MCTemporalDenoise [v1.4.20 - Update 2010/07/02]

[LaTo]

Originally Posted by darkelf5

I've tried it with AviSynth x64 build.
Everything works, except enabling GPU. I've tested it on a simple script

AVIsource("blablabla")
MCTemporalDenoise("GPU=true")

And this is what i get:

It should be a problem with FFT3Dgpu.

[lych_necross]

Is there a chance that this script will become a dll like with SmoothAdjust? If so, what about LSFmod?

[LaTo]

Originally Posted by lych_necross

Is there a chance that this script will become a dll like with SmoothAdjust?

No, too complicated for me.

Originally Posted by lych_necross

If so, what about LSFmod?

Maybe later, but at this time I don't see any big improvement to achieve this and I'm lazy

[asarian]

Earlier I had some trouble with TemporalDegrain, as part of a scene would bleed into the next on certain scene changes. So I looking for a temporal denoiser that will denoise without these faults, and I'm hoping MCTemporalDenoise is it. :) So, is this program 'safe' for general use? Kinda hate spending enother hundred or so hours on encoding, only to find out the output is messed up again.

I got the impression, from the other thread, that temporal degraining is inherently hard to accomplish; so maybe a truly 'safe' filter in that regard does not exist. But I'm hoping people's experiences with MCTemporalDenoise are good.

Thanks.

[LaTo]

Originally Posted by asarian

Earlier I had some trouble with TemporalDegrain, as part of a scene would bleed into the next on certain scene changes. So I looking for a temporal denoiser that will denoise without these faults, and I'm hoping MCTemporalDenoise is it. So, is this program 'safe' for general use? Kinda hate spending enother hundred or so hours on encoding, only to find out the output is messed up again.

I got the impression, from the other thread, that temporal degraining is inherently hard to accomplish; so maybe a truly 'safe' filter in that regard does not exist. But I'm hoping people's experiences with MCTemporalDenoise are good.

Thanks.

Yes, it's safe for general use... Especially with limit=-1 & limit2=-1

[asarian]

Originally Posted by LaTo

Yes, it's safe for general use... Especially with limit=-1 & limit2=-1 ;)

Thank you very much! :) I'll go install it rightaway!

[LaTo]

Originally Posted by asarian

Thank you very much! I'll go install it rightaway!

Just one thing: adjust sigma for your source, it should be the lowest while keeping it's efficiency.

[Didée]

Originally Posted by LaTo

Just one thing: adjust sigma for your source, it should be the lowest while keeping it's efficiency.

That was exactly his problem with TemporalDegrain: usage of default sigma's, which was too strong for his source...

The underlying problem with using FFT3DFilter as prefilter is that FFT3D is generally not safe on scenechanges. Is there any special SC-protection in MCTD to compensate for that?

[LaTo]

Originally Posted by Didée

The underlying problem with using FFT3DFilter as prefilter is that FFT3D is generally not safe on scenechanges. Is there any special SC-protection in MCTD to compensate for that?

No SC-protection in MCTD... But I have never experienced SC problem with MCTD, it may be due to reasonable sigma values in preset.

If FFT3D really miss a scene change, try to use limit=x / limit2=x (instead of -1).
"x" should be a positive value but the lowest possible (it's a simple limiting threshold).
If the clip isn't so grainy, bt=1 should fix the problem also.

Maybe I will implement a SC-protection in MCTD with SCSelect if the problem persist, but I doubt.
Truly I use FFT3D all the time without problem, so I want to see a sample/script (with appropriate sigma) where FFT3D miss a scene change... Someone please? I am curious

[Didée]

Originally Posted by LaTo

where FFT3D miss a scene change...

It doesn't miss "a" scenechange. It does miss *every* scenechange. To my knowledge, temporal processing of FFT3DFilter simply doesn't check or care for that.

Just set bt=5,sigma=16, and watch what happens.

Of course it's mostly not a problem with lower sigmas. But that doesn't mean the issue isn't there. It's just that the problem becomes much harder to notice (and that's why people aren't complaining) ... but *present* it is all the time.

[LaTo]

Originally Posted by Didée

It doesn't miss "a" scenechange. It does miss *every* scenechange. To my knowledge, temporal processing of FFT3DFilter simply doesn't check or care for that.

Just set bt=5,sigma=16, and watch what happens.

Of course it's mostly not a problem with lower sigmas. But that doesn't mean the issue isn't there. It's just that the problem becomes much harder to notice (and that's why people aren't complaining) ... but *present* it is all the time.

Argh you're right... Hardly noticeable but it's here with high sigma in FFT3Dfilter

[senseiam]

Originally Posted by senseiam

Don't know if anyone else has seen this or not but I am getting "This CLIP is NOT MOD4" errors with mod4 resolutions. When I crop to mod16 script works, no problem.

I tried 1920x1080 (mod4) and I get "THIS CLIP IS NOT MOD4" error

I add Crop(0,4,-0,-4) for 1920x1072 (mod16) works like a charm.

