another restore function

[MOmonster]

I just installed the newest avisynth and has no problems to run my script.
I think mg262 is right. You doesn´t set all the clips.
My function needs three different clips, the source clip the bobbed clip for the output and the bobbed clip for the analyse. This is my script:

LoadPlugin("C:\program_me\dgmpgdec\DGDecode.dll")
LoadPlugin("C:\program_me\filters\leakkerneldeint.dll")
LoadPlugin("C:\program_me\filters\TDeint.dll")

import("D:\work\C_package.avs")

MPEG2Source("D:\dvd.d2v")
#ConvertToYUY2(interlaced=true)

#crop(16,32,-16,-32)
ord = last.getparity() ? 1 : 0

a = last.leakkernelbob(order=ord, sharp=false, twoway=false, threshold=4)
b = a.crop(32,32,-32,-32).BilinearResize(640,352)

Crestore_film(bob=a, bclip=b, bthreshmin=48)

It is necessary to set the bob clip and the bclip. If you don´t set the clips direct, you can also do it that way.

ord = last.getparity() ? 1 : 0

a = last.tdeint(mode=1)
b = last.leakkernelbob(order=ord, sharp=false, twoway=false, threshold=4).crop(32,32,-32,-32).BilinearResize(640,352)

Crestore_film(last,a,b)

That a fast smart bobber for bclip and your favourite bobber for bob.The first clip Crestore use is the sourceclip (last). For optimal settings, please read the readme.

[Didée]

@ MOmonster

Save yourself from such hassle ...

Function Crestore_XYZ(clip clp, clip "bob", clip "bclip", [...]
{
warning1 = """clip variable "bob" was not specified"""
warning2 = """clip variable "bclip" was not specified"""
bob   = default( bob,  clp.bob().subtitle(warning1),y=24 )
bclip = default( bclip,clp.bob().subtitle(warning2),y=40 )
[...]

[mg262]

Add to that: as I understand it, putting quotes around an argument allows that argument to be omitted... if I'm correct the following would also work:

Function Crestore_XYZ(clip clp, clip bob, clip bclip,...)
{
[...]

[MOmonster]

@Didee
You´re right, maybe such warnings would be better to understand what is wrong. I´ll implement it in my next function, but I don´t know when (and if) I renew this version.

[MacAddict]

My sincere apologies, I apparently didn't have my bobber set for the bclip even though I read the readme twice. Thanks so much for the explinations and help.

[MOmonster]

Small bugfix (wrong handling with pattern smaller than 8 Frames), Crestore has now version 0.9e.
There are no changes, the development of Crestore has stopped for other projects.

[MOmonster]

Because I got the feeling that Crestore is still used, even though there was no development, yesterday I decided to implement the code of the final Cdeblend into Crestore to also finalize this function.

These are the changes:
- new modes -> hq mode is 20% slower than the old Crestore version but much better, mode<0 is still significantly better and 20% faster than the old one (with simple bobber: realtime)
- instead of Cdeint, Crestore_pal & Crestore_film you have just a rate parameter in Crestore(), so you can output also 24fps
- different easier handling and more tweakable
- mt_masktools needed to run the function (included into the zip)

All in all this version works significantly better than the last one.
You find the downloadlink in my first post.

[Backwoods]

Invalid Attachment specified. If you followed a valid link, please notify the administrator

When clicking the link you posted.

[MOmonster]

A login was needed, I don´t know why? But it should work now.

[Backwoods]

Works now, thanks.

[MOmonster]

Update:
A small bugfix, because a copy and paste mistake, but nothing you will realize during running the function.

[stg]

I can't make this work, perhaps I'm too much of a n00b

Everything looks fine until I start the encoding, then I get some unspecified error.

[Alain2]

No problem for me, works fine, thank you MOmonster (btw the name of the function is now Crestore, not Crestore_film as shown in the first post)

[stg]

Ok, now it works for me too, awesome.

Gotta check out those parameters more carefully, though. The TNG episode I encoded has even more stuttering than the original had.....

[MOmonster]

@Alain2
I don´t find the function name in my first post. But yes the last version has this name, because there was no rate parameter and I seperated the basic function into Crestore_film and Crestore_pal. But now there is only the function Crestore.

@stg
If you have pal material the stuttering shouldn´t be worse than before. Only in really low motion scenes the decimation don´t work so good because Crestore has only a simple patterndetection for the decimating. For anime and highmotion scenes everything runs fine for me, but on low motion scenes and sources with many edits the decimation works not so good.
If you have NTSC footage you should set the rate parameter to 25. Also this function is only tested with two small NTSC samples, because I live in a pal country. If you have a source for me Crestore don´t work good on maybe you can upload a small sample.

Thanks for the kind words.

[stg]

The material I was trying to encode was Star Trek TNG season 1 episodes. PAL/European release.

It seems those DVD's are not very good to begin with, and when they're encoded to XviD, they stutter a lot.


Ok, I uploaded this 30mb sample. This is not even the worst case of such stuttering, but if you look at it you will see what I mean. Especially at 35 seconds, the scene in which the captain's face is shot closely.

Sample: http://s37.yousendit.com/d.aspx?id=3...503ZDJVIFNRQZO

If someone has encoded these episodes succesfully, I would like to know how/what parameters/etc. was in use.

[foxyshadis]

TNG PAL is probably one of the most frustrating to restore, because it's converted hybrid material, and that's always bad news. You're going to have trouble because some scenes were originally 24 fps, some were 30p, and some were 30i, all now merged into 25i. It's impossible to deinterlace that while keeping a natural framerate, which is why the xvids look so choppy. (Unfortunately I have no good suggestions on handling it, my experience mostly ends at ntsc hybrid.)

[MOmonster]

You said your source is a dvd, but you send a avi. I need a unprocessed sample to get the problem. If it is really so as foxyshadis said of course Crestore can´t produce a fluid output. Also restore24 will be far from perfect (but will be more fluid than Crestore). The only thing will be unblend (Cdeblend won´t work with real interlaced parts, but can result in better fluency than Crestore) + hybrid decimating, but I don´t know what tdecimate is able to do in such cases.

[stg]

I must have had a brainfart or something, and posted .avi sample, hehe

Anyway, here's a sample from the DVD, packed with WinRAR:

(had to put it to 3 packs because of the new yousendit limitations, it's total of 145 mb. Considered making a smaller sample but with this you clearly can see the stuttering)


Part1:
http://s38.yousendit.com/d.aspx?id=0...Y1LV0Y0TR00F03

Part2:
http://s38.yousendit.com/d.aspx?id=1...2213ISRYDXMCKQ

Part3:
http://s38.yousendit.com/d.aspx?id=0...A1UOOIG5OQQ98G

[Alain2]

Originally Posted by MOmonster

@Alain2
I don´t find the function name in my first post. But yes the last version has this name, because there was no rate parameter and I seperated the basic function into Crestore_film and Crestore_pal. But now there is only the function Crestore.

Oups sorry, i was looking only at the top, but I am at page 4 of this thread ^^
Good function you created, works nice on a kenshin dvd I was given, better than r24rc1 (or maybe I don't know how to tweak it) ; your function eliminates the worst blends, only leaving small ones, but they are not too visible when watching at normal speed