Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > Capturing and Editing Video > Avisynth Usage

Closed Thread
 
Thread Tools Search this Thread Display Modes
Old 24th December 2010, 20:53   #121  |  Link
palwan
Registered User
 
Join Date: Nov 2010
Posts: 9
Yes, you were right Gavino :-) Now it works fine, thanks alot for pointing me out.
palwan is offline  
Old 28th December 2010, 13:52   #122  |  Link
szabi
Registered User
 
Join Date: Nov 2004
Posts: 221
Quote:
AVISource("C:\my_dv_vhs_video.avi,fourCC="cdvc").ConvertToYV12(interlaced=true).QTGMC( Preset="Slow", EdiMode="NNEDI3" ).SelectEven().FFT3DFilter (sigma=2, sharpen=0.4)
I am using this for deinterlacing and sharpening my video.
I tried preset slower but it was too slow that is why I changed preset slow, but the encoding speed looks like the same as slower.

Are there any way to speed up it?
Multithread support or anything else which can help to keep the quality with better speed.

bye
szabi
szabi is offline  
Old 28th December 2010, 14:26   #123  |  Link
yup
Registered User
 
Join Date: Feb 2003
Location: Russia, Moscow
Posts: 854
szabi!
I am using QTGMC 2.51 wit Set Avisunth 2.6 build for multi-threading and masktools 2.6. For quad core speed up 2.5-3 times comparing to one thread.
For speed up encoding, first make lossless AVI (I am prefer Ut codec) using avisynth script and after make encoding.
yup.
yup is offline  
Old 30th December 2010, 12:35   #124  |  Link
szabi
Registered User
 
Join Date: Nov 2004
Posts: 221
I have DV-AVI (720x576@25fps). Should i transform it to an other avi, before encoding?
I think an unnecessary encoding just decrease the quality.
That is the last avisynth version I know: Avisynth 2.58.
I have to check what masktool is.
It is the first time I hear about.
szabi is offline  
Old 30th December 2010, 13:19   #125  |  Link
Gavino
Avisynth language lover
 
Join Date: Dec 2007
Location: Spain
Posts: 3,384
Quote:
Originally Posted by szabi View Post
I have DV-AVI (720x576@25fps). Should i transform it to an other avi, before encoding?
I think an unnecessary encoding just decrease the quality.
That's why yup said use a lossless codec for the intermediate. The idea is that a 2-pass encoder (for your final result) can be sped up by using the lossless intermediate as input rather than processing the (slow) script twice.
__________________
GScript and GRunT - complex Avisynth scripting made easier
Gavino is offline  
Old 30th December 2010, 13:22   #126  |  Link
Mystery Keeper
Beyond Kawaii
 
Mystery Keeper's Avatar
 
Join Date: Feb 2008
Location: Russia
Posts: 712
Quote:
Originally Posted by szabi View Post
I have DV-AVI (720x576@25fps). Should i transform it to an other avi, before encoding?
I think an unnecessary encoding just decrease the quality.
That is the last avisynth version I know: Avisynth 2.58.
I have to check what masktool is.
It is the first time I hear about.
Your avi is encoded (not sure with what codec, but "avi" is just a container). When you process it with AviSynth, you get uncompressed frames, which you can either store as is or compress with encoder. If you save your video uncompressed, it takes a LOT of space to store and memory to play, so you better compress it. When it comes to using heavy scripts in AviSynth, it is oftenly good to save the output uncompressed or compressed with lossless codec, then compress it with the codec of your choice. That way AviSynth is not loading your CPU at actual compressing and doesn't run twice during 2 pass encoding.
__________________
...desu!
Mystery Keeper is offline  
Old 30th December 2010, 15:49   #127  |  Link
szabi
Registered User
 
Join Date: Nov 2004
Posts: 221
As I wrote my avi is 720x576@25fps (FourCC:dsvd[sony digital video]).
But it is hard to work with it, so in script you see I change it to "cdvc"

