Quote:
Originally Posted by
Mike_WI 
Just curious.
Would that show up in pictures, or too subtle to photograph?
I would never have though it would be possible... but here you are !
I took the pictures in raw format, 100 iso, then set the contrast boost on maximum in the raw -> tiff converter.
The pictures are out-of-focus on purpose, to avoid moire effect and confusing the steps with the panel grid.
Even like this, the effect is difficult to see. It gives a good idea of what it looks like in reality.
I've also checked with Gimp, creating a 8-bit greyscale ramp and increasing its dynamic. I got exactly the same kind of steps, so that's not a faulty contrast process in the Oppo.
In case your PC display posterizes the image so much that the steps are not visible, I took the above picture and boosted the contrast again to help you localize the steps. Now, the result looks completely unnatural.
No correction in the Oppo : no step
Very light correction. Two steps are barely visible in the lower ramp, the right hand one just near the arrow above the "contrast" word, a bit on the right side.
Light correction. Three steps are visible in the upper ramp, in the grey part.
Bigger correction. 6 steps are visible in both ramps. Sharper in the upper one.
Quote:
Originally Posted by
Bob Pariseau
Now your YCbCr 4:2:2 input with Deep Color off should be 12 bits per component (24 bits per pixel -- with only 2 of the 3 components transmitted each pixel time). So if your display is truncating input back to 8 bits per component you would expect to be able to see it there too.
This is the case indeed. But who knows if the Oppo turns on deep color sending the corrected picture to the 4:2:2 output, while the deep color option is actually off ?
Quote:
Originally Posted by
Bob Pariseau
Keep in mind that the video processing path is different in the Oppo for Source Direct than for any explicit output resolution. The type of checking you are doing should be done with an explicit output resolution so that the VRS chip is doing all the heavy lifting.
I quite don't understand what you mean here.
I can say that this test is done with the output resolution set to 1080p.
I've just tried in source direct. the result is the same. Steps with deep color off, no steps with deep color off (dithered).
Quote:
Originally Posted by
wmcclain
Thank you! This is the first report I have ever heard of deep color processing being visible on a test pattern.
Not quite. It's the effect of truncation versus dithering that is shown here, not deep color. The data are all 8 bits.
The advantage of deep color may show up if the display device itself performs heavy processing after the player. For example if black and white levels are adjusted in the player, and the display is in some kind of "high contrast" picture mode. The dither
may become visible without deep color... if you're close enough to see the display's grid, I'd say.
