Search options (Page 8 of 9)
Blackfyre wrote:Any MPV Configuration specialist that's better than me feel free to chime in here please.
I've been scratching my head trying to get VRR to work RIFE + MPV + SVP on my C2, this is the best I have achieved so far:
With Vulkan, VRR Working BUT black levels are slightly crushed + colour banding in certain bright highlights:
ontop
fullscreen=yes
volume=100
volume-max=100
vo=gpu-next
hwdec=vulkan
video-sync=audio
fbo-format=rgba16hf
spirv-compiler=auto
drm-vrr-enabled=yes
vulkan-swap-mode=fifo
target-trc=pq
target-peak=800
target-prim=bt.2020
dolbyvision=no
vf=format:dolbyvision=no
tone-mapping=spline
tone-mapping-mode=luma
tone-mapping-param=bt.2390
tone-mapping-max-boost=2.0
gamut-mapping-mode=perceptual
target-contrast=inf
hdr-compute-peak=yes
target-colorspace-hint=yes
D3D11 - Perfect colours, no banding, perfect black levels, BUT VRR does not work:
ontop
fullscreen=yes
d3d11-exclusive-fs=yes
volume=100
volume-max=100
vo=gpu-next
gpu-api=d3d11
hwdec=d3d11va
gpu-context=d3d11
video-sync=audio
fbo-format=rgba16hf
spirv-compiler=auto
drm-vrr-enabled=yes
d3d11-sync-interval=1
hr-seek-framedrop=no
target-peak=800
target-trc=pq
target-prim=bt.2020
vf=format:dolbyvision=no
dolbyvision=no
tone-mapping=spline
tone-mapping-mode=luma
gamut-mapping-mode=perceptual
tone-mapping-param=bt.2390
tone-mapping-max-boost=2.0
target-contrast=inf
hdr-compute-peak=yes
target-colorspace-hint=yes
Is there anything that can be done to the bottom code that would allow me to use VRR with D3D11?
Or to the Vulkan code at the top to fix crushed black levels?
I have tried everything but cannot find a solution.
hwdec=d3d11va *prevents VRR
drm-vrr-enabled=yes *prevents VRR or zero effect
d3d11-sync-interval=1 *prevents VRR or zero effect
Only these parameters prevents VRR tech, tested 4 times!
Drakko01 wrote:Hey! I encourage everyone to try the new driver hotfix of nvidia 551.46. They address various sttutering and micro-stuttering and it seems to me that this benefited svp. Let me know what you think.
This driver havent do anything for me, also this driver non-WHQL based, so therefore, as NVIDIA themselves stated, this driver is advised only to those who had real problems, no benefit at all, tested 1 full day, very tough tests, hard tasks!
dawkinscm wrote:RickyAstle98 wrote:Im already use mpv player since SVP4 Pro buy, I custom mpv for my preferences, so I cant get extra performance at all? I test d3d11va and RTX 4070 dies at 720@24>144 realtime playback, but d3d11va-copy didnt 
The reason you get better performance with d3d11va over d3d11va-copy is because with d3d11va SVP can't properly process the frames. d3d11va-copy copies frams to system ram where SVP has full access to them. So unless SVP has changed and somehow has direct access to GPU memory, you need to use copy. But you probably already know this and you were just joking.
I think there is some performance bonus for me, but unfortunately no, but yes d3d11va-copy better
dawkinscm wrote:RickyAstle98 wrote:MAG79 wrote:Gippy thanks for information.
In the past D3D11 Native value in LAV decoder was incompatible to SVP.
Need to check if it works correctly and SVP does its work fine (24 to 60 fps conversion).
---
I've checked. D3D11 works good with SVP.
In LAV decoder filter properties it indicates as Active Decoder: d3d11 cb direct
Does d3d11va-copy same thing D3D11 in mpv player? How to force D3D11 thru svp and mpv player, a bit confused, I want more performance!
For SVP, whichever decoder you use in mpv it needs to be a copy-back decoder where the frames are copied back to RAM, wtherwise SVP can't process them. d3d11va-copy is one of the better Windows copy-back decoders. If you want to start using mpv then use the SVP installer to install it then use the default mpv.conf that comes with SVP. It's not perfect but it will 100% work.
Im already use mpv player since SVP4 Pro buy, I custom mpv for my preferences, so I cant get extra performance at all? I test d3d11va and RTX 4070 dies at 720@24>144 realtime playback, but d3d11va-copy didnt
MAG79 wrote:Gippy thanks for information.
In the past D3D11 Native value in LAV decoder was incompatible to SVP.
Need to check if it works correctly and SVP does its work fine (24 to 60 fps conversion).
---
I've checked. D3D11 works good with SVP.
