Philips 630/02 (ESS Vibrato chip) and Custom Matrices incompatibility

[dr.Prozac]

I posted this thread in XviD forum, but Alxemi said it would be better to place it here.
All encoded movies are displayed very well on my PC, but some of them have a very bad quality during watching in my home DVD player (the picture is full of something like macroblocks). I've noticed that it depends on a kind of matrix I have used to encode.
To be sure taht it really depends on matrices I have set all codec's settings to default.

This problem occured when such matrix had beed used: HVS Best, HVS Good and Jawor's Q.M.

There was no problems after compressing with: SixOfNine HVS, MPEG, H.263, Bullet Proof Heavy Comp.

I can add it doesn't depend on a level of bitrate. I made some tests with low and also with high bitrate.
I used Koepi's XviD 1.0.1 / 05062004
and also made some tests using
Jawor's XviD binaries / 27072004

[SeeMoreDigital]

Thanks for this information mateo4x4,

Did you use B-VOP with any of your matrices. And if so, did you select packed bit-stream?

Have you tried muxing any of your Mpeg4/mp3 streams into MP4?


Cheers

[dr.Prozac]

Did you use B-VOP with any of your matrices. And if so, did you select packed bit-stream?
I tried to encode the same movie using BVOP's 2/1.5/1 and also using BVOP's OFF. There was no any difference in playback.
It also didn't depend on PB. I made the same tests using PB ON and also PB OFF. No difference.
Have you tried muxing any of your Mpeg4/mp3 streams into MP4?
Yes, I've also tried doing this. The same results.

[SeeMoreDigital]

Well that's me out of suggestions.

That said, glfinish came up with an interesting 'AQ' obsevation a couple of days ago: -

http://forum.doom9.org/showthread.php?s=&postid=537962#post537962

Maybe this might help. It has 'very slightly' with my Xcard. But with different settings to glfinish's.


Cheers

[dr.Prozac]

That said, glfinish came up with an interesting 'AQ' obsevation a couple of days ago
Thank You SeeMoreDigital.
I've just done some more tests. Always the same effects.

I have also checked on more matrices. The results (23 matrices):

1. H.263 good playback {Preview} (http://www.dreamland.neostrada.pl/H.263.jpg)

2. MPEG good playback

3. HVS Best bad playback {Preview} (http://www.dreamland.neostrada.pl/HVS%20Best.jpg)

4. HVS Better bad picture {Preview} (http://www.dreamland.neostrada.pl/HVS%20Better.jpg)

5. HVS Good like in test nr 3

6. SixOfNine HVS (Max=24) good playback

7. SixOfNine Max=20 good playback

8. BulletProof's Heavy Compression good playback

9. BulletProof's High Quality good playback

10. Low Bitrate Matrix like in test nr 4

11. Very Low Bitrate good playback

12. Ultra Low Bitrate good playback

13. Selur's Matrix like in tests nr 4 & 10

14. Jawor's 1CD Matrix like in test nr 4

15. 28DaysLater good playback

16. Angel BestLow good playback

17. Angel BestVeryLow good playback

18. Andreas Einfache 99er like in tests nr 4 & 10

19. Andreas Doppelte 99er good playback

20. EQM_v1 like in tests nr 3 & 4

21. EQM_v2 like in tests nr 3 & 4

22. Professional like in tests nr 3 & 4

23. Ultimate good playback

SixOfNine is a very nice matrix for higher bitrate movies so I'm glad it works fine on my DVD player.... but
now it seems I'm out off my best matrices for low bitrate movies (Jawor's 1CD Q.M. and HVS Best):mad:

[MrBunny]

I'm not sure if this will work with the DVP-630, but it does work for the 642.

For just about any matrix that I've thrown at it for which you get a result like numbers 3 and 4, go into the system menu, then immediately exit it. That seems to clear it up.

However, for some matrices, it still causes blocking to occur on keyframes. From what I recall, the Fox Entertainment matrix (not included in your list) and the professional matrix will do that. This hack DOES seem to fix the hvs matrices, jawor's and Andreas 78er matrices as well as some others. I am still trying to work out how to determine the exact characteristics of these matrices which will give the different types of playback. Some, but not too much progress so far.

Definately, it involves both inter and intra matrices. So far it seems like a lower threshold for the allowable coefficients. To make a matrix that will work after the hack, I'm looking at a minimum of 12 in the intramatrix (except for the first 8) and 11 for the intermatrix (at least for the 1x2 and 2x1 coefficients, I don't play much with the DC coefficient due to the (old?) ffdshow playback issues). I've actually modified the Professional matrix based on the above guidelines and now it works fine after the hack.

Anyone have any other observations or confirmations so far? From what I'm hearing these Philips players are getting around, so it'd be nice to fully document their abilities.

[dr.Prozac]

First, I want to thank you for your post MrBunny.
For just about any matrix that I've thrown at it for which you get a result like numbers 3 and 4, go into the system menu, then immediately exit it. That seems to clear it up.
I'm afraid I don't understand how this hack works or seems it doesn't work at all according to me.
Could you write something more about it.
Thanks in advance

[MrBunny]

The hacks is really just as follows:
Just press the buttom that will bring you into the the system menu (on the DVP-642 it's just a button on the remote below the display buttom on the top right corner of the directional arrow buttoms). That menu is the one that allows you to change the audio, video, and other player settings. All you have to do is to enter that menu, then exit it again while playing back the file.

Hope that's clearer.

[dr.Prozac]

:thanks: MrBunny
Yes, it's much clearer now ! And I thought you had written about computer's System Menu before (in your previous post) :D So I couldn't believe that it was the idea of this hack :) Now I understand and I have already done some tests. It really works on DVP 630 ! But I am still wondering what is the way it works :confused: As if by magic :)

[SeeMoreDigital]

Jeez,

I think I might buy one of these players myself now!

By-the-way, have any of you guys got a widescreen TV?


Cheers

[dr.Prozac]

Jeez, I think I might buy one of these players myself now! :D


I've got Panasonic TX-32PS11P TV

[MrBunny]

Originally posted by SeeMoreDigital

By-the-way, have any of you guys got a widescreen TV?

Nope. Boring 4:3 for me.

[SeeMoreDigital]

Originally posted by mateo4x4
I've got Panasonic TX-32PS11P TV Great!

Have you tried decoding and playing anamorphic 720x576 encodes in either AVI or MP4?


Cheers

[dr.Prozac]

Have you tried decoding and playing anamorphic 720x576 encodes in either AVI or MP4?
Yes.I do very often.
The best is when aspect ratio is not higher than 1.95 Then I can greatly use TV Zoom and there is no problem with cutting picture to much or positioning the subtitles.

[SeeMoreDigital]

I have some proper Mpeg4 encodes complete with PAR signalling I would like you to try. If you are up for it!

The AVI sample might not auto AR correctly. But the MP4 sample might.

I'll provide a link tomorrow if thats ok with you?


Cheers

[dr.Prozac]

I'll provide a link tomorrow if thats ok with you?
No problem !
I will look at and get them tomorrow after work, ok ?
Today I'm really tired. going to sleep now :)

[minolta]

@SeeMoreDigital: for anamorphic, see my response in SpItoS' thread.

[SeeMoreDigital]

Originally posted by minolta
@SeeMoreDigital: for anamorphic, see my response in SpItoS' thread. Thanks for the heads up!

Were these encodes AVI's or MP4's?


Cheers

[minolta]

I had the 'custom PAR' problem with both AVI and MP4 containers (took me a while to figure out the solution, so I tried just about everything). The 'logic resize' behavior is the same for both as well.
-Minolta