digitalFAQ.com Forums [Archives] - Avisynth: Trying out LimitedSharpen()

Page 4 of 5

Show 40 post(s) from this thread on one page

digitalFAQ.com Forums [Archives] (http://www.digitalfaq.com/archives/)

- Avisynth Scripting (http://www.digitalfaq.com/archives/avisynth/)

- - Avisynth: Trying out LimitedSharpen() (http://www.digitalfaq.com/archives/avisynth/13762-avisynth-limitedsharpen.html)

Dialhot

12-28-2005 08:48 PM

Quote:

Originally Posted by Shibblet

Anybody else get these same results?

You are just experiencing all the reasons why I always say that a sharpener should never be used. And the reason why I do not use any !

BTW, LS is a lot tweakable and you will probably find a way to limit the bad effects (that is the purpose of the function).

gamma

12-29-2005 02:51 AM

Quote:

Originally Posted by Shibblet

I am having three problems with LimitedSharpen().

1. It does clear up the image, and add some missing detail, however it adds edging and pixels to places that weren't really there before. So a stationary object in the background results in some flickering around the edges because of the sharpening from frame to frame.

2. By adding these extra pixeled areas, you are actually increasing the size of your encode. Which results in having to lower your quality level in order to make it fit in the same place it was before.

3. Older movies (which are problematic and noisy to begin with) seem to get a lot of edging issues from frame to frame.

Anybody else get these same results?

Well, I think your settings are too "strong".
Try this:

Code:

Lremovedust(4,1)

limitedsharpen(ss_x=1, ss_y=1, smode=3, strength=50, soft=20)

That works like a charm for most of my encodes. Also, play with the smode setting, try smode=4 for example. You can also try to add "wide=true".

BTW: on a test I did a while ago, the following syntax gave me a better
subjective picture, and a equal filesize than a lancosresize.

Code:

limitedsharpen(ss_x=2.0,ss_y=2.0, lmode=2, wide=true, strength=50, dest_x=..., dest_y=...)

The higher the supersampling is, the slower the encode becomes, but you can increase the strength without getting more artifacts.

rds_correia

12-29-2005 03:35 AM

Nice tips gamma.
Thanks :D.
BTW I find that LRemoveDust(17,2) softens too much for my taste.
I've been trying other methods and 17,1 seems to be much better for clean sources.
Tonight I'll try LS() more deeply too.
This thread http://www.kvcd.net/forum/viewtopic.php?t=16032 has some very good comparing scripts that really help (along with some test encodes) to get a better idea on how this works.
Cheers

Boulder

12-29-2005 04:24 AM

Don't forget to use Soothe() when you use LimitedSharpen, it'll fix some of the issues related to LS. See the link I posted earlier in this thread.

gamma

12-29-2005 04:36 AM

Quote:

Originally Posted by rds_correia

Nice tips gamma.
Thanks :D.

You're welcome :D

Quote:

Originally Posted by rds_correia

BTW I find that LRemoveDust(17,2) softens too much for my taste.
I've been trying other methods and 17,1 seems to be much better for clean sources.

If I remember correctly, the first number (in this case 17) is the Removegrain-mode the function calls. (please, somebody correct me if i'm wrong). Looking at the documentation of removegrain, mode 2 or 5 are the least "destructive" for the details. So lremovedust(5,1) for example should give a good result also.

Boulder

12-29-2005 06:59 AM

Modes 1,2 and 5 are not as effective so I wouldn't recommend using them in LRD. Mode 17 also preserves thin lines better than those mentioned. Using a limit of 1 should be enough to keep the details but press down the noise and grain.

Prodater64

12-29-2005 07:41 AM

I would wish to know how about the crop/resize to transform not anamorphic to anaporphic with LS()?
That is not mentioned in any place here!!!
How must i to do it?
Or is it that the function will do it automatically?

Boulder

12-29-2005 08:43 AM

FitCD will handle the maths.

