They have been designed to do their thing and usually work better separately. My experience in any field of activity is that you should use a specific tool for a specific need. Probably this should be the next thing to try out for ultimate results, combining good denoising and good sharpening. You never know when you're going to want to re-encode just a small section for some reason.įor example if a video was 10,000 long and you wanted a script to encode the first 5000 frames:Īnd I found the Scintilla Guide mentioned to be very infromative also (although geared on anime), especially in regards of shapening filters. Make sure to add -stitchable to the x264 command line. I create a script, make a copy, and then add trim() to the end of each so each script encodes around half the video (I start the second script on the first frame of a scene), then I encode them simultaneously and append the encoded video when it's done. I personally always go for the quality and I'm not on a payrun or anything so I would just stick to it. Very nice shapening with enough details retained.
TemporalDegrain(SAD1=200, SAD2=150, sigma=8)Īnd I really liked the results. But you have to be ready to do the homework also)).Īnd I found the Scintilla Guide mentioned to be very infromative also (although geared on anime), especially in regards of shapening filters. I must say I would be nowhere if some very kind users on this forum would not be helping out with all my requests. They did all the research already for you and God thank them for this. Get the best from the best and hardy you will make a mistake. Also it's good to note down encoding speeds for all of them if speed is important for you.īut the best approach for anything you don't know is to ask a master and this is one of the blessings of a forum. Then get the best of them and try tuning the best out of them.
DENOISER 2 SPEED DOWNLOAD
Download all of them and do the default testing. Now I have an idea what can all those noise reduction filters do with the default settings and that's a good starting place to refine your filters in avisynth, whatever they may be. But the real lessons I learned are as follows: I must say that these are all pretty much speculative conclusions based on personal liking. I'm sure if I would really know all the avisynth filters I would probably do it differently but the main thing is to get the results you are looking for so I guess it beats the sophistication. In my case I used an old and really bad source, and had all the three (qtgmc, temporaldegrain and deen) put on it, and it came out pretty good. Yep, I agree, I also like FFT3DFilter more, some subtle difference is there.Īs far as combining Qtgmc and Deen, after doing all the tests today I think it's not worth it, at least in the case of your sample. I probably should revisit that at some stage.
So I took some settings down and used this scriptĪt one stage I tried getting QTGMC to use dfttest for the denoising (it uses FFT3DFilter by default) but from memory it was a fair bit slower and for some reason I preferred the result with FFT3DFilter doing the denoising. Just by looking at the samples with the default settings, I personally found that dfttest and TemporalDegrain are the best ones, although TemporalDegrain is a bit overkill with the default settings. That said, I did some homework and checked out the samples in this thread. Actually I came to this thread with the desire to dwelve into the matter deeper and develop some skill, so at this point you can't take me too seriously. But I do have some really bad old material, so to remove as much noise as I like, I try different things. I haven't really experienced in avisynth, just starting recently, so I don't have much to say in this regard. I do use sometimes even 3 different ones, but I think that's just my laziness to tweak them to do the same by combining 2 or sticking to one. In what order do you run deen and QTGMC and do yo use the default deen settings? I've never been able to run two different noise filters effectively, but I'll try anything.
0 kommentar(er)
