Log in

View Full Version : Question about iDCT entry


SkVid
31st May 2005, 08:08
I have chosen "64 bit floating point" for "set decoder iDCT" which puts the entry

iDCT=3

into the rebuilder.ini. So far so good.

But the results in the .d2v file shows:

DGIndexProjectFile06
1
24 I:\VIDEO_TS\VTS_01_1.VOB

Stream_Type=1
iDCT_Algorithm=2
YUVRGB_Scale=1 (0:TVScale 1:PCScale)
Luminance_Filter=0,0 (Gamma, Offset)
Clipping=0,0,0,0 (ClipLeft, ClipRight, ClipTop, ClipBottom)
Aspect_Ratio=4:3
Picture_Size=720x576
Field_Operation=0 (0:None 1:ForcedFILM 2:RawFrames)
Frame_Rate=25000
Location=0,0,0,6CEB9

7 0 2048 1 1 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
7 0 51200 1 1 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
7 0 100352 1 1 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
7 0 145408 1 1 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
7 0 579584 1 1 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2


So here the iDCT_Algorithm=2 is set to two. Why this???
Reading the dgdecode helpfile it should be three.


Doing the same .vob with dgindex and the same settings for iDCT the resulting .d2v file looks like this:

DGIndexProjectFile06
1
24 I:\VIDEO_TS\VTS_01_1.VOB

Stream_Type=1
iDCT_Algorithm=3 (1:MMX 2:SSEMMX 3:FPU 4:REF 5:SSE2MMX)
YUVRGB_Scale=1 (0:TVScale 1:PCScale)
Luminance_Filter=0,0 (Gamma, Offset)
Clipping=0,0,0,0 (ClipLeft, ClipRight, ClipTop, ClipBottom)
Aspect_Ratio=4:3
Picture_Size=720x576
Field_Operation=0 (0:None 1:ForcedFILM 2:RawFrames)
Frame_Rate=25000
Location=0,0,0,6CEB9

7 0 2048 1 1 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12
7 0 43008 1 1 2 2 12 12 12 12 12 12 12 12 12 12 12 12 12
7 0 90112 1 1 2 2 12 12 12 12 12 12 12 12 12 12 12 12 12
7 0 139264 1 1 2 2 12 12 12 12 12 12 12 12 12 12 12 12 12
7 0 573440 1 1 2 2 12 12 12 12 12 12 12 12 12 12 12 12 12

Not only the entry for iDCT is different, look at the pattern of the rows.

Can anybody tell me if I did something wrong or is it a bug in the rebuilder?

SK

hobyho
31st May 2005, 08:31
The idct value in the d2v is neglected when you specify it with the decoder, in this case in the DVDRB generated AVS files which are used for encoding cells.

Quite simply the idct value in the d2v file is overrided by the one specified to the decoder. If you were to not have an idct value specified with the decoder, then it will be defaulted to the d2v file idct value.

I hope that helps to clear things up.

SkVid
31st May 2005, 08:48
Thanks for this clearing answer. And just to know that I understood what you said ...

mpeg2source("K:\D2VAVS\V01.D2V",idct=3)

This line in the .avs file including the idct setting is the relevant thing.

But what about the pattern of the rows. There is a difference between many twos and twelves. Has this no relevance?

hobyho
31st May 2005, 09:30
Yes that is correct.

As for the differing patterns, I cannot comment. Though I would speculate that its due to the different version of DGIndex being used (if that is indeed the case, which you have not indicated).

neuron has been doing quite a lot of changes, and the DVDRB generated file is based on the latest stable version (which is quite old compared to the latest version).

SkVid
31st May 2005, 11:00
I used dgindex with the timestamp 21.01.2005 20:11

jdobbs
31st May 2005, 12:30
You're overanalyzing. DVD Rebuilder builds the D2V specifically to work with itself. The only portion of the DGIndex suite of programs used by Rebuilder is dgdecode.dll -- and nothing higher than v1.1.0.

If you are generating a D2V with DGIndex independently and trying to use it with Rebuilder you are in for problems. There are hundreds of things that DVD Rebuilder does and records while it is doing its original scan and building the .D2V file. If you change the D2V -- the results are unpredictable and always bad.

The iDCT method specified in the D2V file is the one that will be used if not overridden by the parameter passed in mpeg2source()

SkVid
31st May 2005, 13:13
I don't want to change the .d2v file or exchange it by an independently generated one.

I discovered this by a coincidence and was just wondering why there is a difference.

So thanks all of you for your help and support! :)

jdobbs
31st May 2005, 15:17
:cool: