I have shifted my video tinkering the last several years toward filtering via Avisynth scripts and nearly completely away from Virtualdub filtering, but merely using Vdub as the outputter with script as input.
Source video is 720p and am making output 1080p. For sharpening, the standard "sharpen" in Vdub. I've not been able to see any difference between this and all the myriad of other sharpeners in either Vdub or Avisynth. Also, I see no functional reason not to ramp it up all the way to 64. My source must be clean enough that sufficient artifacting st 64 doesn't happen. Yes, my final encode may be 20-30 bigger due to micro-noise being sharpened, but that doesn't bother me.
That's sharpening, now I want to resize. I was doing resize first via SuperRes in Avisynth (SR(1920,1080)), then Sharpen(setting=64) in Vdub. Of course the order was dictated as that. BUT then I remembered I can still do SR in VDub. So once the 2 actions are both in Vdub, I can control the ORDER in which they occur.
So, the question is which order is better, SuperRes resize to 1080p first or Vdub Sharpen(64) first? My old eyes, even close up have hard time discerning.
+ Reply to Thread
Results 1 to 5 of 5
Not sure if this helps since I can't really say anything about VirtualDub since I haven't really used it for decades.
Whether it is better to sharpen before or after resize usually depends on the source, the noise it has and the strength or the sharpening.
So I would recommend to simply open two instances of VirtualDub do one order in one instance and the other order in the other and compare them.
(usually I also add some denoising before and and deringing after sharpening)
Cu Selurusers currently on my ignore list: deadrats, Stears555
VirtualDub's built in sharpen is indiscriminate -- it will increase noise as well as sharpen edges. And's it's radius is only one -- it will not help at all with very blurry video.
Last edited by jagabo; 6th Jun 2020 at 09:04.
If you sharpen before you scale, the scaled video will look sharper and not need as high of strength sharpening. If you sharpen after you scale, it will require a higher strength sharpen setting to get close to the same results.
I personally do it before I scale unless the source was already super sharp. If it was super sharp to begin with, then I do it after I scale, if it's even needed afterwards. I also do this with avisynth and not Vdub, so I am only assuming it will be the same way, but might not be as good as all the scaling options you would have with avisynth. I dont think Vdub's internal filters, and built in things come out as nice as avisynth's usually do. It's also kind of obsolete for the most part nowadays and I havent used it since DIVX/XVID AVI was still popular in 2004. If your using avisynth and vdub, I don't really see a reason to scale and sharpen using both. One or the other should be sufficient.
In avisynth, something along the lines of this should give a fairly sharpened upscale without going way too overboard.
LimitedSharpenFaster(ss_x=1.0, ss_y=1.0, strength=12, overshoot=0, undershoot=0, soft=0, edgemode=0) # Raise strength for more sharpening. Try to keep at 40 or below, or else lines can start to alias as you go higher with the supersampling off. NNEDI3_rpow2(rfactor=2, cshift="LanczosResize", fwidth=1920, fheight=1080, nns=4) # Adjust resolution as needed, this assumes its 16:9 aspect ratio
LimitedSharpenFaster(ss_x=1.0, ss_y=1.0, strength=14, overshoot=0, undershoot=0, soft=0, edgemode=0) NNEDI3_rpow2(rfactor=2, cshift="LanczosResize", fwidth=1920, fheight=1080, nns=4) mctemporaldenoise(sigma=1, sharp=false, radius=1, ecrad=1, AGstr=0.6, GFthr=1.00) # Raise sigma to 2-3 for stronger cleaning. Movie it to before NNEDI3 if sigma 3 is still not clean enough for you.
Last edited by killerteengohan; 8th Jun 2020 at 20:37.
If you sharpen before you scale, you will be magnifying the interpolations (incl. halos). Sharpen is almost always best done at the very end.