-only
m_pSeqLumaB	Real-time B channel thumbnail in continuous focus mode	-	Read-only
m_nRefExpIndex	Reference exposure index in continuous focus mode	-	Read-only
m_nRTExpIndex	Real-time exposure index in continuous focus mode	-	Read-only
m_bPDInfoDisable	Flag for PD info usage in continuous focus mode	-	Read-only
m_nLumaAverage	Average luminance	-	Read-only
m_nSceneVariance	Real-time variance of thumbnail	-	Read-only
m_bSceneFlatFlag	Scene flatness flag	-	Read-only
m_nChangeExpIndex	Difference between current and reference exposure index in continuous focus mode	-	Read-only
m_nChangeExpIndexThr	Exposure index threshold for scene change detection in continuous focus mode	-	Read-only
m_nChangeFVSAD	SAD difference between real-time and reference FV in continuous focus mode	-	Read-only
m_nChangeFVSADThr	SAD threshold for scene change detection in continuous focus mode	-	Read-only
m_nChangeLumaSAD	Real-time luminance SAD value	-	Read-only
m_nChangeLumaSADThr	Real-time luminance SAD threshold	-	Read-only
m_nRefocusLumaSAD	Previous frame luminance SAD when CAF triggers SAF	-	Read-only
m_nRefocusLumaSADThr	SAD threshold deciding whether SAF is triggered by CAF	-	Read-only
m_bStableExpIndexFlag	AE exposure table stability flag	-	Read-only
m_bStableFVFlag	FV stability flag	-	Read-only
m_bStableAFLumaFlag	AFM luminance stability flag	-	Read-only
m_bStableAELumaFlag	AEM luminance stability flag	-	Read-only
m_bPDTriggerCAF	PD triggered AF flag	-	Read-only
m_bFVTriggerCAF	FV triggered AF flag	-	Read-only
m_bLumaTriggerCAF	Luma triggered AF flag	-	Read-only
m_bExpIndexTriggerCAF	Exposure index triggered AF flag	-	Read-only
m_bRefocusTriggerCAF	CAF triggered SAF flag	-	Read-only

CAWBFilter Parameter Description

The CAWBFilter (AWB) module is used for automatic white balance control.

Gray World

Parameter Name	Description	Suggested Tuning	Special Notes
m_bGrayWorldEn	Gray World mode enable: 0: Disable Gray World; 1: Enable Gray World; in this mode, white balance gains are calculated by the Gray World algorithm	User Setting

ROI Control

Parameter Name	Description	Suggested Tuning	Special Notes
m_pRoiBound	Control for 16 ROI regions	Yes
m_pRoiLumLutHigh	Upper limit of brightness for statistic blocks	No
m_pRoiLumLutLow	Lower limit of brightness for statistic blocks	No
m_pRoiEn	ROI enable	Yes
m_pRoiWeightLut	Weights corresponding to different brightness indices	Yes
m_pRoiLumThre	Brightness lux index threshold	Yes
m_bRoiLimitManual	Fixed ROI enable: 0: Calculate white point limit area based on auto white point ROI and low color temperature block ratio; 1: White point limit area fixed to RoiCtHigh, RoiCtLow, RoiXMax, RoiXMin
m_nRoiCtHigh	Upper Y coordinate of white point ROI; weights for blocks outside ROI area are zero when calculating white balance (CT = BlockBlue_sum(8bit) / BlockRed_sum * 1024)	Yes
m_nRoiCtLow	Lower Y coordinate of white point ROI	Yes
m_nRoiXMax	Upper X coordinate of white point ROI (X = BlockBlue_sum x BlockRed_sum / BlockGreen_sum / BlockGreen_sum x 1024)	Yes
m_nRoiXMin	Lower X coordinate of white point ROI	Yes
m_pRoiCtHighAuto	Upper Y coordinate of white point ROI at different brightness levels	Yes
m_nRoiCtLowAuto	Lower Y coordinate of auto white point ROI at different brightness levels	Yes
m_nRoiXMaxAuto	Upper X coordinate of auto white point ROI at different brightness levels	Yes
m_nRoiXMinAuto	Lower X coordinate of auto white point ROI at different brightness levels	Yes
m_nValidNum	Minimum number of valid blocks in each ROI; if the number of blocks in the brightness statistics range is less than this value, the ROI does not participate in white balance calculation	No
m_bWeightOnSum	Block weight calculation method: 0: Use total brightness sum of blocks; 1: Use RGB gain of blocks	No

