Auto Gordian Knot: current version 2.55

[len0x]

Originally posted by ydobon
A follow-up question: When AutoGK says "Found 63667 frames" Where does it get that number from? The .d2v file?

not only, its resulting number of frames (i.e. after IVTC etc)

Originally posted by ydobon

Also, do you use that number of frames to determine the chosen resolution?

of course

[len0x]

Originally posted by Taurus
Will do some more testing on the subject to see the behaviour regarding size, quality and speed.
Are you still interested in the results?

Yeah, but I know size and speed factors already (its only obvious that putting denoiser after resize you get better compressibility and faster speed). I'm very interested in subjective opinion about quality of the output. I myself would like to preserve as much visible details as possible (even if its grain). That's why I'm also interested to see how much difference is between mode 1 and 2 of removegrain (mode1=undot) you can see with your eyes.

[Taurus]

Originally posted by len0x
(its only obvious that putting denoiser after resize you get better compressibility and faster speed)

That's the same I found out.
Will do some more testing on visual appearance in the evening.
Mode=1 and Mode=2 compared to undot.
But I have to look for a more suitable source first, caus the last one used
is so grainy and bad (Buffy2). I have to search for another one for that job.

Cheers

Taurus

[ydobon]

Hi,

Originally posted by len0x
of course

Agh. Then I'm busted. The tweak I've done to avoid cutting my .ts files results in lower resolution the bigger is the cut

Well, at least I'm able to cut individual frames...

On a side note, with so many tests I noticed that if you have "no audio track" selected, AutoGK selects the first audio track when you open a new source. Is it a bug or a feature?


Thanks for the patience

[therealjoeblow]

Originally posted by richarddd
I have a movie on 2 dvd disks that I'd like to convert to 3 CDs (2100 Mb) with AutoGK. Can I do this (other than by converting 1 dvd to 2 CDS and the other DVD to 1 CD)?

There's a reference in the FAQ to putting all files in one directory and renaming to *_.1*; *_.2*; etc. If this is the way to do it, what do I do about the two stream info.txt and .ifo files (one each from each disk)? For example, if I have VTS_02 - Stream Information.txt, VTS_02_0.IFO, VTS_02_1.VOB, VTS_02.2.VOB, VTS_10 - Stream Information.txt, VTS_10_0.IFO, VTS_10_1.VOB and VTS_10.2.VOB, how would I rename?

Apologies if I missed a post where this was covered.

This is the way I do it:

you started with:

[files in DIR1]
VTS_02 - Stream Information.txt --> leave as is in DIR1
VTS_02_0.IFO --> leave as is in DIR1
VTS_02_1.VOB --> leave as is in DIR1
VTS_02.2.VOB --> leave as is in DIR1

[files in DIR2]
VTS_10 - Stream Information.txt --> ignore this file
VTS_10_0.IFO --> ignore this file
VTS_10_1.VOB --> rename to VTS_02.3.VOB and copy to DIR1
VTS_10.2.VOB --> rename to VTS_02.4.VOB and copy to DIR1

so now you have:

[files in DIR1]
VTS_02 - Stream Information.txt
VTS_02_0.IFO
VTS_02_1.VOB
VTS_02.2.VOB
VTS_02.3.VOB
VTS_02.4.VOB

[files in DIR2, which you can ignore]
VTS_10 - Stream Information.txt
VTS_10_0.IFO


You should be good to encode from DIR1 now for most 2 disc movies (I understand from others that there are a few oddballs that will give you problems with this method, but all I've tried worked correctly).

NOTE: the only major issue is that you will not be able to encode subs automatically this way, because the stream info for the 2nd disc isn't used in the process, so the subs would be screwed up. If you want subs, BEFORE you do the rename/move -> you should run a 'bogus' AutoGK session on each disc individuallty with any arbitrary video and audio settings, but with the correct subtitle settings for what you want. After the subs are indexed and rar'ed, you can then abort that session (as I said, do for both DVD discs, to produce proper subs for each one). Then SAVE THE SUB FILES SOMEWHERE!, and now you can carry on and do the rename/move bit, run the full encode, setting your target file size to take into account the size of the subs. After the encode, you'll then need to use the directvobsub tools to:

1) recombine your 2 separate sub files into 1 to match the overall length of time for the whole 2 DVD movie, and the
2) split the recombined sub file into separate chunks for each of your 3 CD's

The subs are a bit of a pain for multi-DVD movies, no doubt.