Do i have to transfer it to uncompressed format or any kind of script modification is enough?

szabi
szabi is offline  
Old 30th December 2010, 23:12   #128  |  Link
ajp_anton
Registered User
 
ajp_anton's Avatar
 
Join Date: Aug 2006
Location: Stockholm/Helsinki
Posts: 782
I'm having trouble with a small timer that changes every frame. Instead of just upscaling the changing digit (MCBob does this), two frames are blended together.
Right now I'm using tr0=1,tr2=0,rep2=1 to improve this, but it reduces the quality of the rest of the video. What else can be done?
ajp_anton is offline  
Old 31st December 2010, 00:56   #129  |  Link
-Vit-
Registered User
 
Join Date: Jul 2010
Posts: 448
Quote:
Originally Posted by ajp_anton View Post
I'm having trouble with a small timer that changes every frame. Instead of just upscaling the changing digit (MCBob does this), two frames are blended together.
Right now I'm using tr0=1,tr2=0,rep2=1 to improve this, but it reduces the quality of the rest of the video. What else can be done?
You could set tr1=1 also to reduce the first temporal radius used for the output.

You could reduce blocksize so it is less likely to blend the digit as it finds it a poor block match. Try Blocksize=8. Too small a block size can cause poor motion matches (need enough detail in a block to make a good match)

A similar idea would be to reduce the SADs used for the temporal passes. However, I removed those as arguments in this version (can't remember why - probably cutting down argument count). You can find them in the script - search for _thSAD1 (& 2). The first has the stronger effect on the output. It defaults to 10*8*8, try reducing it to 5*8*8 or similar. This will make it less willing to blend blocks if they match poorly. It may reduce the effectiveness of the deshimmering though. If you don't like editing the script for a single vid, then use TempGaussMC_beta2, which has these values as arguments.

I may reinstate the SAD values as arguments next version.

Last edited by -Vit-; 31st December 2010 at 00:59.
-Vit- is offline  
Old 5th January 2011, 21:51   #130  |  Link
kolak
Registered User
 
Join Date: Nov 2004
Location: UK
Posts: 2,412
Is there a way to use QTGMC with YUY2 in lossless mode?
When I tried it says not supported- converting to YV12 solves problem, but I would like to keep source as YUY2.

Thx
kolak is offline  
Old 5th January 2011, 21:57   #131  |  Link
Didée
Registered User
 
Join Date: Apr 2002
Location: Germany
Posts: 5,394
interlaced_yuy2_video

ConvertToYV12(interlaced=true).QTGMC().ConvertToYUY2().MergeChroma(last.Yadif(mode=1))



It would be possible to make a "proper" YUY2 processing mode (it has been done before for TGMC). It only requires a bazillion of ConvertInterleavedToPlanar()/ConvertPlanarToInterleaved() Back-and-Forth-conversions in the script.
__________________
- We´re at the beginning of the end of mankind´s childhood -

My little flickr gallery. (Yes indeed, I do have hobbies other than digital video!)

Last edited by Didée; 5th January 2011 at 22:03.
Didée is offline  
Old 6th January 2011, 00:02   #132  |  Link
kolak
Registered User
 
Join Date: Nov 2004
Location: UK
Posts: 2,412
Quote:
Originally Posted by Didée View Post
interlaced_yuy2_video

ConvertToYV12(interlaced=true).QTGMC().ConvertToYUY2().MergeChroma(last.Yadif(mode=1))



It would be possible to make a "proper" YUY2 processing mode (it has been done before for TGMC). It only requires a bazillion of ConvertInterleavedToPlanar()/ConvertPlanarToInterleaved() Back-and-Forth-conversions in the script.
What is the quality loss?
What Yadif is doing there?

