RickyAstle98 wrote:

Anybody has mpv shader like Intel denoiser (which makes all videos looks pretty without quality loss)?

I wouldn't recommend ever using a denoiser unless you have poor quality sources. You say "without quality loss" but with any denoiser there is always a risk of losing high frequency detail.

aloola wrote:

mpc-BE + mpc-VR (RTX HDR + RTX SuperRes) + SVP RIFE give you the best experience and easy to config.

also for mpv I found this is a good shader for anime/real-life videos https://github.com/cunnyplapper/CuNNy/t … r/mpv/fp16

I just tried CuNNY and 4x32 is the best option matching FSRCNNX_x2_16-0-4. But CuNNy 4x32 is similar to nnedi32 in that it's heavier on GPU, especially when using SVP.


Drakko01 wrote:
dawkinscm wrote:
Drakko01 wrote:

I've been using RIFE since it became available and I never thought it could improve this much. BTW which is the full resolution of your rips and wich resolution its your display

1920x2160.

It's a odd resolution to use and also a odd screen resolution. Can you explain the reason for choosing that resolution in your rips?

I didn't choose it. This is the standard resolution for 3D blu-ray.

aloola wrote:

mpc-BE + mpc-VR (RTX HDR + RTX SuperRes) + SVP RIFE give you the best experience and easy to config.

also for mpv I found this is a good shader for anime/real-life videos https://github.com/cunnyplapper/CuNNy/t … r/mpv/fp16

Thanks for the link. In general I tend to downscale to FHD rather than upscale because there is zero advantage to upscaling my own sources. But this could be useful for poor quality acquired stuff so I will give it a go. Also I don't use a PC for 4K HDR.

Drakko01 wrote:
dawkinscm wrote:

Today I tried MPC+MadVR with a simpler config and it still couldn't handle my full resolution rips using Rife. However saying all that I'm also happy running my files reduced to FHD since it uses around 40% less GPU with Rife v4.15v2.

I've been using RIFE since it became available and I never thought it could improve this much. BTW which is the full resolution of your rips and wich resolution its your display

1920x2160.

Drakko01 wrote:
dawkinscm wrote:
Drakko01 wrote:

I am exactly at your numbers even using TRT 9.2.

I was always  curious why the devs never updated the TRT Version if there its a reason not to.

It's not stable. When you you are at my numbers with TRT 9.2, are you running large files that have 2x the vertical resolution of a FHD file at 72fps on v4.15?

Maybe i I didn't understood what resolution do you use, but you statement was a little vague...

Generally I play 4k sources downscale to 1920*800 by svp and upscale to 2560*1440 with madvr.

That makes sense. Especially since mpv works much better than MPC+MadVR for SVP+Rife. Today I tried MPC+MadVR with a simpler config and it still couldn't handle my full resolution rips using Rife. However saying all that I'm also happy running my files reduced to FHD since it uses around 40% less GPU with Rife v4.15v2.

I downloaded the latest version of MPC-BE today and I realised a few things. Firstly, for users with lower spec GPUs, MPC is the best choice over MPV because MPV is basically MPC+MadVR. If you want to add MadVR then MPV is more efficient and uses less resources.

The other thing I realised is that if you play a video file with HD audio connected to an external amp it uses about 20% more GPU. Using the GPU makes sense, I just didn't realise how much GPU is potentially used.

Drakko01 wrote:
dawkinscm wrote:
RickyAstle98 wrote:

FHD or 4K?

Higher than FHD but definitely not 4k. At FHD I can run a little higher at 75fps.

BTW After more testing with the default TRT 8.5.1, Rife v4.16 lite and v4.15 lite are almost identical. Rife v4.15 is slightly better than both. I don't know about anyone else but for me, it looks like I should have stuck with the default TRT.

I am exactly at your numbers even using TRT 9.2.

I was always  curious why the devs never updated the TRT Version if there its a reason not to.

It's not stable. When you you are at my numbers with TRT 9.2, are you running large files that have 2x the vertical resolution of a FHD file at 72fps on v4.15?

RickyAstle98 wrote:
dawkinscm wrote:

Now that I'm back on TRT8.5.1 I am able to run Rife v4.15v2 at 72fps which uses about 75-90% GPU. The lite version uses about 60-80% GPU. For me I suppose that it goes without saying that:

4.15v2/72 > 4.15v2 lite/72 > 4.15v2/60 > 4.15v2/60 lite > everything else.

FHD or 4K?

Higher than FHD but definitely not 4k. At FHD I can run a little higher at 75fps.