White Balance Gain Control

Parameter Name	Description	Suggested Tuning	Special Notes
m_pAWBGainLimit	Final WB gain limitation range	Yes
m_nAWBCTShift	WB Gain upward shift distance, used to shift low color temperature scenes toward warmer tones	Yes
m_pAWBCTShiftThr	WB Gain shift Y threshold: When Y < AWBCTShiftThr[0], WB Gain is shifted up by m_nAWBCTShift When Y > AWBCTShiftThr[1], WB Gain is not shifted Intermediate values are interpolated	Yes

Mixed Lighting Scene - Reduce High Color Temperature Weight

Parameter Name	Description	Suggested Tuning	Special Notes
m_sRoiBoundDayLight	Mixed high and low color temperature scene, bounding region for high color temperature statistic blocks; blocks falling in this area are counted as high color temperature blocks	Yes
m_pLowCtNumThr	Threshold for the ratio (per mille) of low color temperature statistic blocks (see the illustration below lowCtLightPermillage-ratio) When the ratio of low CT blocks > LowCtNumThr[0], the weight of high CT blocks begins to decrease; When the ratio ≥ LowCtNumThr[1], the weight of high CT blocks is reduced to the minimum.	Yes
m_pDayLightNumThr	Threshold for the ratio (per mille) of high color temperature statistic blocks (see the illustration below dayLightPermillage-ratio) When the ratio of high CT blocks < DayLightNumThr[1], the weight of high CT blocks begins to decrease; When the ratio ≤ DayLightNumThr[0], the weight no longer decreases. Works together with m_pLowCtNumThr to determine the base strength (baseRatio) for reducing high CT block weights.	Yes
m_pLowCtThr	Y threshold for reducing the weight of high CT statistic blocks in mixed CT scenes (see CtThr – ProtectRatio illustration); blocks with Y < m_pLowCtThr[1] are counted as low CT blocks	Yes
m_pLowCtProtectRatio	Coefficient for reducing the weight of high CT blocks; the smaller the value, the lower the weight of high CT blocks. Together with baseRatio determines block weight (see CtThr – ProtectRatio)	Yes	Changes with CtThr
m_nLog2CwtOverA	Weight ratio of CWF to A light	No

---

CtThr – ProtectRatio illustration

lowCtLightPermillage-ratio illustration

dayLightPermillage-ratio illustration

Green Zone Control

Parameter Name	Description	Suggested Tuning	Special Notes
m_pRoiBoundG	Green zone ROI control	Yes
m_pRoiLumLutHighG	Upper brightness limit for green zone statistic blocks	Yes
m_pRoiLumLutLowG	Lower brightness limit for green zone statistic blocks	Yes
m_bGreenShiftEn	Green zone ROI enable	User setting
m_pGreenLumThre	Weight table for different brightness indices in the green zone	Yes
m_nGreenShiftMax	Max weight for green zone: multiplied by exposure shift to get the final shift weight; max 32, fully biased toward outdoor gain	Yes
m_pGreenNumThre	Threshold for green zone statistic blocks count; if the number of blocks in green zone is within this range, green shift is applied	Yes
m_pOutdoorGain	Target WB gain for green zone	Yes

AWB Frameinfo

Parameter Name	Description	Suggested Tuning	Special Notes
m_bAwbCalStatus	AWB calculation status	-	Read-only
CurrentResult	Currently applied white balance gain	-	Read-only
CurrentTarget	Target white balance gain calculated by AWB	-	Read-only
m_nSceneLux	AWB debug parameter, current AE calculated scene illuminance	-	Read-only
m_nValidBlkNum	Number of valid statistic blocks	-	Read-only
m_nLowCtBlkRatio	Low color temperature statistic block ratio (per mille)	-	Read-only
m_pCtLimit	Current ROI range (X_Max, X_Min, CT_High, CT_Low)	-	Read-only
m_nDayLightBlkRatio	Daylight statistic block ratio (per mille)	-	Read-only
m_nGreenShiftWeight	Green zone shift weight	-	Read-only
m_pStatBlkR	R channel statistics	-	Read-only
m_pStatBlkG	G channel statistics	-	Read-only
m_pStatBlkB	B channel statistics	-	Read-only
m_pBlkCtX	Statistic block horizontal coordinate X	-	Read-only
m_pBlkCtY	Statistic block vertical coordinate Y	-	Read-only
m_pBlkRoiId	ROI of statistic block	-	Read-only
m_pBlkWeight	Weight of statistic block for white balance calculation	-	Read-only
m_pDebugBlcokData[23][31]	RGB values of AWB statistic blocks	-	Read-only

C3DNRFirmwareFilter Debug Description

C3DNRFirmwareFilter (NR) module is used for 2D and 3D noise reduction.