[Taurus]

Originally posted by len0x
That's why I'm also interested to see how much difference is between mode 1 and 2 of removegrain (mode1=undot) you can see with your eyes.

Sorry for the delay. Daytime job.
Well after looking for a new source for my comparision, I finaly decided to take another Buffy episode. This time Buffy 5.
It's much cleaner than the older episodes. PAL 16:9 25fr
Made an AVS scipt using AutoGk. 720*416 Resolution. Xvid Q:2 as used by AutGk. No more filtering; degrainers at the end of the scipt.
Mode=8 just for testing(incredible)(lanczos_undot for comparision)
-----------------------------------------------------
Buffy_lanczos4_removegrain_mode8.avi----42,252 MBytes
Buffy_lanczos4_removegrain_mode2.avi----43,023 MBytes
Buffy_lanczos4_removegrain_mode1.avi----44,537 MBytes
Buffy_lanczos4_undot.avi----------------44,605 MBytes
Buffy_lanczos_undot.avi-----------------44,998 MBytes
Buffy_lanczos4.avi----------------------45,371 MBytes
Buffy_lanczos.avi-----------------------45,604 MBytes
-----------------------------------------------------
So here are my subjective conclusions:

On a phosphor display it's hard to see any difference between undot and removegrain Mode=1&2. Even strong zooming into the clip does not show remarkable difference. If you want to stay as conservative as possible, my opinion is: use Mode=1. If you want some extra encoding headroom,
Mode=2 is the one. As I wrote, to my eyes both look almost identical,
even at zooming into the pictures.
But I think still undot is doing a fairly good job compared to a non filtered file.

But the real headshaker is Mode=8. At this greasy source it dramatical changed necessary bitrate and quality(great).
But this was just for fun

Hope this is on any help.

Cheers

Taurus

[len0x]

Originally posted by ydobon
On a side note, with so many tests I noticed that if you have "no audio track" selected, AutoGK selects the first audio track when you open a new source. Is it a bug or a feature?

May be both in a sense that other behaviour was never requested

[len0x]

Originally posted by Taurus
But the real headshaker is Mode=8. At this greasy source it dramatical changed necessary bitrate and quality(great).

Really? What is so dramatical - size went down just 0.8M, when mode 2 gave 1.5Mb less than mode 1...

[Taurus]

Originally posted by len0x
Really? What is so dramatical - size went down just 0.8M, when mode 2 gave 1.5Mb less than mode 1...

You're right...
Just the visual appearance. A little wax on human skin but the overall
impression was real natural. It all depends on the source.
I would NEVER use a degrainer on clean sources. Purist.
But this is another story.
P.S.: Just compared a few samples lanczos versus lanczos4.
Strange, the filesize is less on lanczos4.
And I can't tell a difference in visuality

Have a good day (night)wherever you are.

[len0x]

Originally posted by Taurus
P.S.: Just compared a few samples lanczos versus lanczos4.
Strange, the filesize is less on lanczos4.
And I can't tell a difference in visuality

ok, lets proceed with lanczos4 and see where it gets us

[richarddd]

Originally posted by therealjoeblow
This is the way I do it:

Thanks. That seems to work.

As a bit of background, I was ripping Lawrence of Arabia, which is approx 2 hours on disk 1 and 1 hour on disk 2, and having sync problems on the second disk. After some testing and searching this forum, I figured out that the problem was caused by the intermission included at the beginning of disk 2 (blank screen & silence followed by music). If I unchecked all the intermission chapters (about 20 chapters of about one second each and about 5 chapters of about one minute each) before ripping, all is well. Doing everything as one job resulted in approximately equal quality for each CD, which is what I wanted.

[haubrija]

I have a quick question (apologize if its been asked before)

I've been using AutoGK to archive my .ts streams recorded from a Fusion3 card. I edit out commercials with HDTV2MPEG2 before.

I had been told that ProjectX would help me sorten out some Audio Sync errors I had been having.

