Auto Gordian Knot: current version 2.55

[Hobojobo]

@therealjoeblow

Would it be possible to add a hidden .crop feature, like miljbee proposed.
I would use GK :D to auto/manual crop the movie. (I like to see the result straight away.)
Then I could put the numbers into the .crop file and I am done.
That would be perfect.

I would like to crop the tv shows to 640x480, 4:3 ratio.
It does not matter losing a few pixel on the borders.

[((( atom )))]

@lenox

when i was a little interrupted while setting up my last pile of movies, i couldn't recall if i had set specific settings right, wich i wanted for specific movies and i ended up deleting the entire queue and did it again.

it would be very practical if one could doubleclick on any job in the queue and have the setting that were used loaded.

this might also be usefull, if i wanted to "copy" settings i used for a specific movie to another job. for example i might want to use different audio settings since i have some old movies, wich wouldn't be worth being kept in ac3, that was my case..

what do you think of it?

[therealjoeblow]

Originally posted by AlmanNewfred
Thanks for this little tool. But the frame calculator seems to be missing an option for film frame rate (23.976 fps).

Did you read the discussion in the readme.txt file??? Please do.

[AlmanNewfred]

Originally posted by therealjoeblow
Did you read the discussion in the readme.txt file??? Please do.

Yes I did. I don't pretend to understand the intricacies of frame rates and the conversions AGK does, but I do know that using 30 fps for the frame calculation gives does not seem to give correct results for a pure film source, i.e. the credit frame ends up greater than the total number of frames AGK finds after demuxing. Using 24 fps does seem to give the right frame - it matches what you get from loading the vobs into vDubMod, and the final encode looks right.

[len0x]

Originally posted by AlmanNewfred
Yes I did. I don't pretend to understand the intricacies of frame rates and the conversions AGK does, but I do know that using 30 fps for the frame calculation gives does not seem to give correct results for a pure film source, i.e. the credit frame ends up greater than the total number of frames AGK finds after demuxing. Using 24 fps does seem to give the right frame - it matches what you get from loading the vobs into vDubMod, and the final encode looks right.

please post all numeric info to support your claim.
(including AutoGK log)

[Hesse]

Originally posted by AlmanNewfred
Yes I did. I don't pretend to understand the intricacies of frame rates and the conversions AGK does, but I do know that using 30 fps for the frame calculation gives does not seem to give correct results for a pure film source, i.e. the credit frame ends up greater than the total number of frames AGK finds after demuxing. Using 24 fps does seem to give the right frame - it matches what you get from loading the vobs into vDubMod, and the final encode looks right.

I can confirm this behavior. If the source is detected as pure film, I must enter the FF frame number to get a correct credits encode. Otherwise, I enter the nonFF frame number.

Jesse

[len0x]

Originally posted by Hesse
I can confirm this behavior. If the source is detected as pure film, I must enter the FF frame number to get a correct credits encode. Otherwise, I enter the nonFF frame number.


And I can confirm that its a bug :) I forgot as usual about pure Film sources which are represented differently in AutoGK. already fixed for the next version.

[Hesse]

Originally posted by len0x
And I can confirm that its a bug :) I forgot as usual about pure Film sources which are represented differently in AutoGK. already fixed for the next version.

Wow Lenox, you are fast! Thanks for the fix. So we should enter the nonFF 29.997 Frame number always in the next version?

Jesse

[len0x]

Originally posted by Hesse
Wow Lenox, you are fast! Thanks for the fix. So we should enter the nonFF 29.997 Frame number always in the next version?


yes, next version is up btw :)

[DSPguru]

Originally posted by len0x
yes, next version is up btw :) great :D.
is 'hybridgain' back to default then ?

[Firewave]

I used AutoGK the first time today. It works fine, better than any other tools I tried. And at the moment there are just two little things, that are on my mind:

First it's a little "freature" request: Could you ass the length of the input to the "Video Info"? It could be quite helpful, if you aren't sure about the length.

