Lanczos resize when using HCenc??

[blizzery]

Hi,

I read somewhere that Bicubic resize filter shoud be used when output width is less than input, Lanczos when more.

In my case the output width is not less than input, and therefore I will use Lanczos.

I read the following in FulciLives guide about FitCD:
"...since you are usig TMPGEnc Plus you should checkmark the RGB-output box. If you were using CCE you would instead checkmark the YUY2-output box. "

My question is what boxes should I check in FitCD when using Lanczos resize filter, and when HCenc will be used for encoding??

[45tripp]

select lanczos from pulldown menu,
untick output boxes.
HC needs yv12 input. your source is likely already yv12, or outputed as such.
if not add:
coverttoyv12()
at the end of your script

gl

[jagabo]

I read somewhere that Bicubic resize filter shoud be used when output width is less than input, Lanczos when more.

That advice is backwards. You want to use Lanczos when downsizing to reduce moire artifacts, bicubic when upsizing because it's faster.

[theewizard]

test it aND BE YOUR OWN GUIDE

I have found Lanczos almost always gives me better results with less artifacting PERIOD.

and bi-cubic only comes close to lanczos when either hlaving the size or doubling the size. which is very very rare situation

[guns1inger]

FitCD only uses Lanczos, but avisynth supports several variations of Lancsoz, including Lancsoz 3 and 4. I always change my scripts to use Lancsoz4Resize after saving them from FitCD or writing my own.

[theewizard]

@guns1inger i didn't know that , i'm going to try those

[AlanHK]

FitCD only uses Lanczos

You can choose from Lanczos, Bilinear or Bicubic.

[guns1inger]

I meant in terms of Lanczos FitCD only uses one variation. I should have made the clearer, obviously.

[jagabo]

AVISynth's LanczosResize() and Lanczos3Resize() are the same thing. Lanczos4Resize() may be a little better but I haven't seen much difference.

[guns1inger]

It seems to be sharper without adding edge rings and artifacts.

[jagabo]

I thought I saw a little more ringing with Lanczos4Resize(). But I haven't tested it very much.

I just noticed that AVISynth's BilinearResize() and BicubicResize() filters are the equivalent of VirtualDub's Precise Bilinear and Precise Bicubic filters -- much better than the plain Bilinear and Bicubic filters.

[manono]

I just noticed that AVISynth's BilinearResize() and BicubicResize() filters are the equivalent of VirtualDub's Precise Bilinear and Precise Bicubic filters -- much better than the plain Bilinear and Bicubic filters.

Bilinear, yes. Bicubic, no. BicubicResize(xxx,yyy,0,0.75) is the equivalent to the VDub Precise Bicubic.

The default BicubicResize(xxx,yyy) is the same as BicubicResize(xxx,yyy,0.33,0.33) and isn't so good (in my opinion):

With b = 0 and c = 0.75 the filter is exactly the same as VirtualDub's "precise bicubic" and the results are identical except for the VirtualDub scaling problem mentioned above. The default is b = 1./3. and c = 1./3.,

http://avisynth.org/mediawiki/BicubicResize

Unless I misunderstood what you wrote (always possible).

[Alex_ander]

AVISynth's LanczosResize() and Lanczos3Resize() are the same thing.

In latest AviSinth versions Lanczos3Resize() even gives error message, LanczosResize() should be always used instead. Lanczos4Resize() and Spline36Resize() give extra sharpening and are useful for increasing a clean (artifact-free) image (otherwise they emphasise those artifacts and the simple LanczosResize is better).