Hi TI experts
During ISP debugging, water ripples were found in the YUV image.
Raw image without water ripples.
This thread has been locked.
If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.
Hi TI experts
During ISP debugging, water ripples were found in the YUV image.
Raw image without water ripples.
Hi Peng,
I am not familiar with the sensor registers.
Could you please share the knee points you used as tuning tool input?
xml files when this issue happens?
Could you also please share the GLBCE, CFAI and CCM settings?
I would need them to reproduce the issue.
Hi peng,
Sorry about the delay.
I am not able to completely reproduce the issue using your xml files in 0601.case.zip without your GLBCE settings.
This is what I get after adjusting image brightness with asym=10 for GLBCE settings and additional digital gain.
/cfs-file/__key/communityserver-discussions-components-files/791/o_5F00_1920x1080_5F00_nv12.yuv
It looks like a problem concerning tone mapping (GLBCE), noise filtering, and sharpening.
I checked your decompanding settings and they look OK and should not cause any contours.
Noise filtering gives smooth output after removing noise and then tone mapping and conversion to yuv may cause quantization error that looks like contours on smooth and neutral image areas.
You may adjust your tone mapping and noise filtering (maybe slightly weaker) a little more to reduce the contours.
Hi Gang
My image was saved after passing GLBCE. So water ripples are not caused by glbce.
So can I assume that this phenomenon is quantization error caused by image compression.
Hi peng,
I suppose you mean GLBCE is by passed.
GLBCE alone won't cause the contours.
It is a combination of noise filtering, tone mapping (global and/or local) and quantization in conversion to YUV.
This kind of gray field gives a smooth gradient output image after noise filtering and it is contrast enhanced by global/local tone mapping.
Then, small quantization error becomes visible because of the nature of the gray field.