BTW After more testing with the default TRT 8.5.1, Rife v4.16 lite and v4.15 lite are almost identical. Rife v4.15 is slightly better than both. I don't know about anyone else but for me, it looks like I should have stuck with the default TRT.

Now that I'm back on TRT8.5.1 I am able to run Rife v4.15v2 at 72fps which uses about 75-90% GPU. The lite version uses about 60-80% GPU. For me I suppose that it goes without saying that:

4.15v2/72 > 4.15v2 lite/72 > 4.15v2/60 > 4.15v2/60 lite > everything else.

RickyAstle98 wrote:

You didnt understand my post, its only realtime playback bug, the models are normal, just a MPV bug, which I can fix just reencode source files! Thats because I dont give rendered example, models frame handling broken only with realtime playback! Anyway I found a fix, maybe someone facing this bug, just reencode files, voila! And how strange its sounds, I record realtime playback, and there I didnt see issue, thats why I dont give examples, even recording ontop!

So maybe I also misunderstood. I thought the video would show the artefacts but I see none. Are you saying that I have to play the clip on PC using Rife?

RickyAstle98 wrote:

This is how issue look like (video is predelayed for showcase purposes) >
https://drive.google.com/file/d/1tGmkW6 … ojEPp/view

I don't see anything even after slowing it down. What am I supposed to see?

RickyAstle98 wrote:

Yes and no, because someone maybe facing my issue with 4.10+ models, for MKV (especially HEVC coded) containers, where 4.10+ models provides improper frame handling!

I mostly have AVC encoded files but I do have a couple of HEVC encoded files which I regularly use as part of my testing and I haven't seen anything obvious.

RickyAstle98 wrote:

Also one more thing, yesterday I tested 4.15 lite and see one artefact, which 4.9 is properly mask! The lite models are less precise I think?

What I noticed was that I saw less artefacts with v4.15 when using TRT 9.2 than using TRT 8.5.1. Meanwhile 4.15 lite/72fps is still pretty good. I'm wondering if it is to do with the other changes I made to the scripts which I reverted back to default. I'm happy with 4.15 lite but I might do some experiments to see if I can improve it further by modding the scripts again.

scb wrote:
RickyAstle98 wrote:
Insindro wrote:

Could someone let me know how I’d be able to change my builder optimization level?

Line 137
Line 1179
Line 1297

Please do elaborate.. what is the impact of this? What are my options and how does it help? smile

There's very little benefit in changing it. I've played with many of the options including this one and the default settings are mostly the best settings.


RickyAstle98 wrote:
Insindro wrote:

In your opinions, which rife model is currently the smoothest?

4.4 smoothest (for me)

For me they are all about as smooth as each other but 4.15 and 4.15 lite have by far the least artefacts. 4.15 lite is the best all round.

Did you follow the MPC-HC/BE installation instructions in the SVP manual pages?

surmast14 wrote:

Thank you so much Dawkin. Your absolutely right. The bits of advice got me all the way to the last step and took days lol. That site you gave me was so much easier and 10 minutes. My copy and paste messed up too, so i reset svp and everything seems to be fine now i think. When I transcode videos it doesn't say rife engine but ignore hfr sources, but i think it's working since i put 72fps and 60fps.

You are welcome. You are almost there but not quite because if it says "ignore hfr sources" then you are still not running Rife. However if you follow the instructions on this page https://www.svp-team.com/wiki/Manual:SVPcode you should get there. Good luck smile

surmast14 wrote:

Please help me sad. Been emailing SVP4 for months. My rife tensor ai not working at all. a command prompt pops up every time.  vsmrt download not working. got a 4080 laptop. Im trying to get the best video possible. Also many youtube videos usually m3u8 8k and hdr aren't downloading for me.

It sounds like you are trying to copy bits of advice you see on here rather than just follow SVP instructions. All the information you need is here: https://www.svp-team.com/wiki/RIFE_AI_interpolation. Check the "Problems" section at the bottom of the page for help. If that doesn't work then uninstall SVP and start again, following the instructions on this page. SVP does almost everything for you.

Using Lanczos resize makes a noticeable difference to GPU usage and performance plus I'm not seeing any scaling issues. So I'm guessing that the resize happens before the interpolation.

dawkinscm wrote:
flowreen91 wrote:

3. Search "resize" in settings and specify your exact resolution downscale value. Try to add various shaders to regain visual quality.

This is interesting. I've tried downscaling before but the quality is clearly worse than mpv's class leading scaling. I think there is an option to change the scaling from bicubic to something better but it's been a while.

