Joined: 2/26/2011
Posts: 21
What's 10bit444? And what's the different to the "normal" movies?
Editor, Emulator Coder, Site Developer
Joined: 5/11/2011
Posts: 1108
Location: Murka
10 bit 444 encodes are encoded using x264 in 10 bit mode with 4:4:4 colorspace. colors are crisper and more detailed, without much of a change in filesize. old crap can't play such files, which is why the "regular" encode is still offered
Joined: 2/26/2011
Posts: 21
natt wrote:
10 bit 444 encodes are encoded using x264 in 10 bit mode with 4:4:4 colorspace. colors are crisper and more detailed, without much of a change in filesize. old crap can't play such files, which is why the "regular" encode is still offered
that means 10bit444 movies have a better quality, right?
Player (66)
Joined: 4/21/2011
Posts: 232
Joined: 2/26/2011
Posts: 21
ok, i see, thank you! :)
Banned User
Joined: 3/10/2004
Posts: 7698
Location: Finland
nanogyth wrote:
Could someone explain in more detail what's happening here? It looks to me like the red channel has been brightened on the version at the right, but otherwise it's hard to tell which one should be "more accurate" (without seeing the original lossless frame). That might be a good idea, actually: Add the original lossless version of the frame besides those, for comparison.
Lex
Joined: 6/25/2007
Posts: 732
Location: Vancouver, British Columbia, Canada
The 4:2:0 frame has chroma subsampling, so the chroma channel is half the resolution of the luma channel, so it needs to be upscaled to 4 times its original size (2× width and 2× height). With small numbers of pixels (low-resolution images), this leads to diffusion of color in pixel art, and it's especially visible in reds. The luma channel (basically a greyscale image of the original frame) is full-resolution, so that's why you can see precise detail in the shapes. All in all, the 4:4:4 image is closer to the original lossless frame since each pixel has its own color. Also, I'm assuming the 4:4:4 image was encoded with 10-bit precision, so conversion back to 8-bit RGB is much more likely to be correct, assuming the same color matrix was used for decoding as encoding.
creaothceann
He/Him
Editor
Joined: 4/7/2005
Posts: 1874
Location: Germany
APNG version (view in Firefox / Chrome with plugin / Opera). The minimap shows the "JPEG-like" compression artifacts on solid blocks.
Banned User
Joined: 3/10/2004
Posts: 7698
Location: Finland
It would still be nice to see a comparison to the original, non-lossy frame.