https://www.svp-team.com/forum/viewtopi … 672#p84672
updated svpflow2_vs.dll, default value for scene.limits.blocks set to 40 for RIFE
You are not logged in. Please login or register.
SmoothVideo Project → Posts by Chainik
https://www.svp-team.com/forum/viewtopi … 672#p84672
updated svpflow2_vs.dll, default value for scene.limits.blocks set to 40 for RIFE
> instead finding rife_sc all the time?
just don't close the window
> Then how to make proper frame blending when RIFE_SC threshold over 15?
this is why you need a MV-based scene change detection instead of dumb image comparison
%APPDATA%\SVP4\override.js
smooth.scene.limits.blocks = 40;
default value is 20
with 40 at least the Hugo intro - the fly over the city and through the train station - is better (almost no scene changes)
btw
smooth.debug.qmode = true;
will draw a color square in the frame corner, when it's red it's a scene change
flowreen91
> I think it's easier to test if u go to line 222 from generate.js and delete the following
how is this easier than choosing "6%" in the gui?
> NVOF algo fixes vertical motion artefacts but causes problems elsewhere.
It just can't "cause problems". It can only:
a. (_very_ unlikely!) miss the scene change that is actually a scene change --> which shows RIFE's artifacts at this frame
b. mark a scene change where there's no scene change --> makes it choppy where it could be not choppy
that Hugo?
"6%" or "8%" mean nothing here
IF "6%" THEN USE "scene change based on SVP's motion vectors"
IF "8%" THEN USE "scene change based on NVOF's motion vectors"
ELSE USE "dumb frame comparison"
Testing RIFE update with SVP's scene change detection added.
Replace script\generate.js, script\base.py and plugins64\svpflow2_vs.dll, restart SVP.
Now THREE different modes are possible, controlled by the "Scene change threshold" profile option:
1. default - old behavior with "dumb" scene change
2. scene change based on SVP's motion vectors - when "scene change threshold" is "6%"
3. scene change based on NVOF's motion vectors - when "scene change threshold" is "8%"
Scene change-related params from override.js should also work, such as smooth.scene.limits.zero, .blocks, .scene (see a short description there - https://www.svp-team.com/wiki/Manual:SVPflow)
UPDATE: svpflow2_vs.dll updated, also supporting blending on scene changes instead of frame repeating.
There's no "processing of scene changes" option in the RIFE profile so you have to set it via 'All settings' - find the profile and set 'fi_scene_changes' to 0 (blend) or 1 (repeat)
UPDATE 2: svpflow2_vs.dll updated, default value for scene.limits.blocks set to 40 for RIFE
UPDATE 3: svpflow2_vs.dll updated, using correct frame for repeating on a scene change
because they are ... different
> I would switch to MPV but I can't stand that it's completely barebones with zero UI
SMPlayer
> I can't actually find any ncnn vs dml benchmarks for rife.
TRT is 2 times faster than ncnn
DML is _10_ times slower than TRT
> And please update MPV to the latest stable v0.38.0
nope, it crashes playing DoVi
flowreen91
step 6: "enjoy" what exactly? 10-15% performance drop?
another post that is actually a bad advice for people who don't understand what their doing, just following "guides"
----
dawkinscm
> If you have to use Rife v2 and have picture shaking issues then updating vslmrt makes sense.
only for TRT>=9
with TRT8 updating vsmlrt.py most likely does nothing
only difference between SVP's vsmlrt.py and current master is one line:
# TensorRT 9.0 or later
"ONNXTRT_Broadcast_*:fp32"
> How can I change the TRT version? And how can I find out the current version? I really don’t like the scene change on any model at any 6%-100% values.
this isn't related to the TRT version at all
TRT version can only affect build time, memory usage and overall performance
one more time, what is the exact point of "v2"?
it's not-faster or even slower, and it takes more RAM. so, why?
just because "v2" is cooler than "v1"? also tensort rt v10 is cooler than v8...
when you're using RIFE with Avisynth Filter, it's always a Vulcan/ncnn version.
> I did install python-3.12.3-amd64 and VapourSynth-x64-R66
SVP already have everything needed
you actually broke things by installing this AND also "set variables" from SVP
you must either use what SVP provides OR external python/vaporsynth, not both
re(install) Avisynth+ manually
https://github.com/AviSynth/AviSynthPlu … redist.exe
why? the only difference is TensorRT libs 10.0.1 which are probably _slower_ than 9.2
поискать в драйверах Интел "улучшатели" видео, и всё выключить
> Huge improvement over ncnn for Radeon and Arc.
any benchmarks?
начиная с mpv-x86_64-20240310-git-a56d8ff - падает на DV
[/пожимает плечами]
проект перешел в фазу когда новые релизы никакой пользы не несут, а токмо добавляют багов рефакторинг кода ради самого процесса рефакторинга, чисто чтоб без дела не сидеть.
Android TV specific
---------------------------
If you want on-device FRC then you really should look at RK3588-based TV boxes (or better). It can handle 1080p, but nothing more.
Anything slower won't be usable for FRC, even NV Shield is too weak.
BUT you can try with SVPcast streaming - i.e. doing FRC on a desktop PC in the same LAN. For this use case hardware requirements are much lower.
Because there're so little TV boxes that could handle SVPlayer, we don't want to make a Google Play version, cause it will obviously have 1.0 rating
So only an APK version for now.
Apple TV specific
---------------------------
Tested with Apple TV 4K 1st gen (2017) and 3rd gen (2022).
4K 2017 based on Apple A10, so it's rather slow. Only 1080p video with low interpolation settings.
4K 2022 based on A15, so it _could_ be powerful BUT someone decided it must be _cheap_ so they removed an active cooling. As a result it always overheating and thermal throttling like hell.
Because of that it can't handle 4K videos too (despite of A15-based iPad can do it).
This mod is a must with 4K 2022:
Modded 4K 2022, with output set to 1080p in Settings and "resize to 1080p" in SVPlayer, can play anything.
SmoothVideo Project → Posts by Chainik
Powered by PunBB, supported by Informer Technologies, Inc.