Originally Posted by Kid Banana
The differences on mouseover 13 are quite pronounced. The actor's left eyebrow and right upper lip are far more detailed in the UK edition.
I think the differences must be to do with the VC-1 implementation that Warner is using. Shame RDjam no longer posts here much as he was very much the go-to man for VC-1 info.
There is only one main implementation of VC-1 codec used by professional studios. There is nothing wrong with it and also with US Watchmen version. It's quite low bitrate encode and that's all. UK version uses highier bitrate (for me on the proper level) and difference is visible- mutch better high frequencies retention (as expected). There are some interesting things about these encodes:
- encoders spread data slightly different- AVC puts much more to I frames when VC-1 to B frames (encoding settings)- this makes VC-1 encode look "better" in static frames, but in motion good quality I frames are more important, as long as there is no visible "pumping effect" (to bad quality B frames in compare to I frames),
- AVC encode has quite small B frames, but even so they are still better quality then VC-1 ones (very visible on the frist frame),
- frames 19 (I frame for VC-1 and B frame for AVC) shows that you should try to compare the same type of frames as I frames are always better quality than B ones- US looks not lot, but still sharper,
- frame 13 shows big difference, becuase both encodes are B frame+AVC is bigger,
- the biggest difference is probably visible on the 3rd frame, becuase AVC is I one and VC-1 is B,
- frame 12 is interesting- even if AVC one is much smaller still VC-1 looks way less detailed- does it prove that AVC is more efficient???
There is nothing wrong with US version- just to low average bitrate, to achieve good transparency. It would be nice to have source frames- than people would see how much is removed during encode
The difference between source and AVC would be probably about the same as between these 2 versions.