Log in

View Full Version : Stand-alone software implementation of NNEDI3


luquinhas0021
24th April 2015, 17:23
Somebody propose yourself to implement the NNEDI3 algorithm in a stand alone software, that open MP4, MKV and AVI containers, and the best types of audio and video encoders? If possible, why not improve NNEDI3? lol
If possible, put into it the Spline 100.

Thank you.

vivan
24th April 2015, 18:16
Somebody propose yourself to implement the NNEDI3 algorithm in a stand alone software, that open MP4, MKV and AVI containers, and the best types of audio and video encoders?What's wrong with avisynth? It can easily do any of that.

If possible, why not improve NNEDI3?Improve? How exactly?

If possible, put into it the Spline 100.Wat.

luquinhas0021
24th April 2015, 18:28
I) Avisynth doesn't open any container neither any audio or video codec. Container, only AVI. Codec, none that has VFW interface.
II) Improve NNEDI in quality, sharpness, less or no artifacts.
Where can I see the NNEDI3 algorithm?

LoRd_MuldeR
24th April 2015, 19:45
I) Avisynth doesn't open any container neither any audio or video codec. Container, only AVI. Codec, none that has VFW interface.

Since when? :confused:

→ https://github.com/FFMS/ffms2

Where can I see the NNEDI3 algorithm?

Download NNEDI3 from the author's web-site, it has the sources included:
http://bengal.missouri.edu/~kes25c/

(And don't miss the important notice in "Source Readme.txt" file!)

foxyshadis
25th April 2015, 07:37
What is it with you and spline100? I don't think I've seen a post from you yet that hasn't mentioned it. Do you think it's a magic algorithm? It's not, it's just gathers more pixels into creating each spline, and those pixels don't necessarily have anything to do with the pixel being interpolated. The end result is usually more like adding an unsharp mask after, and a lot slower. In almost all cases, high-tap filters are good for other kinds of interpolation (think rotation, warping, and so on) but barely make a difference for large-scale image resizing.

Aside from that, I really, really don't understand why you want someone to re-implement avisynth just so you can have nnedi3 be "standalone". A combination of avs2pipemod, avisynth.dll, nnedi3, and your choice of input filters and command-line encoder is literally all that's needed for a portable nnedi3. You could do something similar with Vapoursynth, but that one has significantly more dependencies.

luquinhas0021
25th April 2015, 14:21
I don't think Spline 100 is a magical algorithm, but i like so much of it because it can give good results in so many cases, as good as NNEDI3 or even better, if i take the relationship computational cost x optimal result