In LAV decoder filter properties it indicates as Active Decoder: d3d11 cb direct
Does d3d11va-copy same thing D3D11 in mpv player? How to force D3D11 thru svp and mpv player, a bit confused, I want more performance!
scb wrote:This is good to know, but what are the differences between the models? When should one pick a particular one? What are the different models best at? What are the quality differences?
Any new model is better intermediate quality, 4.9 models good enough for any type of content, movies, videos, anything, main difference between 4.14 and 4.4 ~1.6x performance drop
4.4 for any movies, but be careful, same patterns or objects can be doubled or tripled by error interpolation, 4.9 doesnt have this effect...
scb wrote:Guys,
I still haven't seen a full explanation about the different between the 'normal' and 'lite' models of the same versions? What situations should I pick the 'lite' version over the 'normal' version of a model?
V2 models are ~20% faster
Lite models are ~15% faster
4.14 lite is an exception but with RTX4070 outputs only 3% faster
aloola wrote:RickyAstle98 wrote:Really why SVP developers cant develop RIFE-SVP swapchain? It will suit for everyone, any type of content, any resolution, almost everything! My general idea about this quiet simple, SVP algorithm almost 10 times less resource-sensitive, but intermediate frames quality rises at high frame rate video source, 2K 24>144fps ~40-60 watts with 8-12 px grid and 21 shader, whats the problem RIFE@4K 2x and then SVP with higher pixel grid? Its going to be really smooth!
With this swapchain, 24>96@4K easy! Double peformance for 4K sources...
it's simple to do but not a good idea. it would produce a lot of artefacts.
Founded solution, LSFG can do interpolation!
https://store.steampowered.com/app/9930 … s_Scaling/
jimdogma7 wrote:roshin1401 wrote:Im noob to this software. While using TensorRT, a cmd popups doing compilation. Does this occur for every video file opened? or only once? And also please suggest best rife model for my gpu. Its a 3070 laptop.
I'm having the same problem and unfotunately this post wasn't responded to. I've put the .OONX files in the "models" and/or 'Rife" folders as instructed and yes, they do show up in the AI models pulldown in the SVP manager, but when I select it and try to run a video, a CMD box pops up and after a few seconds goes away. The video then plays but is not converted. Nor does the little green text in the lower left corner come up telling you what setting you're using, etc. Tres bizarre. Plus, if I try to transcode the video it doesn't work either. Says conversion failed.
It seems pretty straitforward, I don't what I'm doing wrong here. Please help!
Make sure you download latest TensorRT library because current SVP outdated library sometimes crash during TRT boost off/on optimizations, TRT 9.1/9.2 FIX this issues for any onnx models!
https://www.playbook.com/s/vadash/MFTzy … mG26wKy7zz
You need only rife folder from this archive!
aloola wrote:RickyAstle98 wrote:Really why SVP developers cant develop RIFE-SVP swapchain? It will suit for everyone, any type of content, any resolution, almost everything! My general idea about this quiet simple, SVP algorithm almost 10 times less resource-sensitive, but intermediate frames quality rises at high frame rate video source, 2K 24>144fps ~40-60 watts with 8-12 px grid and 21 shader, whats the problem RIFE@4K 2x and then SVP with higher pixel grid? Its going to be really smooth!
With this swapchain, 24>96@4K easy! Double peformance for 4K sources...
it's simple to do but not a good idea. it would produce a lot of artefacts.
A lot of artefacts? Wdym? I didnt say SVPtoRIFE video, do the opposite!
Example : 24fps video > RIFE 3x = 72fps then > SVP to any fps (monitor Hz)
VERDICT1 : Better than SVP algo 24fps to any fps imho (any resolution)
VERDICT2 : SVP works better with high frame rate sources (tested)
VERDICT3 : SVP developers, JUST DO IT!
This function can be used purely as a secondary performance boost (and/or for weaker GPUs and high resolution targets)
dlr5668 wrote:RickyAstle98 wrote:RAGEdemon wrote:537.58 eliminated the pan stutters for me.
My permanent setup is now 4K (3840x2160) with RIFE 4.4 with this driver. RIFE/SVP is set to double the frame rate. (with RTX 4090 @ 65% GPU usage)
For those who don't know, this makes all content (24fps) into 48fps. You MUST make a custom resolution to make your refresh rate 48Hz, or it will be stuttery because 48fps does not go into the default 60Hz/120Hz that the vast majority of monitors/TVs are.
4K 48fps @ 48Hz video is perfectly smooth. This is easy for SVP because it is just doubling the frame rate.
Other modes such as sync 24fps video to screen refresh, or fps x 2.5, or whole multiples, is really taxing on the GPU. An overclocked 4090 can't do it, even on the highest performing RIFE (4.4/4.6).
