about fixoversharp

[morsa]

Could its source code be published somewhere?

[MrTibs]

If you are talking about FixVHSOversharp....

Well, the source is so bad that I'm embarrassed to publish it. I've been slowly learning C++ filter programming (I don't know either) and I think I could do a much better job with it now.

If someone is willing to clean the source before publishing it, I'll be happy to provide the source. I think Fix VHS should be GPL. (Under no conditions should the existing source be published.)

Edit: Personally, I think the algo is lousy, I was working on a better one but I stopped doing VHS conversions so the project stalled.

[morsa]

But it was really usefull for any kind of material.Think about those oversharped DVD and Remember bad configurated DV camcorders!!

[MrTibs]

Well, your request got me back working on the filter. Expect to see an improved version in the next week or so.

Suggestions and comments are welcome.

[SoonUDie]

Well, thanks for cranking out a new one ;)

[Boulder]

I think the time has come to wake this one from the dead:D

Are there any new developments? The old filter really is a killer when it comes to VHS processing.

[morsa]

Can't BlindDeHalo replace it?

[Boulder]

It appears to be YV12-only (I do YUY2 captures these days) and also much slower. I can currently run three instances of FixVHSOverSharp in my VHS-processing script (which is slow already due to truly interlaced material;) ) and I get ~0.32 real time speed with CCE. With BlindDeHalo the processing would become too slow, I'm afraid.

I was just wondering whether MrTibs ever finished the improved version he mentioned.

[Didée]

Boulder: Sure you're using BlinDeHalo3? This last version is pretty fast. The original (now obsolete) BlindDeHalo was very slow indeed ... we didn't have MaskTools, those days ;)

However, BHD can not really replace FixVHSOversharp. BDH does a fair job for what it's designed for. But still, it is working "blind", where FixVHSOversharp does a *real reckognition* of halos (looking for small 'overshooting' luma peaks in direct neighborhood of strong luma transitions).

More seriously, BDH works OK for artificially added Edge Enhancement. But it can't do much against these VHS-typical, dropshadow-alike halos present only on one side of edges.

[Boulder]

Oh hell, I should have guessed that the function's grown to be a big boy now;)

I guess I could give number three a go when I get rid of the bigger pile of tapes that are about to die.. I haven't run into any severe halos yet but I know there are some recordings which are at least second or third generation, and I fear them:D