Prodater64

12-29-2005 12:56 PM

Quote:

Originally Posted by Boulder

FitCD will handle the maths.

You mean that i must to load the non anamorphic source, and set the target as anamporphic and that is all?

Dialhot

12-29-2005 01:14 PM

Exactly.

gamma

12-30-2005 04:56 AM

Just for pointers, a fast version of LS() is obtained this way:

read this: http://forum.doom9.org/showpost.php?...&postcount=339

Then download this:
http://people.pwf.cam.ac.uk/mg262/po...t_03Dec05B.dll
and this:
http://people.pwf.cam.ac.uk/mg262/po...penFaster.avsi
and this:
http://manao4.free.fr/masktools-v2.0a13.zip

Then load limitedsupport and masktools 2.0a as plugins, and import the limitedsharpenfaster routine. Call the function with:

Code:

LimitedsharpenFaster(.....)

When using lremovedust at the same time be aware that you'll still be needing the old version of masktools (v1.5.x). This is because the internal names for the filters have changed. Loading both versions at the same time works perfectly.

Using this faster LS() i went from CCE encode speed with the same clip from 0,28 to 0,81. Results are exactly the same as the original LS(), at least for my eyes. :D

supermule

02-01-2006 02:06 AM

the limited support dll is no longer available for download, can you upload it pls.

Also reading so many posts is confusing, can someone post a final fast version of limitedsharpen(in avsi format) with all required plugins and a limitedsharpen(parameter).

Boulder

02-01-2006 02:28 AM

Read what I answered you just a while ago :roll:

Dialhot

02-01-2006 03:33 AM

Quote:

Originally Posted by supermule

the limited support dll is no longer available for download, can you upload it pls.

I did not test yet but it is supposed to be included in last masktool so this dll is not needed anymore.

Racer99

02-01-2006 12:33 PM

If it isn't you can try just checking the root, there are couple of versions there.

http://people.pwf.cam.ac.uk/mg262/posts/

supermule

02-01-2006 11:36 PM

I found the dll and tested it also, doesnt suit my taste so I will keep it for later.

gamma

02-02-2006 03:12 AM

Quote:

Originally Posted by supermule

I found the dll and tested it also, doesnt suit my taste so I will keep it for later.

What do you mean? You'll won't be needing the dll anymore as it is packaged in the latest masktools.

supermule

02-02-2006 05:11 AM

the .dll is part of the tool download now.

Limitedsharpenfaster() sharpened enough to create artifacts and is still slower. So I will stick with the conventional sharpeners for now.

I am currently trying Rsharpen and denoisesharpen, they look promising.

Dialhot

02-02-2006 07:28 AM

Quote:

Originally Posted by supermule

I am currently trying Rsharpen and denoisesharpen, they look promising.

Never heard about them. Please open a toppic on them is there is something interresting to know.

Edit: oups. Just saw that you already did that.
http://www.kvcd.net/forum/viewtopic.php?p=123781#123781

Dialhot

02-16-2006 03:50 AM

To summurize a little :

LimitedSharpenFaster to replace LimitedSharpen is a lot faster !

With LS() I encoded in 0.53x for a half DVD resolution (352*480). For the same source and the same script but using LSF() I encode at 0.80x for a SVCD resolution (480*480).

You simply need RemoveGrain upper to version 0.9 (so 1.0 ios good) and Masktool above 2.0a20.

http://home.arcor.de/kassandro/Remov...emoveGrain.rar
http://manao4.free.fr/masktools-v2.0a27.zip

LFS() is delivered as an avsi file in the masktool zip.

Quote:

Originally Posted by supermule

Limitedsharpenfaster() sharpened enough to create artifacts

These artifacts should be removed by Soothe().
http://forum.doom9.org/showthread.php?t=99679

Note : I dit not have time to test Soothe().

Thanks to Boulder and the guys that did that :)

Page 4 of 5

Show 40 post(s) from this thread on one page