Publisher
Joined: 4/23/2009
Posts: 1283
Which is why you should try nnedi3 and compare to see which you like better. It's your video and your channel, choose what you like, not what TASVideo channel does =p.
Player (170)
Joined: 7/1/2016
Posts: 263
Can you help a little bit more? Too many options. Is there a way to reduce such blending using QTGMC? Like you can see some artefacts from Zombie that died on previous frame. edit: i also got an up / down shaking (jitter) or something. Any simple solution on that?
Publisher
Joined: 4/23/2009
Posts: 1283
You would need to post your script. Also, I assume you didn't recapture with YUY2. Converting to RGB then back to YV12 can cause those bleeding also =p. Hint, get a new source capturing the RIGHT way.
Player (170)
Joined: 7/1/2016
Posts: 263
avisource("1.avi") ConvertToYV12() AssumeTFF() QTGMC( Preset="Placebo", SourceMatch=3, Lossless=2, NoiseProcess=1, NoiseRestore=0.0, Denoiser="dfttest", DenoiseMC=true, NoiseTR=2, Sigma=4.0, Sharpness=0.5, SLMode=1) Crop(8, 0, -8, 0) nnedi3_rpow2(rfactor=4,cshift="LanczosResize",fwidth=3168,fheight=2160) Aim for 4K youtube encode. Denoise to reduce vertical color lines and i would like to get a little bit sharper image. Is those settings ok? I did capture new clip in YUY2.
Publisher
Joined: 4/23/2009
Posts: 1283
Man, I really really wish you would get on IRC than talk through here. Yea your script is a bit off. First, let's try this:
Language: avisynth

Avisource("1.avi") AssumeTFF() QTGMC(Preset="Placebo", SourceMatch=3, Lossless=2, MatchPreset="Placebo", MatchPreset2="Placebo")
In VirtualDub, is the problem still there?
Player (170)
Joined: 7/1/2016
Posts: 263
I know, it's pretty rude to ask something like this, but maybe someone can make deinterlace script for me? Here is lossless source file from GVUSB2 with correct color space: https://www.dropbox.com/s/7tg5aofm3wezkvo/amarec%2820170525-2030%29.avi I read that QTGMC article and i was trying to google, but nothing helps. Converting to YV12 before QTGMC, not converting to YV12 - same blended result. I want to remove such blending if possible: Also reduce such noise in the best way: And also deinterlaced picture shakes up and down every other frame. So i want to eliminate that shaking as well. Again can't find how to do that with QTGMC. Only with different functions / filters. Sorry
creaothceann
He/Him
Editor
Joined: 4/7/2005
Posts: 1874
Location: Germany
Try this script:
Language: Avisynth

AVISource("amarec(20170525-2030).avi") AssumeTFF SeparateFields Interleave(last, BlankClip(last)) AssumeTFF Weave Crop(42, 0, -30, 0)
If nobody else here has an idea you could try asking in the doom9 forum.
Publisher
Joined: 4/23/2009
Posts: 1283
I'll take a look later today. Having a sample clip helps a lot =p
Player (170)
Joined: 7/1/2016
Posts: 263
creaothceann wrote:
Try this script:
Language: Avisynth

AVISource("amarec(20170525-2030).avi") AssumeTFF SeparateFields Interleave(last, BlankClip(last)) AssumeTFF Weave Crop(42, 0, -30, 0)
If nobody else here has an idea you could try asking in the doom9 forum.
Thank you! No blending and shaking, but it have like CRT scanlines / pixels look. This is pretty cool actually! But is it possible to get something like this without "CRT" effect?
Publisher
Joined: 4/23/2009
Posts: 1283
You know, I forgot that NES is 60 FPS non interlaced. My bad. Basically if you remove the lines:
Language: avisynth

Interleave(last, BlankClip(last)) AssumeTFF Weave
and resize, you should be fine.
Player (170)
Joined: 7/1/2016
Posts: 263
So basically it will be the script from my first post, lol So only mistake i made - i use RGB color space instead of YUY2 for capture?
Publisher
Joined: 4/23/2009
Posts: 1283
Yep, brain farted there, my bad again. Course you still did capture to RGB which was a big no no. So now to help you with the rest of it. I still suggest nnedi3, and if you want 4k, it be this line:
Language: avisynth

nnedi3_rpow2(rfactor=16, nsize=0, nns=4, qual=2, cshift="lanczosresize", ep0=2, pscrn=4, fwidth=2880, fheight=2160)
Then you need to convert to YV12 before encoding to x264, so use this line:
Language: avisynth