Send thing is a question to everybody: What's the best setting, if you want to encode a short movie? Like having a around 70 megs result for a 7 minute movie. Is it the "Custom size" option?

So long
Firewave

[gircobain]

Originally posted by Firewave
Send thing is a question to everybody: What's the best setting, if you want to encode a short movie? Like having a around 70 megs result for a 7 minute movie. Is it the "Custom size" option?
If target size is not a concern, I believe the best option is to encode by quality. Most people are satisfied with a target quality within 75-85%. On the other hand, if quality is the main concern, you are better off choosing maximum quality under predefined size.

[ookzDVD]

@len0x,

First I wanna thank for your AutoGK which finally support the world's
finest codec (Xvid 1.0 beta). :)

Second, the hidden options specially the ".credits".
it will be better if we use the timestamp for the credit start
point instead of frame. use the timestamp will avoid the miss
understanding for FF and nonFF frame #.

suggestion:
---
1:23:23
20
FALSE
---

Thank you.

[manono]

Firewave-

There are a couple of ways to get the movie length already. One is from DVD Decrypter when you decrypt the vobs (assuming you're in IFO Mode, as you should be). And the other is by opening the .d2v in GKnot, after AutoGK makes it.

And if you want to make the short video for a specific size, then, yes, use a custom size. But pay attention to the compress test results and the chosen final resolution, as it's often hard to guess the size in advance, and get a good result. Or follow gircobain's suggestions.

[Taurus]

@ DSPguru

is 'hybridgain' back to default then ?

As I can see from the log text, hybridgain is not used.
Instead postgain is used, as have been done in the (all?)older AutoGk betas. (-g1)

Cheers

Taurus

[manono]

OK, I'll take the rap for the Postgain setting. Back in the early days of AutoGK, people were complaining about soft volume. It was a concern of mine also, as I wasn't happy with the volume levels when using the default settings of the BeSweet GUI. I went searching in the Audio Forum for a way to remedy it, and came up with the suggestion to use Postgain, instead of Hybridgain. I tried it out and found that the result was, indeed, louder with Postgain applied. So I mentioned it to len0x. If for some reason that I can't understand with my limited knowledge of audio encoding Hybridgain is better, then, of course, I bow to DSPguru's expertise in the matter.

[madprofessor]

i find autogk awesome and the development and support is almost unbelievable.

one question though, i select divx for codec generally. is autogk non-related to the codec ? or is it more optimized for one of the two codecs (divx/xvid). if so i would like to use the one that you work more onto . i hope this is not a lame question :rolleyes:
i like xvid for its non-commercial touch but divx will always be the oldest ;)



ps: i hate it when an actor wears vivid red clothes in a movie ... pixels all over the place :angry: :D :D :D

[len0x]

Originally posted by Firewave
First it's a little "freature" request: Could you ass the length of the input to the "Video Info"? It could be quite helpful, if you aren't sure about the length.


unfortunately there is no way for doing that. Stream info file doesn't contain that info...

[len0x]

Originally posted by ((( atom )))
symptom: after installing autogk i can run it. ONCE! - it won't start again until i rerun the installer. sometimes, though i even have to rerun the installer twice in order for gk to start.
if it refuses to start, i get: autogk has experienced a problem and needs to close (translated).


HT issue ?

[len0x]

Originally posted by ookzDVD
@Second, the hidden options specially the ".credits".
it will be better if we use the timestamp for the credit start
point instead of frame. use the timestamp will avoid the miss
understanding for FF and nonFF frame #.


Hidden Option tool done by therealjoeblow already does that.

[Tuning]

Thinking no one else has requested before....:p

A Small Feature Request: Direct muxing of DTS audio.
Especially useful for HDD saving and multi CD rips.
Thanks.:)

[((( atom )))]

Originally posted by len0x
HT issue ? no, happens with and without ht enabled.

[((( atom )))]

i wanted to view the printable version of this thread on one page in order to search for something easier and i got:

Fatal error: Allowed memory size of 8388608 bytes exhausted (tried to allocate 2188533 bytes) in /home/httpd/docs/vbb/admin/functions.php on line 582

is this thread to long? ;)

[Spiders]

Originally posted by len0x
Does this happen with regular GK as well ?

