Is there going to be a new Manual for the newest YATTA version?

[Chainmax]

Question says it all. The one posted on YATTA's homepage is severely outdated.

[Chainmax]

From the changelog:

5-71
Fixed ymc telecide settings
Can convert dg projects to plain ones by using user specified offset (not tested)

5-70
Swiched back to mpeg2dec3 because of a bug in dg
I guess that means that you can use DVD2AVIdg in order to make the .d2v but you have to save it as a 1.76 file. I'd also think it means that even though you can use mpeg2dec3dg in avisynth, you have to use plain mpeg2dec3 as a .d2v loader. Am I right?

The latest version has two executables: ymc.exe and yatta.exe. The former seems to be a standalone version of the first screen that popped up in previous YATTA versions (telecide config, new project, etc), only with different options. The latter seems to be a standalone version of the YATTA project config.

[Zarxrax]

My guess would be that you need to use both regular DVD2AVI and MPEG2DEC3.DLL for everything.

Anyone know what bug they are talking about though? I noticed this in the changelog yesterday too, and it's made me curious, as I havent heard of any problems anywhere else.

[Myrsloik]

First the obvious:
Plain dvd2avi and plain mpeg2dec3 should be used. It also seeks a lot faster.

The less obvious:
The possible bug in dg almost never appears and seems to return different frames depending on if you seek or play to the same frame number. You'd probably only notice this if you used overrides in telecide since it then produces bad matches.

Disclaimer:
This theory has only been verified by one other person using one m2v. This post also doesn't come with any real evidence since that m2v happens to be 200MB after cutting and is in no way inteded to scare people away from mpeg2dec3dg.

[Chainmax]

So, I have to copy the mpeg2dec3.dll from YATTA's website and DVD2AVI 1.76 into the YATTA folder but I can use DVD2AVIdg for creating the .d2v and mpeg2dec3dg.dll for loading it into avisynth, right? Sorry if I sound n00bish, I'm not quite clear on that point.

As for the first question, how does YATTA operate now?

[Zarxrax]

No, I'm pretty sure you would have to use the same versions for EVERYTHING. Because they count different numbers of frames, if you tried to use a different one then your overrides would probably be totally messed up.

[Chainmax]

It doesn't matter if DVD2AVIdg can save to 1.76 format then?

And again, how does YATTA operate now?

[Chainmax]

:confused:

[Chainmax]

gizmotech? Myrsloik? Anyone?

[Mentar]

Fairly simple. You put mpeg2dec3.dll into the YATTA directory and use the original dvd2avi 1.76 to create your .d2v file. Then, you use ymc (yatta metrics collector) to create the .yap project file, and start your work by opening this .yap file in yatta.

The rest is the same as before.

Myrsloik will kill me (with good reason), since I'm about 4 months late with writing a good step-by-step introduction to yatta. The tool has become tremendously powerful, but I guess also very intimidating to use.

I hope I'll be able to write at least the first chapter today.

[Chainmax]

Thanks, now I can finally start ripping my Simpsons DVDs :). I hope you can someday start working on the guide.

[Chainmax]

IIRC, there are a couple of versions of DVD2AVI 1.76. The one I have was made by LOLI.J, is that the one I should use?

[Chainmax]

So I finally started to use YATTA and I already have a problem:

I am using a vtresh of 27. After almost an hour of using the up arrow and switching between matches and realizing I hand't gone much farther than the title screen I checked what value was on the "V Search" section...it was 30. I assume that it doesn't adjust itself automatically, so it was my fault for not noticing. No problem. However, since it was taking so damn long I decided to change the V Search valure to 50. After all, the frames with 28-50 vthresh will get deinterlaced anyway.
The problem is that it still is taking too damn long. I've been at it for four days (~45 minutes per day) and I'm still not even at half the episode's length. Do you guys think I'm doing something wrong or maybe the DVD is just immensely fucked up?

[Zarxrax]

Wow... I go through episodes in about an hour... just scrolling through every frame... i dont even use the vsearch untill after i went through it all manually. If you are getting a lot of bad field matches or something maybe you have the field order set wrong or something?

[Mentar]

What source are you working on?

I tend to configure the v search to "25 absolute difference to both neighboring frames", this tends to find most fixable bad field matches very quickly and reliably (use cursor up/down for quick skipping with the right hand and leave the left hand on "s"). For a normal anime episode of 25 mins, this usually takes 5 minutes and you're through for the first time.

Then I tend to make a constant quality 2.0 encoding with show=true enabled in telecide. Watch it, and write down the frame names of the remaining combs (usually only mouths), then fix them in yatta, remove the show=true and make the real encode.

If you've got to invest THIS much time, you either have misconfigured decomb (wrong order, for example), or you've got a DVD with blended frames. So I ask again - what source are you working on?

[Chainmax]

I am trying to rip "Simpsons Complete 1st Season". It's a Region 4 NTSC box set. V Search was set to "V". Maybe I was not paying attention, but VMetric values seemed to be much higher on the YATTA config screen than on the Telecide metrics gathering screen :confused:
Is the version of DVD2AVI 1.76 I have (made by LOLI.J) the correct one?
Mentar, your approach is interesting. I'm going to try that, but using a lower percentage (15% or 20%) and use postprocessing instead of the review quant2 encode.
Is it possible to use YATTA and then KernelDeint as a postprocessor (KD needs post=1 and hints=true)?