ConvertToYV12(matrix="Rec709", chromaresample="lanczos")
Site Admin, Skilled player (1235)
Joined: 4/17/2010
Posts: 11264
Location: RU
Koh1fds wrote:
So basically it will be the script from my first post, lol
Told ya :D
Warning: When making decisions, I try to collect as much data as possible before actually deciding. I try to abstract away and see the principles behind real world events and people's opinions. I try to generalize them and turn into something clear and reusable. I hate depending on unpredictable and having to make lottery guesses. Any problem can be solved by systems thinking and acting.
Publisher
Joined: 4/23/2009
Posts: 1283
feos wrote:
Told ya :D
If you knew what was wrong with deinterlacing, why not speak up then? =p
Player (170)
Joined: 7/1/2016
Posts: 263
Any suggestion on dfttest()? Maybe i should use different denoiser? It helps by a lot. Left side denoised by dfttest(). Right side without denoise.
Publisher
Joined: 4/23/2009
Posts: 1283
Just to be sure, make sure the conversion to YV12 is last. You want all the information on color you can to do things, before throwing it out for YouTube.
Player (170)
Joined: 7/1/2016
Posts: 263
Language: avisynth

avisource("movie.avi") AssumeTFF SeparateFields Crop(8, 0, -8, 0) dfttest() nnedi3_rpow2(rfactor=16, nsize=0, nns=4, qual=2, cshift="lanczosresize", ep0=2, pscrn=4, fwidth=3168, fheight=2160) ConvertToYV12(matrix="Rec709", chromaresample="lanczos")
Any mistakes / suggestions? I want same aspect as source file.
Publisher
Joined: 4/23/2009
Posts: 1283
the source is the wrong aspect ratio as TV has non square pixels. The nnedi3 line already fixes this to the correct 4:3 aspect ratio. Do you still want the same aspect ratio as the source? Edit: My nnedi3 line fixes it, yours is back to the wrong aspect ratio.
Player (170)
Joined: 7/1/2016
Posts: 263
Ok. Let's test it with some bubble bobble.
Publisher
Joined: 4/23/2009
Posts: 1283
I hope you have a video camera to record your actual TV (prefer a CRT TV) to show it is the same.
Player (170)
Joined: 7/1/2016
Posts: 263
Ok. Seems like you win this round. http://i.imgur.com/0keU5hf.png I wanna have "round" bubbles. So bubbles with your aspect is more round. Thanks everyone! I think now this script are pretty good.
Publisher
Joined: 4/23/2009
Posts: 1283
No problem, though I wasn't trying to "win" you know. I was trying to help you get the best quality.
Site Admin, Skilled player (1235)
Joined: 4/17/2010
Posts: 11264
Location: RU
Aktan wrote:
feos wrote:
Told ya :D
If you knew what was wrong with deinterlacing, why not speak up then? =p
How could I know about deinterlacing before you suggested it? You looked confident so I didn't bother actually pondering your suggestions (I just dislike how these plugins make it look).
Warning: When making decisions, I try to collect as much data as possible before actually deciding. I try to abstract away and see the principles behind real world events and people's opinions. I try to generalize them and turn into something clear and reusable. I hate depending on unpredictable and having to make lottery guesses. Any problem can be solved by systems thinking and acting.
Publisher
Joined: 4/23/2009
Posts: 1283
feos wrote:
How could I know about deinterlacing before you suggested it? You looked confident so I didn't bother actually pondering your suggestions (I just dislike how that previous plugin makes it look).
I was saying to speak up after I mention deinterlacing! As you know (that I forgot), NES/SNES are actually not interlaced and displays all even (or odd, I forget) fields at 60 FPS. Hence the scanlines. If I had been reminded of this fact, I would have stop trying to deinterlace right away! Which previous plugin did you dislike? I didn't get you on that part.
Site Admin, Skilled player (1235)
Joined: 4/17/2010
Posts: 11264
Location: RU
I edited my post regarding plugins. And as I said, I didn't bother pondering, so I didn't recall that nes is progressive scan.
Warning: When making decisions, I try to collect as much data as possible before actually deciding. I try to abstract away and see the principles behind real world events and people's opinions. I try to generalize them and turn into something clear and reusable. I hate depending on unpredictable and having to make lottery guesses. Any problem can be solved by systems thinking and acting.