Edit: Using latest version of script

Originally Posted by LaTo

No problem with 1920x1080 in my side, provide the exact script used and a little sample clip please.

Don't have a specific sample clip handy right now but I get this on multiple different sources so I have to crop after or crop to mod16.

Script example:

Import("\AviSynth 2.5\plugins\MCTemporalDenoise.avs")
Import("\AviSynth 2.5\plugins\GradFun2DBmod.v1.5.avs")
SetMemoryMax(2048)

AVCSource("something.dga")

Crop(0,20,-0,-20)

MT(""" MCTemporalDenoise(settings="very low",sigma=6,chroma=true,stabilize=true,GPU=false,ncpu=8,enhance=true,sharp=false) """, threads=8, overlap=16)

Even though it's 1920x1040, mod16, it doesn't want to work, reports this isn't mod4.

[LaTo]

Originally Posted by senseiam

Don't have a specific sample clip handy right now but I get this on multiple different sources so I have to crop after or crop to mod16.

Script example:

Import("\AviSynth 2.5\plugins\MCTemporalDenoise.avs")
Import("\AviSynth 2.5\plugins\GradFun2DBmod.v1.5.avs")
SetMemoryMax(2048)

AVCSource("something.dga")

Crop(0,20,-0,-20)

MT(""" MCTemporalDenoise(settings="very low",sigma=6,chroma=true,stabilize=true,GPU=false,ncpu=8,enhance=true,sharp=false) """, threads=8, overlap=16)

Even though it's 1920x1040, mod16, it doesn't want to work, reports this isn't mod4.

MT is the problem, use SetMTMode... (or adjust threads in MT: "height/threads" must be mod4)

[Gavino]

Originally Posted by LaTo

MT is the problem, use SetMTMode... (or adjust threads in MT: "height/threads" must be mod4)

Or use splitvertical=true, since width/threads is already mod4.

[Boulder]

Originally Posted by Didée

It doesn't miss "a" scenechange. It does miss *every* scenechange. To my knowledge, temporal processing of FFT3DFilter simply doesn't check or care for that.

I wonder if anyone with some programming skills could add scenechange detection if Fizick doesn't want to do that. It should be quite simple since there are many filters to borrow code from (of course it depends on the case, but still..)

[senseiam]

Originally Posted by Gavino

Or use splitvertical=true, since width/threads is already mod4.

ok thanks!

[asarian]

Originally Posted by Boulder

I wonder if anyone with some programming skills could add scenechange detection if Fizick doesn't want to do that. It should be quite simple since there are many filters to borrow code from (of course it depends on the case, but still..)

I second that motion. :) I'm not versed enough in the intricacies of motion estimation to even begin to think about doing it myself; but it would be great if someone would actually fix it.

[asarian]

LaTo, I may not be doing it entirely right yet. :) I'm using this avs script on a 1-minute clip:

LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\FFMS2.dll")
Loadplugin("C:\Program Files\AviSynth 2.5\plugins\FFT3DFilter.dll")
Loadplugin("C:\Program Files\AviSynth 2.5\plugins\mt_masktools-25.dll")
Loadplugin("C:\Program Files\AviSynth 2.5\plugins\mvtools2.dll")
Loadplugin("C:\Program Files\AviSynth 2.5\plugins\RemoveGrainSSE3.dll")
Loadplugin("C:\Program Files\AviSynth 2.5\plugins\TTempSmooth.dll")
Loadplugin("C:\Program Files\AviSynth 2.5\plugins\RepairSSE3.dll")
Loadplugin("C:\Program Files\AviSynth 2.5\plugins\deblock.dll")
Loadplugin("C:\Program Files\AviSynth 2.5\plugins\DctFilter.dll")
Loadplugin("C:\Program Files\AviSynth 2.5\plugins\EEDI2.dll")
Loadplugin("C:\Program Files\AviSynth 2.5\plugins\SangNom.dll")
Loadplugin("C:\Program Files\AviSynth 2.5\plugins\gradfun2db.dll")
Loadplugin("C:\Program Files\AviSynth 2.5\plugins\AddGrainC.dll")

video=FFVideoSource("M:\jobs\2010test.mkv").ConvertToYV12()
video=MCTemporalDenoise(video, settings="very high")
video=Crop(video, 0, 142, 0, -142)

return video

Nothing happens for a long while, though. First I get a Windoze error, saying it can't use "mt_masktools-25-x64.dll" (of course it can't: I'm on 32bit; and I didn't ask for it to be run, either). Then x264 keeps hovering around 25% CPU, steadily increasing in memory size (to around 1.4GB). And finally it tells me that 1 minute clip, at 0.00fps, (sic!) will take over 300 hours! Why, that seems a mite extreme. :)

x264 remains at ca. 25% CPU. So, I guess I'm doing something wrong. Any hints?

[tormento]

Originally Posted by Boulder

I wonder if anyone with some programming skills could add scenechange detection if Fizick doesn't want to do that.

And a x64 version too =P

[Didée]

Originally Posted by tormento

And a x64 version too =P

There already is one. (Without SC detection, of course.)