NR Enable

Parameter Name	Description	Suggested Modification	Special Notes
m_nTnrEb	3DNR enable, controls 3DNR only	User setting
m_pTnr3dYEb	3DNR luminance noise reduction enable	Yes, adjustable by Gain/Layer
m_pTnr3dUVEb	3DNR chrominance noise reduction enable	Yes, adjustable by Gain/Layer
m_pTnr2dYEb	2DNR luminance noise reduction enable	Yes, adjustable by Gain/Layer

2D NR Basic Control Parameters

Parameter Name	Description	Suggested Modification	Special Notes
m_pMax_diff_thr	Maximum luminance reference threshold, effective when Dynamic enable is on	No
m_pSig_gain	Luminance reference threshold; smaller values cause more areas to be denoised	Yes	Adjustable by Gain/Layer
m_pAback_alpha	Luminance noise re-addition; larger values re-add more luminance noise	Yes	Adjustable by Gain/Layer
m_pCnrSig_y	CNR reference Y threshold; larger values cause more areas to be denoised	Yes	Adjustable by Gain/Layer
m_pCnrSig_uv	CNR reference UV threshold; larger values cause more areas to be denoised	Yes	Adjustable by Gain/Layer
m_pCnrAback_alpha	Chrominance noise re-addition; larger values re-add more chrominance noise	Yes	Adjustable by Gain/Layer
m_pCnrSig_uv_shift	CNR threshold left shift factor (1 means shift left by 1 bit, i.e., multiply by 2)	Yes	Adjustable by Gain/Layer
m_pGainIndex	Gain index, suggested to keep default value	No
m_bDynamic_en	Adaptive denoise threshold enable; when enabled, luminance denoise ratio is dynamically calculated based on image content	No
m_nGlobalYnrStrength	Global luminance noise reduction strength	Yes	Adjustable by Gain/Layer
m_nGlobalCnrStrength	Global chrominance noise reduction strength	Yes	Adjustable by Gain/Layer

---

Some parameters adjust according to Gain/Layer, for example, m_pCnrSig_y:

• Columns represent gain levels:
  • Column[0] means 1x gain
  • Column[11] means 2048x gain
• Rows represent different layers:
  • Row[0] means parameters for layer 0
  • Row[4] means parameters for layer 4

Higher layers correspond to processing higher frequency regions of the image.

2D NR Luminance Control Parameters

Parameter Name	Description	Suggested Modification	Special Notes
m_bLuma_en	YNR (Y noise reduction) enable according to luminance	User setting
m_pLuma_sig	YNR strength varies with luminance, levels 0-16 correspond to luminance 0-255, divided into 16	Yes	Adjustable by Gain/Luma
m_bCnrLuma_en	CNR (color noise reduction) strength control enable according to luminance	User setting
m_pCnrLuma_sig	CNR strength varies with luminance, levels 0-16 correspond to luminance 0-255, divided into 16	Yes	Adjustable by Gain/Luma

---

2D NR Radial Noise Reduction Strength Adjustment

Parameter Name	Description	Suggested Modification	Special Notes
m_bRadial_en	Enable noise reduction strength adjustment along the radius: 0: Disable 1: Enable	User setting
m_pRadial_ratio	Control of noise reduction strength increase along the radius; higher values mean stronger edge noise reduction	Yes	Adjustable by Gain/Layer
m_bCnrRadial_en	Enable chroma noise reduction strength adjustment along the radius: 0: Disable 1: Enable	User setting
m_pCnrRadial_ratio	Control of chroma noise reduction strength increase along the radius; higher values mean stronger edge chroma noise reduction	Yes	Adjustable by Gain/Layer