Chainik wrote:

> What is the downscale algorithm in SVP?

bicubic resize

SVP 4\script\generate.js line 477 big_smile
available options: https://www.vapoursynth.com/doc/functio … esize.html

This is what I was talking about but I rarely ask direct questions on here nowadays because I rarely get an answer from devs unless it's to correct an assumption I made because the devs didn't answer a previous question. Round and round it goes.

flowreen91 wrote:

Interesting. Are you talking about this particular instance? https://www.svp-team.com/forum/viewtopi … 121#p84121
Share the video if not.

Sorry I didn't see this earlier. No I'm talking about the intro to the movie Hugo which is a little difficult for Rife unless you get the SCT right or turn it off completely. But after more testing I find I can use SCT at 12 for just about everything now.

It's interesting to see how many of the changes I have recently made have now been reverted since yesterday because of how well Rife v4.15 lite works. I'm back to using the default SVP install and TRT and back to using SCT set to 12. The only remaining change is vmslrt. With v4.15 lite/72fps the remaining stubbon artefacts I come across have definitely improved over any other version while still using only 60-70% GPU.

pensioner600 wrote:

Installed MPV. Did I understand correctly that this is a player without a settings interface and everything needs to be written manually in files? And this will take a very long time to figure out.

If you installed MPV using SVP then you don't "need" to do anything else because SVP comes with a default mpv.conf file in the mpv64 folder. If you installed it separately then you will need to manually add a couple of things. I suggest you install using the SVP installer if you haven't already done so.

So this is a little weird. Before I got v4.15 Lite to be "almost" as good as v4.15/60fps by running the lite version at 72fps. But now, for the Alita battle drome scenes the Lite/72fps version actually handles artefacts a little better than base 4.15/60fps. I've made chages to my Nvidia Low latency settings which might make a processing speed difference. But I think maybe v4.15 has regressed a little with TRT 8.5 because I don't remember seeing these artefacts before but I have no actual evidence for this. I'm not prepared to do any new TRT testing until maybe 10.0.0.1 because even running at 72fps, the Lite version uses between 30 and 40% less GPU.

flowreen91 wrote:

If you set a low Scene change threshold like 1-15 then it will compare the two images and it might come to the conclusion that the difference between the two source images is big enough that it will consider it a scene transition.

On a related note, while I have been using the features of later vsmlrt versions I also set my SCT to 100.  After reseting everything to default (but still using the latest vsmlrt), I found I had to turn SCT on for best overall smoothness.The default value of 10 works best although 12 works better for one particular instance. YMMV.


flowreen91 wrote:

1. Both work, MPV slightly faster.

I would say MPV is significantly faster when stressed.


flowreen91 wrote:

3. Search "resize" in settings and specify your exact resolution downscale value. Try to add various shaders to regain visual quality.

This is interesting. I've tried downscaling before but the quality is clearly worse than mpv's class leading scaling. I think there is an option to change the scaling from bicubic to something better but it's been a while. BTW Is the number being used in your attachment 2560x1440 resolution?


flowreen91 wrote:

4. 2 threads+, enabled. Use which one you have better results.

I leave 2 threads on by default but tbh with my card I see no real difference between 1 and 4 threads.


flowreen91 wrote:

5.  Lite = uses less resources,

Except v4.14 "lite" which depending on your card can actually use more GPU resources.

OK so I'm using the latest vsmlrt script because it supposedly properly detects the TRT version. When I did that I notice that I loose some performance with the 9.2 test versions. So I reset everything back to default including the default SVP TRT version (8.5.1) and got the performance back. So where I am now is similar to where I was before. Rife 4.15v2 lite should be the new SVP default version because it uses similar GPU to v4.9 but has better artefact handling. But Rife v4.15v2 is still the best version overall because it removes or reduces every artefact I've ever come across.

abraxas wrote:

But - your milage may vary, I suggest doing some testing yourself. Especially if you are using VR.

Already have. HAGS is a problem for the VR streaming app I use.

RickyAstle98 wrote:
aloola wrote:
unrealit wrote:

Hi!

Sorry for the stupid Question!

But , How can i get Benchmark Values?

And what is HAGS?

What do you mean?

Thanks smile

use the script here.
https://github.com/AmusementClub/vs-mlrt/discussions/19

HAGS = Hardware Accelerated GPU Scheduling
you could google it for more info.

In RIFE with HAGS on performance might drop up to 25%.

HAGS ON is 25% perf drop? I think it would increase performance, no?

The reviews I've read of the feature suggests that any difference it makes to games is minimal at best. But it seems that for non games it can make things worse. I think I learned about turning it off a year ago or so from a chat I was having about VR performance issues.