PDA

View Full Version : Wrong iDCT Algorithm in DVD-Rb 1.06


Fishman0919
17th January 2006, 05:46
I noticed this pb with DVD2SVCD Version 1.2.3 Build 1 when it changed to the new DGDecode and now DVD-RB does it too. With the old DGDecode 1.10 IEEE-1180 Reference was 4, now in DGDecode 1.45 it's 5... so with you pick IEEE-1180 Reference with DVD-RB 1.06 you don't get IEEE Ref...you get 64-bit Floating Point which is 4 in DGDecode 1.45.

Carpo
17th January 2006, 09:58
little swine that jdobbs is - he always releases a new version when i start a batch encode off ;)

so if we want ieee-ref we have to select 5 instead of 4 that right fishman - bit early here - just wake up and stilll asleep ;P

jdobbs
17th January 2006, 13:16
Never expected that... are the other settings correct?

Fishman0919
17th January 2006, 17:47
Never expected that... are the other settings correct?

As for idct... as far as I can tell... just the remapping of the algo's

Carpo
17th January 2006, 17:51
is there a poss fix for this little issue ? have 1 or two dvds to backup and dont really want to waste time if the idcts are wrong

Fishman0919
17th January 2006, 18:01
is there a poss fix for this little issue ? have 1 or two dvds to backup and dont really want to waste time if the idcts are wrong

Yes , pick 32 bit sse2/mmx and you will get IEEE-ref

Here....

DGDecode 1.10 DGDecode 1.45
------------- -------------
1. 32 Bit MMX 32 Bit MMX
2. 32 Bit SSE/MMX 32 Bit SSE/MMX
3. 64 Bit Floating Point 32 Bit SSE2/MMX
4. 64 Bit IEEE-1180 Ref 64 Bit Floating Point
5. 32 Bit SSE2/MMX 64 Bit IEEE-1180 Ref (Fixed)
6. 32 Bit SSE/MMX (Skal) 32 Bit SSE/MMX (Skal)
7. 32 Bit Simple MMX (XviD) 32 Bit Simple MMX (XviD)

jdobbs
17th January 2006, 18:21
is there a poss fix for this little issue ? have 1 or two dvds to backup and dont really want to waste time if the idcts are wrong???? As long as it works, why would it matter. The only noticable difference is speed.

Fishman0919
17th January 2006, 18:42
http://forum.doom9.org/showthread.php?p=682312#post682312

A test I did some time back...

if anyone can see a diff from 99.93% to 99.63%...please let me know.

samuelal
17th January 2006, 20:45
???? As long as it works, why would it matter. The only noticable difference is speed.

Dear jdobbs

Would it be possible to ask, that the next release of DVD-RB, the issue will be fixed in such a way that the application's engine will 'know' and change the iDCT's GUI correctly in accordance to the detected DGDECODE version?

Thank you very much for considering it.

P.S. also thanks to Fishman0919 for the research on the table+speed difference between the DGDECODE versions!


Samuel.

jdobbs
17th January 2006, 21:33
Certainly. I would have done it this time, but I hadn't noticed the change.

Carpo
17th January 2006, 23:37
???? As long as it works, why would it matter. The only noticable difference is speed.
to avoid confusion i would have to say

msobetzko
18th January 2006, 00:03
OK, I'm confused now. I'm sorry but I don't understand all the different IDCT Algorithms and such.

I just installed the full 1.06 Pro release. Is there something wrong with the installation? Should I NOT use the current version.


Mike

Carpo
18th January 2006, 00:19
if you dont use them or use dvd-rb defaults in this area there is no problem and its safe to use :)

blueboyec
18th January 2006, 00:43
I presently have my idct option set to 32 Bit Simple MMX (XviD). So if I now use DGDecode 1.45, I should change it to 32 bit sse2/mmx?

Yes , pick 32 bit sse2/mmx and you will get IEEE-ref

Here....

DGDecode 1.10 DGDecode 1.45
------------- -------------
1. 32 Bit MMX 32 Bit MMX
2. 32 Bit SSE/MMX 32 Bit SSE/MMX
3. 64 Bit Floating Point 32 Bit SSE2/MMX
4. 64 Bit IEEE-1180 Ref 64 Bit Floating Point
5. 32 Bit SSE2/MMX 64 Bit IEEE-1180 Ref (Fixed)
6. 32 Bit SSE/MMX (Skal) 32 Bit SSE/MMX (Skal)
7. 32 Bit Simple MMX (XviD) 32 Bit Simple MMX (XviD)

SpazzHH
18th January 2006, 00:49
I presently have my idct option set to 32 Bit Simple MMX (XviD). So if I now use DGDecode 1.45, I should change it to 32 bit sse2/mmx?

No, that is IDCT 7 in both versions, so it won't matter. Only 3,4, and 5, changed places.

jdobbs
18th January 2006, 00:50
No. It will work correctly and you will use the correct iDCT algorithm. The ones affected are also the ones rarely used because they are so slow.

I'm still trying to figure out why they were changed at all...

This is truly much ado about nothing.

apfraats
18th January 2006, 01:38
Come on JDOBBS, youre a software developper for heavens sake ;)

I use 64 bit floating point.....

Why, even I don't know. Suppose I decided after a few beers it should be more precise.

And always nice to use youre floating point engine in youre CPU also.....

A long time ago, we had to buy a second chip for that purposes :D

In my relatively low-level software development period , I was even annoyed by the most little mistake I made..... There was no room for pragmatic explanations...

Well tha'ts because maybe making a DVD-backups is something else as operating automated weapon systems..... :D :D :D

TheSeeker
18th January 2006, 01:59
Is the version of dgdecode that comes packaged with dvdrb 1.06 version 1.45 or the older version?

samuelal
18th January 2006, 02:23
Is the version of dgdecode that comes packaged with dvdrb 1.06 version 1.45 or the older version?

Hello man.

To answer your question briefly: Yes, it installs the DGDecode v1.4.5 .

To answer it while qouting the author himself:
Added support for DGDECODE v1.4.5. This update
includes many enhancements/bug fixes and makes
DVD-RB's support consistent with Neuron2's most
current release. Support for MPEG2DEC3DG and
DGDECODE v1.1.0 remains. This update includes
a fix which corrects a rare problem in which an
encoder may "freeze" while trying to encode a
segment. The DVD-RB Installer now includes and
installs DGDECODE v1.4.5 by default.
This was written in the Email that every DVD-RB Pro user received (and probably update in the history/readme file as well).

Samuel.

onesoul
18th January 2006, 05:06
I don't understand all the impatience, jdobbs denoted awareness about this little "issue" since 3rd post so it means sooner or later it will be fixed. In my opinion there is no need to nag so much...

be cool