Divx 5 final best of the best

[BiaTch 5.0]

I'm a Xvid encoder because I'm aim for 1CD encodes a find it to do a good job however Divx5 imo will be the highest quality mpeg4 encoder, (strong)Psycovisual enhancementshave imo amazing potential I've found them to give me about maybe 5% less quality than DVD the only catch is that backgrounds become shakey, like there is a ghost in the background. When/If this is resolved Divx will be a very hard to beat high quality mpeg4 encoder.

[DJ Bobo]

psychostuff is a non interesting feature anyway, they fix it or not, this won't do anything, as I won't use it, this is all marketing, what do you think MPEG compression is based on? it's already based on the weaknesses of our eyes, since it removes all details we don't see or hardly see.

The only usefull feature of DivX5 is bidirectional encoding, all other options are for the dog!

[yingx2]

The only usefull feature of DivX5 is bidirectional encoding, all other options are for the dog!

I agree. But I would have all of the MP4 features on using 0% QB for credits encoding jobs. They're there to help reduce file sizes.:)

[BiaTch 5.0]

But then why is divx5 so high quality with strong Psycovisual enhancements?

[DJ Bobo]

I tried strong psychocrap on some DBZ episodes, it makes picture only worse, I wasn't satisfied at all.

[Got Milk?]

Originally posted by DJ Bobo
I tried strong psychocrap on some DBZ episodes, it makes picture only worse, I wasn't satisfied at all. I'm assuming the psychovisual stuff is optimised/designed for film, and thus you won't get the best results when using it on animation?

[DJ Bobo]

@ Got Milk
I don't know, I never tried it on real movies.

[BiaTch 5.0]

The point is that psychovisual works, you said it’s crap, it is all marketing yet you only tried it once, on animation. Before you acquit Psycovisual enhancements why don’t you do some testing?

BTW I found it’s useful with 2CD encodes otherwise it seems to get overwhelmed & cant do a good job of hiding artifacts


If Iam wrong someone explain to me why Psycovisual enhancements produce such good results.

[yingx2]

I've tried this psycho-enhancement hype several times. I thought that it was a pretty good feature at first because I was using 100% QB for my tests.

Then I tried 600kbps and tested it again. The one without PE turned out to be fine, while the other one with PE got slightly (but percievable) degraded picture at the same size.

If your two cd rip looks good with PE, I'd assume it will look even better without PE.

The point is that two cd rip (at about 1500kbps) almost guarantees good result, with or without PE feature.

just my two cents.:)

[DJ Bobo]

@ yingx2
Thanks for your support

@ BiaTch 5.0
Yeah yeah, and you are the great tester... -.-
If it doesn't work well with anime, why should it work better with real movies?!
This is all because Got Milk is "assuming", as if Got Milk was the programmer of DivX5 or something, or is he a prophet or something?
And because of his "assuming", I'm automatically wrong right?
Man man, and you wonder why I get nervous sometimes -.-

[Got Milk?]

Originally posted by DJ Bobo
@ BiaTch 5.0
Yeah yeah, and you are the great tester... -.-
If it doesn't work well with anime, why should it work better with real movies?!
This is all because Got Milk is "assuming", as if Got Milk was the programmer of DivX5 or something, or is he a prophet or something?
And because of his "assuming", I'm automatically wrong right?
Man man, and you wonder why I get nervous sometimes -.- Does that mean I'm not a prophet? Damn, wish people would tell me.

I wasn't saying you were wrong, just suggesting there would be a greater benefit when used with normal film. I don't know, I haven't done tests with animation, like you haven't done tests with movies. It was just a suggestion, wasn't designed to be a personal jibe at you or anything - not sure how you interpreted it as that.

[TheXung]

The only usefull feature of DivX5 is bidirectional encoding, all other options are for the dog!

bframes actually does close to nothing for quality, at least in 1 CD films. Just make a 1CD rip with and without it on, the quality is pretty darn close. Maybe at lower bitrates . . . but that's just too low for me

The feature that will give you real quality gains is qpel. At least in film . . . .

what do you think MPEG compression is based on? it's already based on the weaknesses of our eyes, since it removes all details we don't see or hardly see

Actually, the only weaknesses of our eyes built into MPEG spec/design is use of the DCT/iDCT and 4:2:0 colorspace and an optimized quantum matrix. All of these concepts are basically ripped from JPEG, thus it's only exploiting spatial weaknesses of the eye. The key feature that separates MPEG from MJPEG is motion estimation/compensation which is really just a more accurate model of how to represent motion with. MPEG leaves room for other features, like bframes, adaptive quant, etc but these features are pretty much implementation dependant (there is no garantee that using bframes, qpel, or anything will make a better video codec). If you think about it, Nandub is really psychovisual enhancements that are "implementation specific" instead of "mpeg design specific"


The thing about Divx 5 PE, video streams shouldn't become garbled like that when making a 1CD rip. For this reason, I think it's not producing an mpeg-4 compliant bitstream. The results aren't that impressive for me to go through the whole movie and check for garbled blocks.

[BiaTch 5.0]

Originally posted by DJ Bobo
@ yingx2
Thanks for your support

@ BiaTch 5.0
Yeah yeah, and you are the great tester... -.-
If it doesn't work well with anime, why should it work better with real movies?!
This is all because Got Milk is "assuming", as if Got Milk was the programmer of DivX5 or something, or is he a prophet or something?
And because of his "assuming", I'm automatically wrong right?
Man man, and you wonder why I get nervous sometimes -.-

YOU’RE the one who is assuming, all I said was you make claims without having fully tried PE more than once & you have to admit that compressing a cartoon & compressing a real movie is quite different. To be truthful I haven’t tried Divx5 2CD without PE I just compared with Xvid 2CD. I didn’t say I was 100% sure I was right I just wanted an explanation, It was a pointless thread anyway.

BTW what’s this about great tester? , do you have eyes? , can you encode & compare? , Then I guess you’re a great tester too.

[theReal]

First I thought the psychovisuals don't do anything, then I found that "strong" made movies look worse, but recently I found out you should always try stuff on many different movies before judging...

For some movies with a low bitrate, psychovisual settings can really improve edges, especially around people's faces - it makes cheeks look less "artificial", if you know what I mean. This is only the case if the bitrate is rather low, and it depends on the movie.

Then, I also found that even with very high quality movies (compress test ~70%), light psyvi. can improve some single scenes, where it makes some light but typical "Divx4 artefacts" look more natural, more like Divx3.11 SBC artefacts. On most parts of that movie, you don't see any differences at all, it's only a few scenes where it slightly improves the looks.

Alltogether I'd say the psy settings sharpen the edges a bit and make Divx5 artefacts look more "grainy" rather than "blocky". That can be good or bad, you have to test it for every single movie.