View Single Post
Old 18th December 2017, 08:38   #47801  |  Link
edcrfv94
Registered User
 
Join Date: Apr 2015
Posts: 84
Quote:
Originally Posted by madshi View Post
In case you guys wonder how NGU Sharp compares to mpv's latest FSRCNN(X), here's a little comparison:

Blu-Ray screenshot
downscaled (PNG) | (JPG, 100 quality)
latest FSRCNN32
latest FSRCNNX32
NGU Sharp - High
NGU Sharp - Very High

To make things as fair as possible I've downscaled the image with Bicubic/Catrom, which is exactly what FSRCNN and FSRCNNX were trained for.

Here are benchmark numbers, for 720p doubling:

Code:
Nvidia 1070:
FSRCNN16: 15.270 ms
FSRCNNX16: 26.397 ms
FSRCNN32: 46.290 ms
FSRCNNX32: ? (estimated: 80.021 ms)
NGU-Sharp High: 3.940 ms
NGU-Sharp Very High: 11.800 ms
Code:
AMD 560:
FSRCNN16: 14.289 ms
FSRCNNX16: 24.412 ms
FSRCNN32: 45.235 ms
FSRCNNX32: ? (estimated: 77.282 ms)
NGU-Sharp High: 12.970 ms
NGU-Sharp Very High: 37.100 ms
These are very weird benchmark results, to say the least. We know that NGU doesn't run as well as it should on AMD Polaris GPUs. But FSRCNN(X) running (ever so slighty) faster on my AMD 560 than on my Nvidia 1070 is just plain weird.
Look well, very expect NGU can be used at AviSynth also can NGU use for Deinterlacers to replace nnedi3(QTGMC)?

Last edited by edcrfv94; 18th December 2017 at 11:06.
edcrfv94 is offline   Reply With Quote