Blackfyre wrote:RickyAstle98 wrote:TechnoStone wrote:V2 models for some reason have jiggling, the more to the right and down, the worse the effect. Almost unnoticeable for 1080p, but very annoying for 1440p and 4K (both native and downscaled to 1440p, here's the example - https://cdn.discordapp.com/attachments/ … WGll8w.mp4). For this one, I'm running Game of Thrones S1 E1 at 4K on 3440x1440 (so SVP downscales to 2560x1440) at x2 framerate on 2080 Ti with 4.4v2, 4.6v2, 4.9v2 and 4.13_litev2. There's no such effect on V1 models (tried all same versions)
I dont have this problem, maybe you use target frame rate instead of interpolation factors? Because I do slomo video and see what happens on 25FPS video converted to 2.5x (no issues) (62.5FPS)
Then I do frame rate target at 60FPS and see your issue all the time, even without slomo effect! Dont use target framerate, factors work better!
Also this, make sure it's either 2x, 3x, 4x etc
Then open the drop down menu to the right of the slider, and enable at the bottom (Force Exact Value).
cws wrote:It's true there's no significant performance improvement, but there may be fixes and other minor improvements.
Specifically, when using v13 (with TensorRT 8.5 / CUDA 11.8) with the v2 models, I can see some blurriness and pixels shifting in the bottom right corner if there is some obvious static content there like a logo. If this is something you can reproduce, try using the latest v14 (with TensorRT 9.x / CUDA 12.x)
I checked by making a backups of both old SVP TensorRT and latest TensorRT v14 and switching them back and forth by looking at the mentioned trees scene.
I noticed that issue was present on all RIFE (v2) i tested even latest RIFE 14.4 (v2) on old SVP TensorRT.
Issue persisted when trying fixed FPS with Force Exact value on/off or FPS x2, which means it's unrelated.
I could not reproduce it anymore when replacing the old SVP TensorRT with latest TensorRT, which means it got FIXED on a version between old and latest.
Current default SVP installation has 4.6/4.9 v1 models which means average user never saw this jiggling issue.
In the future when SVP devs will want to add the v2 models they will surely update the default old TensorRT libraries too so your average user will never see this issue.
Until then we should follow @cws guide on how to update our own TensorRT library from https://github.com/AmusementClub/vs-mlrt/releases if we want to play around with latest RIFE models:
cws wrote:Right, TensorRT 9.x is in "beta" and not officially released. Still, v13 was released when TensorRT 8.5 was the latest version, when there is a new release of TensorRT 8.x (which is 8.6.)
There may be fixes and enhancements in the latest v14 releases which is not in v13. If you're willing to help test, I would try downloading https://github.com/AmusementClub/vs-mlr … 4.test3.7z and unzipping
and the folder
into
C:\Program Files (x86)\SVP 4\rife