[Chainmax]

Forget about the KD question, I decided to quit dicking around and use post=2.
I have one last question though. I am using guide=0, but I'm not 100% sure if that's the correct setting. While the source does jump wildly between "in pattern" and "out of pattern" in some spots, they are relatively few and far between. Should I set guide=1 in this case? After all, whatever pattern mismatches that resulted from that will be detected by YATTA, right?

[Chainmax]

Well, I finished using YATTA but another problem has popped up. I clicked on the "save" button numerous times, but I never was presented with a "save as" window or anything like that. I tried quitting and answering "yes" to the saving question and still got nothing. Am I missing something here? All I have now is the .yap file.

[Zarxrax]

If you right-click you get options to save the override files. I was confused with this for a while too.

[DarkNite]

Is it possible to use YATTA and then KernelDeint as a postprocessor (KD needs post=1 and hints=true)?

Whatever for? It will only process what Telecide returns as interlaced, which is probably wrong most of the time, or only correct in the most extreme of circumstances.

I was under the impression that YATTA was for when you don't (and shouldn't) trust the automatic decisions of a filter.

Anyways, there is a sections function. I would set up sections for wherever you have trouble finding any correct matches, maybe insert TMCPost (and relevent settings) for those sections, or possibly (hint: this is some good stuff) look into figuring out VFR and using matroska as your finalized format.

Since we're talking about The Simpsons here I might be half tempted to just use TomsMoComp and decimate later, if at all... I know that's "evil", but so is that source. I know, I have 3 seasons worth of those boxsets. :)

Best of luck.

[Chainmax]

Originally posted by DarkNite:
quote:
--------------------------------------------------------------------------------
Is it possible to use YATTA and then KernelDeint as a postprocessor (KD needs post=1 and hints=true)?
--------------------------------------------------------------------------------



Whatever for? It will only process what Telecide returns as interlaced, which is probably wrong most of the time, or only correct in the most extreme of circumstances.

I was under the impression that YATTA was for when you don't (and shouldn't) trust the automatic decisions of a filter.
From everything I read, YATTA is for exactly that, but I wanted to try KD as a postprocessor (since I was enabling postprocessing in yYATTA anyway) using something like:

Telecide(order=0,guide=1,post=1,back=0,hints=true,ovr="C:\MPEG-4\Simp\Episodio1\EspNav.txt")
KernelDeint(order=0,sharp=true)

I don't know if KD can avoid blending at scenechanges like Telecide's post=2 does, though. Therefore, as posted before, I decided to quit dicking around.
What is TMCPost? Also, why VFR should be used? This disc seems to abide well to a common 3:2 pattern.
As BoNz1 suggested me in a PM following this thread (http://forum.doom9.org/showthread.php?s=&threadid=67586) (in which you suggested me to try YATTA), I'm only fixing bad matches. After that I'm thinking to use mipsmooth and maybe lumafilter to reduce remaining artifacts.

[Myrsloik]

For the current version (Decomb 5.10) you have to use post=2 for the internal postprocessing and post=0 if you want to use something like kerneldeint. Post=1 will just waste time since the override file sets postprocessing on or off for every frame and will give results identical to post=0.

Some dvds just aren't vfr and then you don't use it and sometimes kerneldeint looks bad and then you don't use it either. It always blends. I assume TMCPost is the ultra rare version of tomsmocomp that takes hints from telecide.

[Chainmax]

From the KernelDeint readme:
To use this filter as a post-processor for Telecide(), use the following script:

Telecide(...,post=1,hints=true)
KernelDeint(...)
Anyway, you're saying it blends at scenechanges? I will definitely not use it then. These DVDs have some pretty nasty blendings on scenechanges. I fixed most of them with YATTA, but some of them just couldn't be avoided with different matches.

[DarkNite]

I fixed most of them with YATTA, but some of them just couldn't be avoided with different matches.

If it's not high motion, and it's not more than a frame in a sequence that has no match then the D key is your best freind.

Just don't forget to load the .dec.txt into your decimate line if you do go that route. Not that I've done that or anything... :o

[Chainmax]

What does the D key do? About not loading the stuff...I didn't RTFM and thought that just hitting the "save" button was enough to get the override files and quit the project. I am redoing an episode for that. :o++

[gizmotech]

Alternative to the d key, which overrides the automatic decimation strategy, using the freeze frame is usually more effective.

There really isn't any reason to use forced decimation's with the latest yatta, unless correct a bad decimation stretch. The reason I say this is that at any time in the source where you would need for decimation you could have easily just freezed framed the video instead, otherwise the frame was part of a pan/motion set which would requite post processing (and not to be decimated/frozen).

At any rate.

@chainmax.

Please save the dev and forum members a bit of trouble and a) read up on what Telecide does. B) Read up on what Decimate does c) read up on what a deinterlacer does ... and once you've done all that you'll have a much better starting ground to using yatta given decomb is a core tool to yatta operations, and the rest is just basic knowledge that is handy prior to using hte application.

Gizmo.

[Chainmax]

Well, I had almost figured out that trying KD was stupid before I got any answers. I know I'm a bit slow, but at least I corrected myself without help.
You sound offended or annoyed, if that's the case then I apologize.

[DarkNite]

Using a postprocessor isn't a completely bad idea. It's very useful for some hybrid material.

You'd just specify which frames need to be postprocessed in YATTA and then set Telecide to post=0 and your chosen postprocessor afterwards will pick up the hints only from the frames marked as combed in your ovr (lines ending with a + that are listed after the field matching info).