Thx
kolak is offline  
Old 6th January 2011, 00:50   #133  |  Link
Didée
Registered User
 
Join Date: Apr 2002
Location: Germany
Posts: 5,394
As they are, Q/TGMC can not process YUY2 video. However, the most obvious benefits of Q/TGMC is what they achieve on the luma plain. Therefore, that scriptline suggests to use Q/TGMC for luma (which is identical for YUY2 & YV12 anyways), and one of the "simple" deinterlacers for chroma.
__________________
- We´re at the beginning of the end of mankind´s childhood -

My little flickr gallery. (Yes indeed, I do have hobbies other than digital video!)
Didée is offline  
Old 6th January 2011, 00:58   #134  |  Link
kolak
Registered User
 
Join Date: Nov 2004
Location: UK
Posts: 2,412
Sounds good and I assume processing speed will increase a bit also.


Andrew
kolak is offline  
Old 6th January 2011, 01:09   #135  |  Link
Didée
Registered User
 
Join Date: Apr 2002
Location: Germany
Posts: 5,394
No. It will be a little slower. Q/TGMC is still doing "all" of the work, including chroma. A 2nd deinterlacer is used additionally, and its result replaces that of Q/TGMC (on chroma channels). More overall work => slower processing.

Such a "cheat mode" definetly could be integrated in the function, and then it would be possible to avoid a major amount of the overhead. (Not all, but most).
__________________
- We´re at the beginning of the end of mankind´s childhood -

My little flickr gallery. (Yes indeed, I do have hobbies other than digital video!)
Didée is offline  
Old 17th January 2011, 23:29   #136  |  Link
-Vit-
Registered User
 
Join Date: Jul 2010
Posts: 448
Thanks for the YUY2 version. It's good you didn't add it to the lossless code, because that part has been almost completely removed in the next version - it's been superseded by something quite new. I'll merge the changes into the version after next, when I'll also be looking at chroma optimizations.
-Vit- is offline  
Old 17th January 2011, 23:39   #137  |  Link
kolak
Registered User
 
Join Date: Nov 2004
Location: UK
Posts: 2,412
Thanks- I was one of them, but interested in lossless part- hehehe


Andrew
kolak is offline  
Old 18th January 2011, 07:43   #138  |  Link
yup
Registered User
 
Join Date: Feb 2003
Location: Russia, Moscow
Posts: 854
Stephen R. Savage!
My source is analog capture in YUY2 colorspace. Now we can forget about wrong conversion for YUY2 to YV12 for interlaced source.
You use 2.58 Avisynth ot MT 2.57 build? Set 2.6 do not support planar video.
yup.
yup is offline  
Old 20th January 2011, 14:35   #139  |  Link
pokazene_maslo
Registered User
 
Join Date: Apr 2009
Location: Martin, Slovakia
Posts: 79
Hello.
I have 3 questions about denoising.
  1. fft3dfilter can be used to denoise interlaced and progressive video. If I use QTGMC like this
    Code:
    QTGMC( Preset="Very Slow", Sigma=2.2 , NoiseBypass=1, NoiseRemove=1.0, NoiseRestore=0.0 )
    fft3dfilter is used before deinterlacing. However suggested order of filters is "deinterlace->crop->resize->denoise".
    My question is which provides better quality? Denoise first and then deinterlace or deinterlace first and then denoise?
  2. After answering question number 1, how to make fft3dfilter motion compensated? I'm using 64 bit avisynth and don't know how to do that.
  3. Is it OK to replace fft3dfilter with fft3dgpu?
Thanks in advance!
pokazene_maslo is offline  
Old 20th January 2011, 17:24   #140  |  Link
2Bdecided
Registered User
 
Join Date: Dec 2002
Location: Yorkshire, UK
Posts: 1,673
I just wanted to post to say that I love QTGMC. The overall speed increase on a full encode seems to be much greater than it appears to be during the first few frames. I don't know if that's right/wrong, but the output is fine so I'm very pleased.

Cheers,
David.
2Bdecided is offline  
Closed Thread

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 10:04.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2019, vBulletin Solutions Inc.