Hi,
I wonder if AI could be leveraged to expand color and contrast detail in a video by being content aware?
So one example would be that the model might recognise the sky in a scene and brighten up the sun and clouds to HDR levels while perhaps keeping and even creating details in the clouds as well as preserving the blue in the actual sky.
Same thing in dark objects, For example a man wearing a black t-shirt on a beach.
The AI would then darken the black T-shirt while creating/improving shadow details.
Also object colours could be improved with less banding by recognising objects and improving color gradients. Example would be a zoomed in picture of an apple that has a color gradient from green to red, the AI model would recognise that this is an apple and make sure that the color gradient is smooth, realising and has no banding.
Would it be possible?