3D NR Motion Area Detection and Noise Reduction Strength Control

Parameter Name	Description	Suggested Modification	Special Notes
m_pTnrDAvg	Inter-frame luminance average difference threshold; smaller values make motion detection easier	Yes	Adjustable by Gain/Layer
m_pTnrWAvg	Weight of inter-frame luminance average difference; larger values increase reliance on luminance diff	Yes	Adjustable by Gain/Layer
m_pTnrQWeightAvg	Coefficient for inter-frame luminance average difference; larger values make motion detection easier	Yes	Adjustable by Gain/Layer
m_pTnrDSad	Threshold for sum of absolute inter-frame luminance differences; smaller values make motion detection easier	Yes	Adjustable by Gain/Layer
m_pTnrWSad	Weight for sum of absolute inter-frame luminance differences; larger values increase reliance on this metric	Yes	Adjustable by Gain/Layer
m_pTnrQWeightSad	Coefficient for sum of absolute inter-frame luminance differences; larger values make motion detection easier	Yes	Adjustable by Gain/Layer
m_pTnrDuv	Threshold for inter-frame UV average difference; smaller values make motion detection easier	Yes	Adjustable by Gain/Layer
m_pTnrWuv	Weight for inter-frame UV average difference; larger values increase reliance on UV difference	Yes	Adjustable by Gain/Layer
m_pTnrQWeightUV	Coefficient for inter-frame UV average difference; larger values make motion detection easier	Yes	Adjustable by Gain/Layer
m_nTnrLumaBase	Base luminance difference threshold; smaller values make motion detection easier	No
m_pTnrLuma	Threshold for luminance-based inter-frame average difference; smaller values make motion detection easier	Yes	Adjustable by Gain/Layer
m_pTnrYStrengthQ6	TNR global luminance noise reduction strength; larger values increase luminance noise reduction in static areas	Yes	Adjustable by Gain/Layer
m_pTnrUVStrengthQ6	TNR global chroma noise reduction strength; larger values increase chroma noise reduction in static areas	Yes	Adjustable by Gain/Layer
m_pStrongSig_gain	Reference threshold for luminance noise reduction in motion areas; smaller values increase noise reduction	Yes	Adjustable by Gain/Layer
m_pStrongAback_alpha	Luminance noise re-addition in motion areas; larger values increase re-added noise	Yes	Adjustable by Gain/Layer

3D NR Other Parameters

Parameter Name	Description	Suggested Modification	Special Notes
m_nTnrLumaWeight	Window size used for luminance calculation	No
m_pTnrBlockWeight0	Window size for luminance difference calculation in Layer 0	No
m_pTnrBlockWeight1	Window size for luminance difference calculation in Layer 1	No
m_pTnrBlockWeight234	Window size for luminance difference calculation in Layers 2, 3, and 4	No
m_pTnr2dMappingCurve	Fusion coefficient for noise reduction intensity between motion and still areas: 256 = full weak (still area) noise reduction, 0 = max strong (motion area) weight	No
m_p*Ratio	Reserved

Chromatic Aberration Suppression

The chromatic aberration suppression function is controlled jointly by overexposed points and the hue_pf interval. Color fringes that meet the overexposed points condition and fall within the hue_pf interval will be suppressed.

Parameter Name	Description	Suggested Modification	Special Notes
eb	Color detection switch; when enabled, denoising for skin and blue sky regions is enhanced	User setting
m_ppf_en	Purple fringe removal enable: 0: Disable 1: Enable	User setting
m_nwp_en	Overexposure area detection enable for purple fringe removal: 0: Disable 1: Enable	User setting
m_nwp_th	Threshold for overexposed pixel determination; higher value means fewer pixels considered overexposed (see Uv_wp_gain - num_wb diagram)	Yes
m_nnum_wp_min	Threshold to start overexposed area effect; purple fringe removal starts when overexposed points exceed this number (see Uv_wp_gain - num_wb diagram)	Yes
m_nnum_wp_step	Controls the transition interval (see Uv_wp_gain - num_wb diagram)	Yes
m_nhue_pf_min	Lower bound of the purple fringe hue interval (see Uv_pf_gain - hue_pf diagram)	Yes
m_nhue_pf_max	Upper bound of the purple fringe hue interval (see Uv_pf_gain - hue_pf diagram)	Yes
m_nhue_pf_tr_step	Transition control for the purple fringe hue interval (see Uv_pf_gain - hue_pf diagram)	Yes
m_nuv_wp_gain	Weight of overexposed points; higher value means overexposed points have greater impact on final saturation (see Uv_wp_gain - num_wb diagram)	Yes	Can vary with Gain
m_nuv_pf_gain	Weight of the purple fringe interval; higher value means greater impact on final saturation (see Uv_pf_gain - hue_pf diagram)	Yes	Can vary with Gain

---

Uv_pf_gain – hue_pf Diagram

Uv_wb_gain – num_wp Diagram

---

Raw HDR Parameter Description

The Raw_HDR_Soft module controls software HDR processing.

