Log in

View Full Version : Gordian Knot uses a different IEEE-1180 to DGIndex?


Skywalkerjen
8th January 2006, 17:55
Hi there
I posted this in the DVD2AVI forum but Neuron couldn't help so he suggested posting in this forum:

DGIndex 1.4.6b2
Gordian Knot 0.35.0

I process the vob using DGIndex and the iDCT Algorithm: IEEE-1180
When I load the .d2v into Gordian Knot I use the advanced Save .avs dialogue.

I then use Gordian Knot to create the .avs file and one option is to change the iDCT used. If I set the idct to 0 then the .d2v value is used - the IEEE-1180 that I specified in DGIndex.

However, if I chose option 4: 64bit IEEE-1180 Reference that Gordian Knot specifies I get a different file - checked with an md5 checksum.

Which one is correct? The DGIndex IEEE-1180 or the Gordian Knot IEEE-1180?

I tested this further and found the following:

DGIndex and Gordian Knot produce exactly the same file when either is set to 32-bitMMX or Skal SSE MMX and any video length.

DGIndex and Gordian Knot produce exactly the same file when both are set to IEEE-1180 Reference but only when the video length is small (around 50 frames).

Thanks for any help

jggimi
9th January 2006, 21:16
I can't answer the question, but I will move this to Development, since its a question that can only be answered by someone on the GK development project (or someone who reads through the source.)

Guest
10th January 2006, 00:11
The latest DGIndex has a fix for the IEEE IDCT. The version of DGIndex in GK probably does not have the fix.

Skywalkerjen
10th January 2006, 01:16
Thanks for the replies.
That would definitely explain it - I'm using the DGIndex only and creating my own script.

Thanks for your time.

DrStoooopid
10th January 2006, 07:49
Personally I think a more useful thing would to be able to enable and disable plugin usage and some of the automated scripting.

For example:

on my machine I use for backups. It's a Celeron 500, (yes I know, not impressive, but that's all this machine has to do). It can't handle UnDot because the c500's don't use SSEMMX instructions, so when attempting to make an encode, Gknot barfs, and the only way to get it to work, is to edit the script by hand, and comment out any references to UnDot.

I've scoured my Gknot and other programs directories for any type of default template (I'd be happy being able to edit that once and forget about it.), but I can't even find a default template that I can edit for my own uses.

or I'm just retarded and it's staring right at me.

Skywalkerjen
10th January 2006, 14:47
I've created a template script that I copy relevant parts from Gordian Knot into:

#Source
#Deinterlace
#Cropping
#Resize

That's all I'm using.

iradic
10th January 2006, 19:24
What about "Advanced AVS window" usage?

Skywalkerjen
10th January 2006, 21:08
I've given up using the advanced AVS window as I'm not sure about the IDCT settings that are added to the script.

I also sometimes use other steps that have to be added to the script such as sharpening so I'm often editing the script anyway.

iradic
12th January 2006, 01:36
one question: what gknot has to do with idct ... i mean it doesnt put values that you didnt tell gknot to put ... or it does?

and yes ... adv avs window isnt perfect ...

Skywalkerjen
12th January 2006, 03:49
I have no idea what process is going on - just that Gordian Knot outputs a different file to DGIndex when both are supposedly doing the same thing.

swaaye
24th January 2006, 21:15
Is it bad to choose a different IDCT for MPEG2Source vs. the D2V (from DGIndex)?

Skywalkerjen
24th January 2006, 21:20
I assumed that the IEEE-1180 Reference was a mathematical constant and was the same which ever software used it?

swaaye
24th January 2006, 22:48
No, I didn't mean to say that it is bad. I'm asking the powers among us whether it is bad or not to choose different IDCTs between DGIndex and MPEG2source

Guest
24th January 2006, 23:28
I'm asking the powers among us whether it is bad or not to choose different IDCTs between DGIndex and MPEG2source No, it is not bad.