Really why SVP developers cant develop RIFE-SVP swapchain? It will suit for everyone, any type of content, any resolution, almost everything! My general idea about this quiet simple, SVP algorithm almost 10 times less resource-sensitive, but intermediate frames quality rises at high frame rate video source, 2K 24>144fps ~40-60 watts with 8-12 px grid and 21 shader, whats the problem RIFE@4K 2x and then SVP with higher pixel grid? Its going to be really smooth!
With this swapchain, 24>96@4K easy! Double peformance for 4K sources...
You can probably modify base scripts (start at "C:\Program Files (x86)\SVP 4\script\base.py") and chain rife+regular
Idk how sorry, but I will start think about it!
RAGEdemon wrote:537.58 eliminated the pan stutters for me.
My permanent setup is now 4K (3840x2160) with RIFE 4.4 with this driver. RIFE/SVP is set to double the frame rate. (with RTX 4090 @ 65% GPU usage)
For those who don't know, this makes all content (24fps) into 48fps. You MUST make a custom resolution to make your refresh rate 48Hz, or it will be stuttery because 48fps does not go into the default 60Hz/120Hz that the vast majority of monitors/TVs are.
4K 48fps @ 48Hz video is perfectly smooth. This is easy for SVP because it is just doubling the frame rate.
Other modes such as sync 24fps video to screen refresh, or fps x 2.5, or whole multiples, is really taxing on the GPU. An overclocked 4090 can't do it, even on the highest performing RIFE (4.4/4.6).
Really why SVP developers cant develop RIFE-SVP swapchain? It will suit for everyone, any type of content, any resolution, almost everything! My general idea about this quiet simple, SVP algorithm almost 10 times less resource-sensitive, but intermediate frames quality rises at high frame rate video source, 2K 24>144fps ~40-60 watts with 8-12 px grid and 21 shader, whats the problem RIFE@4K 2x and then SVP with higher pixel grid? Its going to be really smooth!
With this swapchain, 24>96@4K easy! Double peformance for 4K sources...
Blackfyre wrote:Question, regarding RIFE I assume the 4070, 4070 Super, 4070 Ti Super, etc are all better than the RTX 3090 with regards to RIFE yes?
I'm asking because I can sell my RTX 3090 used for around $1100 in Australia, and just buy an RTX 4070 Super or 4070 Ti Super for example on special for around the same price.
3090 can no longer handle 4K at 2x with the latest RIFE models sometimes, if the 4070 and higher can handle it, I am willing to do the switch, despite having half the vRAM.
Tested! My 4070 provide 2x 4.14v2 (and lite) under 3200x1824opt (3520x1984opt unstable) At exact 4K nah, so I think 4070Ti close enough for latest models!
Windows 10 fresh (3 month) and NVIDIA driver 546.17 without mods...
dawkinscm wrote:cemaydnlar wrote:v1.14 v2 is the smoothest version of them all for me somehow...
v1.14 v2 is so smooth that I made sure to back everything up
cemaydnlar wrote:It gave a weird error while caching but is somehow working without doing that caching cmd again.2nd i don't see any improvement for gpu usage. İt's the same as v.14. Is it normal?
If the messages come after the "profiling" statement then my understanding is that the warnings are part of the profiling process. As for the performance, that might depend on you GPU memory bandwidth. The Nvidia 3070Ti and above start at 600GB/s memory bandwidth. The 4070Ti has the same memory bandwidth as a 4080 at 700GB/s while the 3080 has even more than the 4080. Below these cards I don't know what the performance will be like with 4.14 "lite".
Talking about 4.14v2 model? Because you say 1.14 v2 too much, slightly confused...
aloola wrote:Output 8000 frames in 34.47 seconds (232.12 fps) 4.0
Output 8000 frames in 34.14 seconds (234.35 fps) 4.4
Output 8000 frames in 33.90 seconds (235.99 fps) 4.6
Output 8000 frames in 45.65 seconds (175.26 fps) 4.7
Output 8000 frames in 46.70 seconds (171.29 fps) 4.9
Output 8000 frames in 63.21 seconds (126.56 fps) 4.11
Output 8000 frames in 63.36 seconds (126.26 fps) 4.12
Output 8000 frames in 50.69 seconds (157.83 fps) 4.12 lite
Output 8000 frames in 63.61 seconds (125.76 fps) 4.13
Output 8000 frames in 50.31 seconds (159.01 fps) 4.13 lite
Output 8000 frames in 64.95 seconds (123.18 fps) 4.14
Output 8000 frames in 65.58 seconds (121.99 fps) 4.14 lite
4.14 lite is the most resource-heavy, dunno about the quality though.