I do not know since use only Auto GK. À on idea computer must be switched off? The Program was tested on Windows 98?

[DSPguru]

Originally posted by manono
OK, I'll take the rap for the Postgain setting. Back in the early days of AutoGK, people were complaining about soft volume. It was a concern of mine also, as I wasn't happy with the volume levels when using the default settings of the BeSweet GUI. I went searching in the Audio Forum for a way to remedy it, and came up with the suggestion to use Postgain, instead of Hybridgain. I tried it out and found that the result was, indeed, louder with Postgain applied. So I mentioned it to len0x. If for some reason that I can't understand with my limited knowledge of audio encoding Hybridgain is better, then, of course, I bow to DSPguru's expertise in the matter. hybridgain iZ the way to go, and should be the default mode.
if someone is into extreme normalization, he can change his personal default mode to using -azid( --maximize ).
postgain of 18db [like you could see in the logfile of "-ota( -G 1 )"] will never make sense to me.

[len0x]

Originally posted by DSPguru
hybridgain iZ the way to go, and should be the default mode.
if someone is into extreme normalization, he can change his personal default mode to using -azid( --maximize ).
postgain of 18db [like you could see in the logfile of "-ota( -G 1 )"] will never make sense to me.

OK, I'll change that to hybridgain then and after I have a config everyone can customize that (the main problem now in AutoGK that if you're not satisfied with audio level there is nothing you can do about it apart from encoding it yourself and remuxing)

P.S. Personally I have no opinion on the subject as very rarely do mp3 myself.

[GizmoDerMokwai]

hi, dunno know if it's really a bug, but autogk saying chosen resolution ***x304, but final avi is ***x288...


===========================================================
[21.04.2004 08:45:59] Duration was: 12 minutes 35 seconds
[21.04.2004 08:45:59] Speed was: 8,07 fps.
[21.04.2004 08:45:59] Compressibility percentage is: 20,38
[21.04.2004 08:45:59] Using softer resizer.
[21.04.2004 08:45:59] Using softer resizer.
[21.04.2004 08:45:59] Chosen resolution is: 512x304 ( AR: 1,68 )
[21.04.2004 08:45:59] Predicted comptest value is: 32,35
[21.04.2004 08:45:59] Running first pass.
[21.04.2004 08:45:59] Writing the following script to D:\tmp\gk_tmp\asterixinamerika_movie.avs
===========================================================
LoadPlugin("c:\PROGRA~1\DVDRIP~1\AutoGK\filters\mpeg2dec3dg.dll")
LoadPlugin("c:\PROGRA~1\DVDRIP~1\AutoGK\filters\autocrop.dll")
LoadPlugin("c:\PROGRA~1\DVDRIP~1\AutoGK\filters\undot.dll")
LoadPlugin("c:\PROGRA~1\DVDRIP~1\AutoGK\filters\kerneldeint.dll")

function getOrder(clip c) {
order = GetParity(c) ? 1 : 0
return order }

movie = mpeg2source("D:\tmp\gk_tmp\asterixinamerika.d2v")
cropclip = autocrop(movie,mode=0,wmultof=4,hmultof=4,samples=10,aspect=0,threshold=35)
c_width = width(cropclip)
c_height = round(height(cropclip) / 1.09401709401709)
input_par = float(c_width)/float(c_height)
fixed_aspect = input_par > 1.4 ? 1.09401709401709 : 1.06666666666667
c_height = round(height(cropclip) / fixed_aspect)
input_par = float(c_width)/float(c_height)
out_width = 512
out_height = round(float(out_width) / input_par)
hmod = out_height - (floor(out_height / 16 ) * 16)
out_height = (hmod > 4) ? (out_height + (16 - hmod)) : (out_height - hmod)
new_aspect = (float(out_width) / float(out_height)) / fixed_aspect
KernelDeInt(movie,order=getOrder(movie),sharp=true)
autocrop(mode=0,wmultof=4,hmultof=4,samples=10,aspect=new_aspect,threshold=35)
Undot()
BicubicResize(out_width,out_height,0,0.5)


Gspot info:
XviD, 876 kb/s, 512x288 (1.78:1) [=16:9]

just wanna to drop you this info...

