High quality on 3D renders?

[CheloXL]

Hi!,
I´m trying to compress some 3D renders and i found almost impossible to get a nice compression. No matter how high the bitrate i set, i always get an ugly squarified picture (like a bad quality jpg).

My settings:
Profile: unrestricted, h263, adaptive quant, quarter pixel, GMC, no packed frames.
Target br: 3072 (i´ve tried lower values first, of course, because normally with 1500-1700 i always get nice results).
Quality: MSP: 6, VHQ: 4, Use VHQ for bframes, use chroma motion, trellis.

Suggestions?

[Sharktooth]

remove adaptive quant. and use EQM V2 custom matrix.

[CheloXL]

Still the same problem... :(

[Sharktooth]

can you post some screenshots or a small clip of the xvid encode?

[Sagittaire]

Motion level affect average bitrate : 3000 kbps can be very bad for some source and 300 kbps can be very good for other.

Make your encoding in quality mode: q2 or q3 H263 done always good visual quality without blocking.

[CheloXL]

Here you have:
5527
This is how the source file looks

5528
This is how the encoded file looks. Using Sharktooth advice and at 3072kbps.

@Sagittaire: I will try your suggestion and post the results.

[xyloy]

Hi!,
I´m trying to compress some 3D renders and i found almost impossible to get a nice compression. No matter how high the bitrate i set, i always get an ugly squarified picture (like a bad quality jpg).

My settings:
Profile: unrestricted, h263, adaptive quant, quarter pixel, GMC, no packed frames.
Target br: 3072 (i´ve tried lower values first, of course, because normally with 1500-1700 i always get nice results).
Quality: MSP: 6, VHQ: 4, Use VHQ for bframes, use chroma motion, trellis.

Suggestions?
Have you tried without Quarter Pixel and adaptive quant. ?
(as for chroma optimizer, I was about to suggest you to use it too, but I don't think it would help. That doesn't mean you can't test it and see by yourself if it improves quality or not ;) )

AFAIK, isn't the MPEG quant. matrix more adequate for high bitrates such as 3 Mbps(or near 0.3 BPF encodes) than H.263 ?:confused:
(ie. Doom9 XviD guide were H.263 is described to be good for 900 Kbps and lower, and MPEG good for 1000 kbps and higher)

[foxyshadis]

Renders typically have lower compressibility than movies, though I can't tell if it's saturating without seeing the output. They might benefit from a totally specialized matrix, I don't know.

[Sharktooth]

During the tests for choosing the "sharp matrix" for autogk, eqm v2 was particularly appreciated for CG.
That's why i suggested it.

[xyloy]

Ok for CG. Let's say I said nothing. ;-)

Is EQM V2 the same as CG-animation matrix given with XviD v1.1.0 final, or is it better ?

[Sharktooth]

Ok for CG. Let's say I said nothing. ;-)

Is EQM V2 the same as CG-animation matrix given with XviD v1.1.0 final, or is it better ?
no, it's different.
sincerely i dont know if it's better bug it gave very good results.