RTX 4070 output 2880 frames 1280x736opt TRT 9.1
60>120 ~304fps (4.4v2)
60>120 ~284fps (4.6v2)
60>120 ~260fps (4.7v2)
60>120 ~250fps (4.9v2)
60>120 ~210fps (4.12v2)
60>120 ~200fps (4.14v2)
dawkinscm wrote:So far v14.4 lite uses more GPU than the non lite version. I thought it was may be because of the changes I made but I reverted back to SVP standard configs and it's the same. Maybe there is an issue they need to fix.
Update:
It uses about the same amount of GPU resources but it works in a different way which makes it more dependant on memory bandwidth than GPU resources. I've read up a little on the new methodology and it is interesting the result is that the "lite" version may not be as "lite" before.
IDK
dlr5668 wrote:RickyAstle98 wrote:dlr5668 wrote:v14 runtime + 4.14 v2 is perfection. Really nice update
Without +10% performance boost and +5% quality - not really nice update at all
Then please upload your v14 runtime and all folders - I will test
Cant you repeat steps? Here we go https://www.playbook.com/s/vadash/MFTzy … 49DgFraywF
Thank you, but as I say there is no significant performance boost, but very fast timing cache building time, under 1 minute for any models!
HD 24>120 insta 150W usage RTX 4070 (4.14v2) quite loud hehe but works!
Anyway thank you for uploading whole SVP4 folder, appreciate!
dlr5668 wrote:v14 runtime + 4.14 v2 is perfection. Really nice update
Without +10% performance boost and +5% quality - not really nice update at all
Then please upload your v14 runtime and all folders - I will test
aloola wrote:RickyAstle98 wrote:Sometimes force exact value prevent RIFE working, with new models too (mpv_cfr argument can also kick RIFE from workload) but I only use X factors because I have GSync monitor tech, I have smoothness all the time!
Force exact value = issues
Fixed frame rate = issues
Frame rate factor = no issues
are you using the newest vsmlrt.py?
then it won't work with fraction numbers.
https://github.com/AmusementClub/vs-mlr … 1847153063
It work with fraction numbers (issues as jiggling with these settings), nope, I didnt use newest .py but I already modify it myself!
Blackfyre wrote:RickyAstle98 wrote:zxcvbnm821 wrote:I have the same problem here too
tried using v14.4, but the problem was not fixed 
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).
Sometimes force exact value prevent RIFE working, with new models too (mpv_cfr argument can also kick RIFE from workload) but I only use X factors because I have GSync monitor tech, I have smoothness all the time!
Force exact value = issues
Fixed frame rate = issues
Frame rate factor = no issues
zxcvbnm821 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 have the same problem here too
tried using v14.4, but the problem was not fixed
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!
barryma wrote:Is the bigger the number of the scene change threshold, the better?And why is my maximum number only 15%?
Higher values = more smoothness and scene transitions become funny
Lower values = scene transitions become duplicated frames not funny
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 jiggling at 4K or 2K downscaled or not, in my case this effect only present when you put factors like 2.5x or halfs, odd values (tried all same versions)
Try to do without downscale option too!
Chainik wrote:RIFE only works for 32-px aligned frame size, so we have two options (for example for 1920*1080 frame):
1. add 8px height black line, interpolate, cut away 8px line
2. cut 24px line --> interpolate 1920*1056 frame
SVP now does (1) for frame height <= 2560 and (2) for >2560.
If you want always do (1) then for now you can edit SVP 4\script\generate.js, line 458, change it to:
Thanks!
Xenocyde wrote:RickyAstle98 wrote:Wait me here, I will edit my message in next 5 minutes! Check plz!!!
So you need to put 4.13.onnx into v2 folder where SVP located
https://ibb.co.com/RYCzGRh
If your 1st timing cache build failed, you need enable TRT boost option, 4.7-4.9v2 cant stable buld timing cache for general optimization... Maybe 4.13v2 too? Me too cant build timing cache without TRT enabled!
Yes I tested 60Hz and only CPU usage change a little, anyway I have modern 165Hz GSync monitor, any FPS factor, smooth always!
Ya so, for me, 4.13 V2 only works when I activate performance boost. Seems to be a bit more resource-intensive compared to non-V2, 50-60% 3D utilization. Need to specify that I capped GPU performance to 60 fps so it's using only 150 W instead of 300+ W, that's why it looks like my performance is not scaling well compared to your 144 Hz performance. I've been trying to get a new OLED screen for some time now, but every damned TV model up until now came with some annoying issue... maybe 2024 is the year I get a 55-inch OLED so I can use the GPU at its fullest.
Hehe))
Posts found: 176 to 200 of 213