digitalFAQ.com Forums [Archives] - Avisynth: Lanczos Vs Spline resizers

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

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

- - Avisynth: Lanczos Vs Spline resizers (http://www.digitalfaq.com/archives/avisynth/14515-avisynth-lanczos-vs.html)

supermule

04-23-2007 06:51 AM

Lanczos Vs Spline resizers

I have seen mostly lanczos being used for most of the resizes and in some scripts also like LSF etc.

Whereas I also found that lanczos creates more artifacts than spline and then we have to use cleaning filters to remove them.

Although spline also tends to oversharp sometimes but atleast it doesnt add artifacts.

That was me, but whats your opinion about these resizers and what do you prefer when :

- Upscaling
- Downscaling
- Low bitrate encodes

and why ???

Dialhot

04-23-2007 07:01 AM

Re: Lanczos Vs Spline resizers

Quote:

Originally Posted by supermule

Whereas I also found that lanczos creates more artifacts than spline and then we have to use cleaning filters to remove them.

I don't think anyone is using cleaning filters to remove artifacts due to resizers. Or you are talking about filters I don't know ?

Quote:

- Upscaling
- Downscaling
- Low bitrate encodes

I need to do upscaling to send picture from my PC to my HD rear-projector (display in 1920*1280). For that I use spline. But for my KDVD encode, that I always do in downscale, I'm still using Lanczos. I remember flicking lines sometimes in my tests with spline.
For low bitrate I prefer to remove LS() but keep lanczos. Actually I never used other thing but maybe LS+bicubic can be also a good combo.

supermule

04-23-2007 11:51 PM

Sorry, I think I was not clear enough. I mean we use filters to remove those DCT effects and for the ringing caused by Lanczos.

In the all three Matrix on one DVD also, lanczos was used.

but on low bitrates I agree with you that Bicubic gives a good result. I use it with c=0.44 parameter to avoid the artifacts and retain some sharpness.

As far as LS is concerned, I read the script, it uses lanczos internally for the resize. So i really wonder if lanczos is actually the preferred filter to use for all types of encodes.

supermule

04-24-2007 06:51 AM

Found this on the avisynth mediawiki

Code:

LanczosResize is an alternative to BicubicResize with high values of c about 0.6 ... 0.75 which produces quite strong sharpening. It usually offers better quality (fewer artifacts) and a sharp image. 



Lanczos was created for AviSynth because it retained so much detail, more so even than BicubicResize(x,y,0,0.75). As you might know, the more detail a frame has, the more difficult it is to compress it. This means that Lanczos is NOT suited for low bitrate video, the various Bicubic flavours are much better for this. If however you have enough bitrate then using Lanczos will give you a better picture, but in general I do not recommend using it for 1 CD rips because the bitrate is usually too low (there are exceptions of course).

Crux : Lanczos is NOT suited for low bitrate video

Also interesting comparison between some resizers is given here.

Dialhot

04-24-2007 07:16 AM

Quote:

Originally Posted by supermule

Crux : Lanczos is NOT suited for low bitrate video

Com'on ! That's not a scoop !

They are just saying that the sharper the picture, the worse is the compressibility. Of course Lanczos if worse than bicubic, the same way than bicubic 0.75 is worse than bicubic 0.33 for the exact same reason !
I don't see what is so surprizing for you in these 8 words.

Let me tell you an other scoop : Spline is NOT suited for low bitrate video (as spline is as sharp as lanczos is...).

supermule

04-24-2007 10:35 AM

nothing scoppy, just posting some information I found :)

Dialhot

04-24-2007 12:29 PM

Quote:

Originally Posted by supermule

nothing scoppy, just posting some information I found :)

BTW the comparison article is interresting.

supermule

04-25-2007 01:04 AM

although lanczos was not part of the tests, but amongst the Bilinear, Bicubic, Spline16 and Spline36 resizers known to us, spline36 emerged as the winner closely followed by spline16.

I wonder what Sinc resizer is, since it was the best amongst them.

Dialhot

04-25-2007 05:23 AM

Actually the test is not about resizer but about interpolating methods when doing positional operation (here rotation) on a picture. May be sync is not usable for developping a resizer.

supermule

04-25-2007 11:35 PM

Code:

Manipulation of digital images often requires resampling of the pixel data. Usually, the pixel positions in the transformed image do not coincide with grid positions. Therefore, an interpolation algorithm has to be applied to somehow derive the pixel data for these in-between positions. The effort spent for this interpolation is crucial for the image quality of the result image.

Code:

 Several authors have judged interpolator quality by testing enlargements. However, there are quite a few problems associated with that, the most important being that there is no mathematically unique solution for digital image enlargements, and, hence, no really objective test. A more practical test is a rotation of an image by some arbitrary angle, which can be easily tested against the original image. The more such rotations the image survives intact, the better is the algorithm. This test is also readily available, since almost any Graphics-program has a rotate function.

They do talk about image manipulations and enlargements.

Dialhot

04-26-2007 03:26 AM

Read again and you see that they say exactly what I'm saying : by opposition to other tests, in this article they decided to not speak about interpolator in resizer but in positionnal transfo.

Sync does not exists as resizer, that is why I suppose (only that) that it is not suitable to developp a resizer. Or it's just because noone tried to :).