is this useable for Avisynth?

[Terka]

http://auricle.dyndns.org/ALE/
http://www.soe.ucsc.edu/~milanfar/software/superresolution.html
http://www.ece.lsu.edu/ipl/Demos.html

[Terka]

deinterlace:
http://image.diku.dk/sunebio/KellerLauzeNielsen.pdf

[Terka]

http://ericpbennett.com/VideoEnhancement/BennettMcMillanSIGGRAPH.pdf

[Didée]

Nice finding.:)
Note on the last one: It's really not very hard to implement something quite similar with a script.

The following should give the idea (example is actually not-working, it's just to show a possible general flow)

dark = last

bright = dark .levels(...) # or Tweak(), ColorYUV(), HDRAGC() ... or whatever suits you

diff = mt_makediff( dark, bright, U=3,V=3 )

bvec1 = dark .MVAnalyse( isb=true, idx=1 )
fvec1 = dark .MVAnalyse( isb=false,idx=1 )

diff_temp = diff1.MVDegrain1( bvec,fvec, idx=2 ) # or use MVDegrain2/3 with more vectors

diff_spat = diff1. MinBlur(2) # or use tbilateral; MinBlur is the fast+lazy shortcut

err_bw = dark.MVMask( bvec1, kind=1, ml=50 ) # tweak ml, perhaps also gamma
err_fw = dark.MVMask( fvec1, kind=1, ml=50 ) # (kind=1 -> motion-compensated , kind=0 -> motion adaptive)

err = Merge( err_bw, err_fw, 0.5 )

diff_spatemp = diff_temp .mt_Merge(diff_spat, err, U=3,V=3)

result = dark .mt_makediff( diff_spatemp, U=3,V=3)

That's pretty basic, and of course one can stuff in lots more things at free will.

After a quick glance through the paper, there are a few points ... they don't use mo-comp at all, the method is 'only' motion-adaptive. Temporal-Bilateral and Volumetric-Bilateral ... well, those we don't have in Avisynth. At least Temporal-Bilateral could be self-cooked with ttempsmooth, via pfclip and according masks.

Reminds me, one feature that's missing in MVTools: create vectors on clip A, then exchange compensation with clip B (MVChangeCompensate) and re-derivate SADs from clip B accordingly. Like "MVChangeSAD(vectors, newclip)".
Occasionally, this might become handy ... e.g. for stuff like this here. ;)

[Terka]

http://www.imse.cnm.es/Xfuzzy/xfpapers.html
http://www.imse.cnm.es/online/2005/ANNIE2005.JGR.pdf

[Terka]

http://www.escet.urjc.es/~asanz/investigacion_en.html

[Terka]

http://www.hindawi.com/GetArticle.aspx?doi=10.1155/2009/381673

[Terka]

http://www.nashruddin.com/phase-correlation-function-in-opencv.html#results
http://www.wedesoft.demon.co.uk/hornetseye-api/files/phasecorrelation-txt.html
http://www.google.cz/url?sa=t&source=web&ct=res&cd=1&url=http%3A%2F%2Fscien.stanford.edu%2Fclass%2Fee392j%2Fprojects%2Fprojects%2Fliang_report.pdf&ei=G0t5SrrzJtig_AbGkvWMBg&usg=AFQjCNH0G8ayBnhNW0y_-_7gWYkXXJjSyA&sig2=xatg3KuIlEHfeyoXz8PSOA
there are many articles about phase correlation, could this be improvement when implemented to mvtools? Depan already uses phase-shift.

[Terka]

http://www.faculty.idc.ac.il/toky/Publications/Journal/superRes.pdf
http://www.utia.cas.cz/files/Soutez_08/Aplikace/Sroubek/gui_help_v12.pdf
the matlab application is free for non commercial application

[Terka]

http://www.cs.technion.ac.il/~elad/publications/journals/2008/SuperResPaper_IEEE_TIP.pdf