[len0x]

Originally posted by GizmoDerMokwai
hi, dunno know if it's really a bug, but autogk saying chosen resolution ***x304, but final avi is ***x288...


Nice catch. Unfortunately there is nothing I can do - aspect ratio displayed is the same as comp test AR. Since autocrop crops automatically during encode its only a guess since autocrop does everything inside. So in some rare cases you would see this...

[GizmoDerMokwai]

hehe, ok, just wanted to give you that info... :)

[hujer]

lenox > is there public cvs server for autogk ? i would like to fix the HT bug coz its annoying me very much, when i set compatibility with windows 98 for autogk it runs virtual dub in some weird mode
with one cpu affinity and makes my whole computer superslow, i have to run into taskmanager and change affinity for the vdub process everytime, if there isnt public cvs please contact me on hujer@users.sf.net and let me know if i can help you fixing this crash somehow, its really pissing me off, thanx for autogk btw :)

[ookzDVD]

Originally posted by len0x
Hidden Option tool done by therealjoeblow already does that.

@len0x,
I mean in your .credits directly use the timestamp instead of the frame #. So the conversion is done by your internal code.

thank you.

[len0x]

Originally posted by ookzDVD
@len0x,
I mean in your .credits directly use the timestamp instead of the frame #. So the conversion is done by your internal code.


well, then we would need reverse conversion if a person knows only frame number (like many ppl actually do including me - they have no problems getting frame number from regular GK). So it doesn't really matter to me if we have tools that do the job and something is already done then I don't see a reason of re-writing it :)

[therealjoeblow]

Originally posted by len0x
well, then we would need reverse conversion if a person knows only frame number (like many ppl actually do including me - they have no problems getting frame number from regular GK). So it doesn't really matter to me if we have tools that do the job and something is already done then I don't see a reason of re-writing it :)

@len0x,

I sent you a new version which can read CLI parameters - you might be able to add a button to AutoGK now to launch the "Hidden Options.exe" after the user selects the input directory, that would probably simplify things a bit for many users.

v-0.6 (April 22, 2004)
----------------------
-Added CLI (Command Line Interface) - 'Hidden Options.exe' can now
accept the target path as a command line parameter. I coded it
to be fairly forgiving, you can use one of the following formats:

"z:\hidden options.exe" x:\path name
"z:\hidden options.exe" -x:\path name
"z:\hidden options.exe" /x:\path name

You can enter the path in all caps, all lower case, or a mix, it
doesn't matter. You can also either add or omit the trailing "\"

Also checks if the drive and path exists, and provides error
messages if not.

-Added error checking when selecting an empty CD or Floppy Drive
from the drivelist pick-box - program would crash with runtime
error before (not like you're going to actually work from one of
those sources, but you might pick it by accident).

[nightrhyme]

Originally posted by hujer
lenox > is there public cvs server for autogk ? i would like to fix the HT bug coz its annoying me very much, when i set compatibility with windows 98 for autogk it runs virtual dub in some weird mode
with one cpu affinity and makes my whole computer superslow, i have to run into taskmanager and change affinity for the vdub process everytime, if there isnt public cvs please contact me on hujer@users.sf.net and let me know if i can help you fixing this crash somehow, its really pissing me off, thanx for autogk btw :)

Would really be nice if you could help make AutoGK run on HT cpu's.
That bug really irritates me too.
I usually disable Hyper Threading in bios before encoding with GK :(

[hujer]

Originally posted by nightrhyme
Would really be nice if you could help make AutoGK run on HT cpu's.
That bug really irritates me too.
I usually disable Hyper Threading in bios before encoding with GK :(

well when i disable HT in bios my windows doesnt boot :), anyway i would really like to fix it, but i need the src first

[rotvel]

It would be nice to have a priority setting for AutoGK. Similar to the one in DVD2SVCD ("Process priority for all used programs" -> Idle|Normal|High).

