Access Violation

[Caliber]

I am getting the following error while trying to use Avisynth.
Avisynth: caught an access violation at 0x0ae81d1a, attempting to read from 0x0aeab7c8
I tried to reinstall windows, but still I get the error. I never used to get this error until I got my new computer - so either it's are hardware or driver issue I guess.
Soyo Dragon+ Mobo
Athlon 1800+ Proc
Xtasy GeForce 3 Ti200 Graphics
Onboard Lan, Sound
Crucial 256MB x 2 DDR-DIMMS
Windows XP Professional
Any suggestions? I've ran many tests on the Ram, tried removing one or the other and still no help.
Any suggestions at all would be appreciated.

[Guest]

Are you running in Turbo mode or pushing the FSB? Try throttling those back if you are. Some memory cannot tolerate that.

[Dali Lama]

Yes, I also get this error message sporatically. Even thoug I am using the same hardware as before. I think it may be due to some "incompatabiliy" in filter use in avisynth. For instance I couldn't use bilinear resize and decomb together. Instead I used smart resize and it worked.

This may not be the "real" solution, but it worked for me,

Dali

[Caliber]

Surprisingly enough I've tried it with no filters at all - no resize or anything and I still get the error. I think it may have to do with the VOB files I'm using as a source. When I open them in DVD2AVI, I notice some artifacts that appear at the exact time I get the error when using the avs file. So it must be with my source - I just ripped it again to test if this is true - the artifacts are in a different place this time and now AVISynth throws the error at that spot.
I think I am having hardware issues. I am using the VIA KT266A chipset and I do have issues with my DVD-ROM drive. It will not do autoplay - I get errors when trying to read certain things from it, so I don't think the movie is being properly copied from it. I don't know how to prevent these errors though. Guess I'll keep trying different Drivers.
Thanks!

[meleth]

Did any of you guys install something lately? Cause i started getting these errors myself yesterday. It seems like it is telecide() that's throws them.

[Guest]

I'm tellin' ya! I have the same motherboard and had the same errors. They all went away when I went into the BIOS and set Normal mode and put the FSB at 133, where it should be. :)

I've run 4 hour encodes with Telecide(postprocess=true), Decimate(), Crop(), and BicubicResize(). It's solid as a rock.

[meleth]

Originally posted by neuron2
I'm tellin' ya! I have the same motherboard and had the same errors. They all went away when I went into the BIOS and set Normal mode and put the FSB at 133, where it should be. :)

I've run 4 hour encodes with Telecide(postprocess=true), Decimate(), Crop(), and BicubicResize(). It's solid as a rock.

Well i have an a7v and also get eceptions all of a sudden. To give you some info about how the problems manifest.

If i run telecine without post processing i get an exception straight away. If i run it with it enabled, i will first get some noise, then a frame or two of the movie and then it will throw an exception. So it's clearly something with telecide cause running decimate without it works fine.

Only thing i did that day was upgrade the ogg directshow filters and for the first time tried running telecide with postprocess off. That was the first time i got the exception and i've gotten it every single time after that.

[Guest]

Postprocess off runs *less* code, not more or different code, so just changing that is not going to account for crashes.

I suggest that you back out the new Ogg stuff and try again without it.

[meleth]

Well i did remove the ogg submux filter but that didn't help.

[Guest]

You said that before you installed the Ogg stuff Telecide worked fine, is that correct?

What makes you think Telecide is throwing the exception?

[meleth]

Ok, i've found the problem here. The problem is the cropping. The help file says that "Telecide requires that the input width be a multiple of 4."

mpeg2source("G:\blah.d2v")
SwapFields()
Telecide(false,false,false)
Decimate(5)
trim(30501,0)
crop(12,2,696,476)
TemporalSmoother(2,1)
BicubicResize(512,384,0,0.5)

174*4=696 so it is multiple by 4, still commenting it out solves the whole exception problem.

[Guest]

It's good to know you have found a cause, but I wouldn't stop there if I were you. Telecide and Decimate come *before* your crop call, so it's very hard to imagine that the crop is affecting them. It must be one of the filters following the crop: TemporalSmoother() or BicubicResize(). You might try leaving the crop and then adding them one at a time to see which is reacting badly.

Also, I'm a little surprised at your SwapFields call before Telecide. Do you know why it is required?

[meleth]

I assumed you change the field order with it, top/bottom odd/even or however you chose to name them. Am i wrong?



I'll test that leaving crop in and adding the other one by one. Whatever it is that might be causing this, it is without a doubt telecide that is throwing the exceptions. I'll let you know my findings when i get home from work in 9h or so.

[Guest]

Telecide does not change the field order unless you set swap=true.

I'll ask again. What makes you think that Telecide is throwing the exceptions? Since you have no doubt, you ought to be able to answer that. :)

[meleth]

Originally posted by neuron2
Telecide does not change the field order unless you set swap=true.



I'll ask again. What makes you think that Telecide is throwing the exceptions? Since you have no doubt, you ought to be able to answer that. :)

Oh, so swapfields() is of no use? or do i use it together with swap=true?

Cause the same avs script use with avisynts ivtc filter, greedyhma and ivtc4. It's just decomb that doesn't work, when i use telecide that is. Using just decimate works however.

[meleth]

Ok, all i can say is that i'm baffled it doesn't matter which one i remove aslong as i remove one of them.

It almost seems that telecide,temporalsmoother,bicubicresize and crop don't mix. Remove any of the 4 and it works.

[meleth]

@neuron

So what do you think? Is the problem located in telecide or not?
As it looks to me atleast that is where the problem lays in telecide since the other 3 filters works with the other ivtc filters out there.

Only other place were i think the bug might be located is in avisynth itself, but one would think a bug like that would have been found by now.

I'm using ME on a TB 800mhz / Asus a7v.

[Guest]

@meleth

I cannot find any problems in Telecide and many people, including myself, use it without any problems.

Telecide will put a heavy load on memory and CPU bandwidth. It is possibly exposing instabilities in your system. Are you overclocking your system in any way?

I also return to observations like the one where you said removing a crop call that was *after* Telecide made the problem go away. There is simply no way that the crop could affect the operation of Telecide in that situation.

The only way we can proceed is if you can make available a short clip and corresponding avs script that causes the problem. Unless I can duplicate the crash, it is all speculation.

[meleth]

K, i'll see what i can do about it when i get home from work.

[Guest]

Please answer my questions if you expect me to help. :)

Are you overclocking your system in any way?

[meleth]

Nope, no overclocking at all.

[Guest]

@meleth

Thank you for sending me your test clip. I was able to immediately duplicate the problem, and, which will be good news to you I'm sure, find the cause and solution.

Apparently, TemporalSmoother() is not coded to properly handle different source and destination pitches, as well as to require a multiple of 4. This means it can crash after a Crop(). By moving it before the crop, as follows, things start working:

LoadPlugin("mpeg2dec.dll")
LoadPlugin("decomb.dll")
mpeg2source("G:\tmp\test.d2v")
Telecide()
Decimate(5)
TemporalSmoother(2,1)
crop(12,2,697,478)
BicubicResize(512,384,0,0.5)

This example shows how difficult it is to determine the culprit. Things are not always what they seem. :)