Raw HDR Parameters

Parameter Name	Description	Suggested Tuning	Special Notes
m_nHDRDownSampleEb	Downsampling enable; should be enabled when input image width or height exceeds 8225 pixels	No
m_nINTFlag	Error handling flag	No
m_pGTMCurveSet	GTM curve set; 5 sets corresponding to long and short exposure multiples: 1, 4, 16, 64, 256	No

File to File

File to File Function Description

he platform supports hardware simulation called File to File (f2f), which pushes a VRF file into the device, replacing the sensor image data, and obtains the ISP-processed image. This is suitable for single-frame simulation.

File to File Usage Instructions

1. Connect the device to the PC via USB.

2. Use ADB to push the VRF file to the specified path with the specified filename (different filenames correspond to different tuning parameters):

   • Main camera

     adb push xxx.vrf vendor/etc/camera/rear_primary_sensor.vrf

   • 副摄

     adb push xxx.vrf vendor/etc/camera/rear_secondary_sensor.vrf

   • 前摄

     db push xxx.vrf vendor/etc/camera/front_sensor.vrf

3. Start the camera to see the VRF preview processed by the ISP. You can also capture photos saved as JPG.

Quick Validation of Tuning Parameters

Quick Validation Function Description

1. The platform supports quick validation of tuning parameters. Enable the following property:

   adb shell setprop persist.vendor.camera.enable.settingfile 1

2. Save the tuning parameters from the Tuning Tool, then push them to the device at the specified path with the specified filename via ADB. Restart the camera for the changes to take effect:

   adb push xx.data /vendor/etc/camera/tuning_files/

Storage Path:

/vendor/etc/camera/tuning_files/

Common File Names:

FlickerSetting.data
front_cpp_nightshot_setting.data // Front Camera
front_cpp_preview_setting.data front_cpp_snapshot_setting.data front_cpp_video_setting.data front_isp_setting.data front_isp_setting_video.data front_nightshot_setting.data rawhdr_default_setting.data rear_primary_cpp_nightshot_setting.data // Main Camera
rear_primary_cpp_preview_setting.data rear_primary_cpp_snapshot_setting.data rear_primary_cpp_video_setting.data rear_primary_isp_setting.data rear_primary_isp_setting_video.data rear_primary_nightshot_setting.data
rear_secondary_cpp_nightshot_setting.data // Rear Secondary Camera
rear_secondary_cpp_preview_setting.data rear_secondary_cpp_snapshot_setting.data rear_secondary_cpp_video_setting.data rear_secondary_isp_setting.data rear_secondary_isp_setting_video.data rear_secondary_nightshot_setting.data

VRF viewer

VRF Viewer Interface

Click on a VRF file to launch the VRF viewer. The interface is shown below:

Menu: Functional Area

• Open: Open an image (supported formats: *.vrf, *.nv12, *.nv21, *.yuv, *.bmp, *.jpg, *.jpeg, *.png)

• Save: Save the image (supported formats: *.bmp, *.jpg, *.jpeg, *.png)

• Options: Image processing options (click Apply to apply options, click Save to save current settings)

• Options for VRF images:

  • Demosaic: Apply demosaicing
  • Linearization: Apply linearization processing
  • WB: Apply white balance gains from VRF
  • Dgain: Apply digital gain from VRF
  • Gamma: Apply sRGB Gamma curve
  • Split Channel: Display the four Bayer channels separately

• Options for YUV images:

  • Color Standard: Choose BT601 or BT709
  • Swing: Set YUV data range (Full range or Standard range)
  • YUV: NV12 / NV21 conversion
  • Gray Image: Display grayscale image

Other Function Buttons

• ZoomIn: Zoom in the image
• Normal: Display the image at original size
• ZoomOut: Zoom out the image
• Move: Drag the image to view different areas
• Select: Select a region in the image by drawing a rectangle
• Histogram: View the brightness and color histograms of the image
• Info: Display metadata of the VRF image (e.g., format, resolution, gain)
• Window: Switch between opened image windows
• Previous: Show the previous image
• Next: Show the next image