Mainly for running AutoGK and still be able to use the PC for other stuff (using the "idle" setting).

Regards /max

[len0x]

Originally posted by hujer
but i need the src first

Source will not be provided, please stop asking about them.

I had another ago but still unable to detect the problem so far...

[hujer]

Originally posted by len0x
Source will not be provided, please stop asking about them.

I had another ago but still unable to detect the problem so far...

just wanted to help, heh

[jeremymacmull]

RE HT bug(?)

lenox is in the process of trying to figure out what the bug is so far after numerous attempts theres been no luck

he is hard at work at it so give him some time and support as it must be frustrating when he cant test directly as he does not have HT cpu



otherwise on my cpu 1.18b starts up 50% of the time and the other 50% it says runtime error 216 at XXXXXX

I dont have it set to compatibility mode ie im running it normally

when it does start up it runs perfectly with both cpus as affinity and on the right priority without slowing down my comp for those of you who are using compatibility mode or turning off ht try not doing so and just keep clicking on the exe till it starts up (as i said on my comp its 1/2 chance of it working but once started everything works!)

JEREMY

[hujer]

Originally posted by jeremymacmull
RE HT bug(?)

lenox is in the process of trying to figure out what the bug is so far after numerous attempts theres been no luck

he is hard at work at it so give him some time and support as it must be frustrating when he cant test directly as he does not have HT cpu

otherwise on my cpu 1.18b starts up 50% of the time and the other 50% it says runtime error 216 at XXXXXX

JEREMY

well, as i have experiences with debugging delphi projects, and it does crash on my machine everytime, i thought i could fix it in few minutes so offered help, i undrestand this is nearly impossible to fix without access to ht system

lenox if you still have problems finding the problem after some time, feel free to mail me, to debug it one only needs the code that runs on startup, that should be only few onshow/onload functions and you can strip rest of the project..., well anyway, good luck, its a nasty problem

[nightrhyme]

Originally posted by jeremymacmull
RE HT bug(?)

lenox is in the process of trying to figure out what the bug is so far after numerous attempts theres been no luck

he is hard at work at it so give him some time and support as it must be frustrating when he cant test directly as he does not have HT cpu



otherwise on my cpu 1.18b starts up 50% of the time and the other 50% it says runtime error 216 at XXXXXX

I dont have it set to compatibility mode ie im running it normally

when it does start up it runs perfectly with both cpus as affinity and on the right priority without slowing down my comp for those of you who are using compatibility mode or turning off ht try not doing so and just keep clicking on the exe till it starts up (as i said on my comp its 1/2 chance of it working but once started everything works!)

JEREMY

I have no problem starting up the exe and starting the encoding. AutoGK just crashes half way through encoding with HT enabled.
Also the pc acts funny with AutoGK running and HT enabled.
It acts like no more memory is available very sluggish.
AutoGK uses about 400 - 500MB memory here. I have 1024MB.

BTW: My system : Asus P4p800, p4c 3.0GHz, 2 x 512Mb pc3200 Mushkin special.

[hujer]

Originally posted by nightrhyme
I have no problem starting up the exe and starting the encoding. AutoGK just crashes half way through encoding with HT enabled.
Also the pc acts funny with AutoGK running and HT enabled.
It acts like no more memory is available very sluggish.
AutoGK uses about 400 - 500MB memory here. I have 1024MB.

BTW: My system : Asus P4p800, p4c 3.0GHz, 2 x 512Mb pc3200 Mushkin special.

do you have win98 compatibility mode on ?, also check in taskbar how many cpus (affinity) is the process using

[jeremymacmull]

Ive no got the compatibility mode on (as when i do the system is sluggish)

but even without it on ive never had autogk crash on me

JEREMY

[nightrhyme]

Originally posted by hujer
do you have win98 compatibility mode on ?, also check in taskbar how many cpus (affinity) is the process using

I don't have compatibility mode on. Would never use it :-)
It runs on both logical CPU's. But still only uses 50% CPU which is an indication that the program don't utilize the HT feature properly.