I edit the stream with HDTV2MPEG2 first. I then loaded the .ts into ProjectX. I chose to save it to a .ts stream.

When ProjectX is done doing its thing, I attempt to load the resulting .ts into AutoGK. For some reason the audio is not read after running it though ProjectX.

It looks like the PID have been changed. What was once 0x21 and 0x24 (audio and video respectively) have become 0xE0 and 0xFFFFF. Is there something I've been doing wrong?

Thanks in advance.

[TreborPugly]

Putting the files on CD.

I've looked over the How-to guides for Auto GK, and some other DivX creation tools. Nowhere do I see anything about the format of CD's you make for stand-alone devices. Do you just burn data CD's, with the *.avi files on them and some stand-alone DVD players will recognize the AVI's? Or do you need to create something like VCD's?

And are there any stand-alones that will recognize DVD's with DivX files on them? For example, I like the idea of creating one DVD with several kid's DVD's worth of shows on them, reducing the number of disks I need to have available to the kids.

[btothec2]

Yes,

You just burn them as data cd's. And almost all f not all standalone Divx players will allow you to burn several avi's on on dvd.

B

Originally posted by TreborPugly
Putting the files on CD.

I've looked over the How-to guides for Auto GK, and some other DivX creation tools. Nowhere do I see anything about the format of CD's you make for stand-alone devices. Do you just burn data CD's, with the *.avi files on them and some stand-alone DVD players will recognize the AVI's? Or do you need to create something like VCD's?

And are there any stand-alones that will recognize DVD's with DivX files on them? For example, I like the idea of creating one DVD with several kid's DVD's worth of shows on them, reducing the number of disks I need to have available to the kids.

[Carraway]

Originally posted by Carraway
I recall a while ago that there was talk that AutoGK was producing dark encodes. [...] I have several just generally dark sources [...] how feasible (or infeasible!) would a .filters file be, in which you could add internal avisynth filters like tweak(bright=xx) or levels(1,1.x,255,0,255) that would then be appended to the end of the comptest and movie avs files?

Originally posted by len0x
wait for config file support (which is at least 1 month away from next stable release date).

Two questions:

First, len0x: any update on support for levels() and tweak() to brighten unusually dark Xvid encodes?

Second, anyone can answer, but len0x will probably know best: if avisynth filters are appended to the end of both the movie avs files as well as the comptest avs files, will that still throw off the values of the comptest?

EDIT:

About the second question, by "appended to the end of the avs file" I of course mean inserting them immediately prior to the resizing. Applying filters after resizing would change their use in the comptest since the comptest uses a different resolution than the movie.avs file would.

[len0x]

Originally posted by haubrija
It looks like the PID have been changed. What was once 0x21 and 0x24 (audio and video respectively) have become 0xE0 and 0xFFFFF.

that can be a problem. AutoGK and its applications only support four digits PIDs, so FFFFF may not work. You can assign another PID manually in options of ProjectX, I believe.

[len0x]

Originally posted by Carraway

First, len0x: any update on support for levels() and tweak() to brighten unusually dark Xvid encodes?

no

Originally posted by Carraway

About the second question, by "appended to the end of the avs file" I of course mean inserting them immediately prior to the resizing. Applying filters after resizing would change their use in the comptest since the comptest uses a different resolution than the movie.avs file would.

There is absolutely no difference for comp test where filters are...
Please read my reply to therealjoeblow on the previous page to see what is affecting comp test for AutoGK's purpose (basically everything, but noise filters the most).

P.S. by comp test I mean resizing algorithm based on comp test and used in AutoGK.

[len0x]

As seen in today's news in www.doom.org jdobbs came up with an interesting idea of distributing an application to the donators first and some time later to the public. That is very appealing idea for me - what does the public thinks?

[bourtzovlakas]

My opinion is,that AutoGK is your program,your creation,your "child"!Therefore,you are justified to distribute it however you like....
But,I have one question.....How can you prevent a donator to release it to the public,before you do....
Anyway,if you choose to distribute AutoGK like this....I will gladly donate an amount,for no other reason,but to akwnoledge and support your efforts....

[Mc|Atm0s]

How about new XviD-1.1?