Psychovisuals and water -> improvement ??

[jggimi]

I've noticed, that with light psychovisuals, the macroblocks we've come to note with water scenes are not as noticeable.

Here are two frames (in one 70K jpg) from two water scenes near the end of City of Angels. The attachment will appear once Acaila has a chance to approve it.

Has anyone else noticed this phenomenon?

[Acaila]

There ya go :)

And btw, I still see quite a few blocks on those pics. The lower pic definately has blocks on the water part, however on the upper one I can't see much blocky water, but a lot of blocky sky though. :(

Anyway, I don't use psyco-stuff much myself, so I'm not the expert.

[Mac Sidewinder]

I use the light mode and have noticed that the real dark areas seem to have less blocks than they used to. I've also noticed and improvement on my home videos that have water in them.

Mac

[jggimi]

Acaila. Yeah, the blocks in the sky show on the still -- but not in the moving image -- I don't know how to explain it, but the thing I noticed most about these scenes is the huge reduction in noticable in-water blocks. The lower image, to my eye, has far fewer than any previous encoding I've done, and you don't notice blocks on the surface during playback at all.

(Perhaps I need to visit my optometrist. At the moment I use either contacts with reading glasses -- or bifocals. :cool: )

[theReal]

Have you got the same movie without psychovisuals? This would make a real comparison. Otherwise we can't really compare anything.
btw. what bitrate have you used for what resolution?

[jggimi]

You're right. I should re-encode without them to compare. I won't be able to until this weekend at the earliest, being out of town.

I encoded at 640x272, 23.976fps... I calculate a 659Kbps bitrate as follows:

According to Vdub file information: 16,927KB in key frames + 547,650K in delta frames = 564,577KB in total video. 01:54:14 = 6,854 secs. 564,577 / 6,854 = 82.3719KB/sec x 8 =

[jggimi]

I've re-run the same encoding without psychovisuals. It's hard, in these still jpg images, to see the differences.

The scenese with surf in them ... to my eyes, the water looks about the same. But the sky looks blockier without light psychovisuals. And underwater scenes or water surface scenes also look blockier without the feature turned on.

I've attached another jpg of the same frames, this time without psychovisuals.

[kappa]

My impression is that real problems arise in underwater scenes, not water surface scenes.
I too use light PE for all my rips (even with Divx 5.0 - I had found that using full processing mode (RGB output) the known artifacts disappeared). I also use ligh pre-processing, it has absolutely no visible difference and it improves compresability.