For instance try TMPGEnc which is optimized for HT and uses 100% CPU with HT enabled.

[AlmanNewfred]

I don't have a HT CPU, and I have no idea if this is relevant, but sometimes vDubMod will stop and say a thread deadlock was detected. This usually happens at the end of the first pass. It looks like vDubMod has at least one bug in the way it handles threads.

[jeremymacmull]

It uses about 70-80% cpu here not 50 and I know divx is optimised for HT

although that extra 30% would make quite a jump in encoding speed!!!

JEREMY

[len0x]

Originally posted by hujer
lenox if you still have problems finding the problem after some time, feel free to mail me, to debug it one only needs the code that runs on startup, that should be only few onshow/onload functions and you can strip rest of the project..., well anyway, good luck, its a nasty problem

If I'm correct in understanding the problem (I might be not):

Hyperthreading (but not dual systems) execute the code into two pipelines within the same processor (like two threads you may imagine), so if you are accessing the same variable twice in a row for instance it may be that access instructions will be put to different pipelines and if second pipeline reaches the execution point first then variable will get inconsistent state. Moreover if variable was supposed to be initialized with the first access then second access can get famous runtime error 216. I think this is what causing it - and if you think 3GHz limit can be even explained: since the code is executing faster then there is more chances that some parts of the code will overtake others. I don't know whether its hardware/os/compiler problem, but as programmatically I'm not doing any threading on startrup (unless delphi does CreateForm in a separate thread on its own) I would not expect this behavior. I'll try to put artificial delay between calls to global variables to see if that helps...

Any other thoughts ?

[hujer]

Originally posted by len0x
Any other thoughts ?

well, if you dont have other threads in your app i think it may be external problem or maybe problem in some delphi component ? do you use some 3rd party components ?

"IF" it is some global variable causing it can be fixing with proper locking with entercriticalsection()

createform dont run in second thread unless you do something in special way

you may want to try adding some debug msgboxes into startup functions and contructors to find the section in which it crashes

i was googling a little and i found 2 notes about delphi and HT crashes, one was related to some broken component from 3rd party, and one was some bug in TListView in vsReport mode, thats why if you dont use threads on startup i think it could be some component causing

btw it is (i think, dont have delphi here at work) possible to compile exe file with debug info, so i can run debugger here and check callstack in which function it crashed, that may be helpful for you

[len0x]

Originally posted by hujer
well, if you dont have other threads in your app i think it may be external problem or maybe problem in some delphi component ? do you use some 3rd party components ?


I have removed all VCL GUI components plus all 3rd party stuff and it still crashes, so I think it is global vars which are causing this (I removed all of them and waiting for confirmation if it works or not)

I obviously do have some threads (nowhere near startup), but I removed them as as well for testing...

Originally posted by hujer
"IF" it is some global variable causing it can be fixing with proper locking with entercriticalsection()


I was about to try that, but had to be sure that it is the cause.
It is still beyond my understanding how this is possible if there is no threading? Critical section may not even help since runtime error 216 means that variable was not initialized and therefore nothing can be accomplished via critical section (how global var is not initialized - again good question)

Originally posted by hujer

btw it is (i think, dont have delphi here at work) possible to compile exe file with debug info, so i can run debugger here and check callstack in which function it crashed, that may be helpful for you

I'll check.

[hujer]

Originally posted by len0x
Critical section may not even help since runtime error 216 means that variable was not initialized and therefore nothing can be accomplished via critical section (how global var is not initialized - again good question)


maybe you try to access variable that you have just destroyed, but no clue how that can happen in fast ht cpus only with only one thread :)

btw which delphi version do you use ?

and are you on some irc channel ? may be faster, i can also run the test binaries for you if needed