Dfftest so slow.
- Samuru21
- Joined: Sat Sep 27, 2008 8:54 am
- Location: Upper Darby,PA
Dfftest so slow.
Is there a faster alternative that gives the same results. Converting a 30 sec clip to lossless takes 20 to 30 minutes!
Yeah i guess so....
- Samuru21
- Joined: Sat Sep 27, 2008 8:54 am
- Location: Upper Darby,PA
- mirkosp
- The Absolute Mudman
- Joined: Mon Apr 24, 2006 6:24 am
- Status: (」・ワ・)」(⊃・ワ・)⊃
- Location: Gallarate (VA), Italy
- Contact:
Re: Dfftest so slow.
Could try to use fft3dgpu, but then again it'd help only if you have a good gpu...
Otherwise, for the same results, you'd hardly find anything as good, with better speed. Perhaps vaguedenoiser could be an alternative. Can sometimes be better than fft3d, but is always faster.
I'm not entirely sure how the speed of fft3d compares to dfttest, but IIRC it's about the same speed if not slower.
Otherwise, for the same results, you'd hardly find anything as good, with better speed. Perhaps vaguedenoiser could be an alternative. Can sometimes be better than fft3d, but is always faster.
I'm not entirely sure how the speed of fft3d compares to dfttest, but IIRC it's about the same speed if not slower.
- Samuru21
- Joined: Sat Sep 27, 2008 8:54 am
- Location: Upper Darby,PA
- Qyot27
- Surreptitious fluffy bunny
- Joined: Fri Aug 30, 2002 12:08 pm
- Status: Creepin' between the bullfrogs
- Location: St. Pete, FL
- Contact:
Re: Dfftest so slow.
What else is in the script?
Be aware that sometimes script rendering times can multiply themselves inexplicably based on which filters are being run and how they operate on the frames in a video.
For example, DFMDeRainbow and LUTDeRainbow perform different forms of derainbowing, and together can knock out practically all the rainbows from the sources I've tried them on (the original AnimEigo release of the Oh My Goddess! OVA, for example - egregiously bad rainbowing, but that combo gets rid of something like 99% of them). Separately, they can get the job done in a matter of maybe 10-20 minutes each (my comp is really old), but having both activated in the same script will usually inflate the conversion time to over an hour.
Be aware that sometimes script rendering times can multiply themselves inexplicably based on which filters are being run and how they operate on the frames in a video.
For example, DFMDeRainbow and LUTDeRainbow perform different forms of derainbowing, and together can knock out practically all the rainbows from the sources I've tried them on (the original AnimEigo release of the Oh My Goddess! OVA, for example - egregiously bad rainbowing, but that combo gets rid of something like 99% of them). Separately, they can get the job done in a matter of maybe 10-20 minutes each (my comp is really old), but having both activated in the same script will usually inflate the conversion time to over an hour.
My profile on MyAnimeList | Quasistatic Regret: yeah, yeah, I finally got a blog
- Samuru21
- Joined: Sat Sep 27, 2008 8:54 am
- Location: Upper Darby,PA
Re: Dfftest so slow.
lanczos4resize(640,480)
tweak(cont=1, sat=1.1)
levels(0,0.946,241,0,255)
sangnom()
dfttest()
removedirt()
deen("a2d",2,2,2)
LimitedSharpenFaster(strength=119)
FastLineDarken(strength=170,luma_cap=225,thinning=0)
tweak(cont=1, sat=1.1)
levels(0,0.946,241,0,255)
sangnom()
dfttest()
removedirt()
deen("a2d",2,2,2)
LimitedSharpenFaster(strength=119)
FastLineDarken(strength=170,luma_cap=225,thinning=0)
Yeah i guess so....
- Qyot27
- Surreptitious fluffy bunny
- Joined: Fri Aug 30, 2002 12:08 pm
- Status: Creepin' between the bullfrogs
- Location: St. Pete, FL
- Contact:
Re: Dfftest so slow.
Ok, so there's one resizer, two correction filters, an anti-aliaser, 3 smoothers, and 2 sharpeners/masks. And that deen line is pathetically weak. I doubt it even does anything noticeable at those settings.Samuru21 wrote:lanczos4resize(640,480)
tweak(cont=1, sat=1.1)
levels(0,0.946,241,0,255)
sangnom()
dfttest()
removedirt()
deen("a2d",2,2,2)
LimitedSharpenFaster(strength=119)
FastLineDarken(strength=170,luma_cap=225,thinning=0)
Doesn't Sangnom have a reputation of being a slow filter? I seem to remember so, but it's been a couple years since I toyed around with it.
My profile on MyAnimeList | Quasistatic Regret: yeah, yeah, I finally got a blog
- mirkosp
- The Absolute Mudman
- Joined: Mon Apr 24, 2006 6:24 am
- Status: (」・ワ・)」(⊃・ワ・)⊃
- Location: Gallarate (VA), Italy
- Contact:
Re: Dfftest so slow.
Sangnom is also a pretty bad antialiaser (if just called like that anyway)...
I think when it comes down to speed, daa is pretty much the best anti-aliasing choice, though maa still is the best antialiaser choice overall.
I think when it comes down to speed, daa is pretty much the best anti-aliasing choice, though maa still is the best antialiaser choice overall.
- Samuru21
- Joined: Sat Sep 27, 2008 8:54 am
- Location: Upper Darby,PA
Re: Dfftest so slow.
Can't find anything on daa and maa? mind giving me a download link to these plug ins
Yeah i guess so....
- Qyot27
- Surreptitious fluffy bunny
- Joined: Fri Aug 30, 2002 12:08 pm
- Status: Creepin' between the bullfrogs
- Location: St. Pete, FL
- Contact:
Re: Dfftest so slow.
daa() is part of AnimeIVTC; I would assume maa() is as well, since it's mentioned in AnimeIVTC's release post on doom9.
So, in other words, if you've got the latest AMVapp, you already have them, as the AMVapp includes AnimeIVTC, if I remember correctly.
So, in other words, if you've got the latest AMVapp, you already have them, as the AMVapp includes AnimeIVTC, if I remember correctly.
My profile on MyAnimeList | Quasistatic Regret: yeah, yeah, I finally got a blog