2,151 Reply by Drakko01

  • Thanks: 53
  • Thanks for the post: 2

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

narkohol wrote:
Chainik wrote:

> Install VapourSynth R70

yeah, great! but why? big_smile

To pass time with placebos until you come up with a new real improvement? wink


Chainik wrote:

everyone following these guides must understand clearly that this will improve absolutely NOTHING

Wasn't so sure about that, but no harm in testing when reverting the change is as easy as copying back a 200MB folder...

Adding to that , and don't even know if it has something to do but a weird thing happen. My display its a samsung tv that it has compatible Gsync and VRR that never engage on playback, odd as it sounds now on MPC Be for some reason sometimes engage VRR and you see a indiscutible improvement on playback.

Post's attachments

Screenshot_20250105_222244_Gallery.jpg, 34.91 kb, 853 x 266
Screenshot_20250105_222244_Gallery.jpg 34.91 kb, 36 downloads since 2025-01-05 

2,152 Reply by Ironclad (edited by Ironclad 09-01-2025 04:04:29)

  • Thanks: 4
  • Thanks for the post: 1

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

RAGEdemon wrote:

This is probably a stupid question however, with AI software, I have noticed that on a 4090, RT is faster than CUDA but not by much.

A 4090 has 128 RT cores and >16000 coda cores.

Is it true that with SVP, when RT cores are being used, CUDA cores are not really being used, and vice versa?

In that case, half the power of the GPU is wasted.

Would it be possible to implement both TensorRT AND CUDA simultaneously to double the performance?

RT cores aren't actually the part doing the AI stuff, it's the Tensor Cores. Also look at the power consumption of your gpu while using RIFE, and you'll see there isn't as much headroom as you're imagining. I do think MIGraphX support should be looked into. Give Radeon users more options.

2,153 Reply by flowreen91 (edited by flowreen91 11-01-2025 06:42:48)

  • Thanks: 109
  • Thanks for the post: 1

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

Guys, can anyone check if they made any significant improvements since last time by testing the 20x frame generation on 4k videos?
https://www.reddit.com/r/nvidia/comment … n_30_with/

2,154 Reply by RickyAstle98 (edited by RickyAstle98 11-01-2025 12:24:31)

  • Thanks: 78
  • Thanks for the post: 2

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

flowreen91 wrote:

Guys, can anyone check if they made any significant improvements since last time by testing the 20x frame generation on 4k videos?
https://www.reddit.com/r/nvidia/comment … n_30_with/

LSFG3 vs LSFG2 on any multipliers - smoother and better! Quality increased, same fast transitions now look more accurate, not just a fluid!

Im using RIFE x3 + LSFG x2 for 24>144 movies conversion, usually GPU usage looks like RIFE x4 with 6x real FLUIDITY rate!

On videos with LSFG3 only, looks pretty normal, conversion from 24FPS is less accurate, but smooth enough!

2,155 Reply by Vequa

  • Thanks: 3
  • Thanks for the post: 1

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

Can someone please post updated instructions on installing and using V2 rife models?
I tried to get it done based on this thread but all the significant information is either from 2023 with dead links, doesn't work or is impossible to understand for us common folk.
I attempted to add the "4.25 heavy" v2 model with and without replacing "vsmlrt.py". It appears in the "AI model" dropdown menu but no matter what I do I can't get it to work.

  • Thanks: 78
  • Thanks for the post: 1

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

Vequa wrote:

Can someone please post updated instructions on installing and using V2 rife models?
I tried to get it done based on this thread but all the significant information is either from 2023 with dead links, doesn't work or is impossible to understand for us common folk.
I attempted to add the "4.25 heavy" v2 model with and without replacing "vsmlrt.py". It appears in the "AI model" dropdown menu but no matter what I do I can't get it to work.

Rename the 4.25 heavy to 4.30 for example, worked for me without replacing vsmlrt.py a long time ago!

2,157 Reply by raider10

  • Thanks: 2
  • Thanks for the post: 1

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

RickyAstle98 wrote:
flowreen91 wrote:

Guys, can anyone check if they made any significant improvements since last time by testing the 20x frame generation on 4k videos?
https://www.reddit.com/r/nvidia/comment … n_30_with/

LSFG3 vs LSFG2 on any multipliers - smoother and better! Quality increased, same fast transitions now look more accurate, not just a fluid!

Im using RIFE x3 + LSFG x2 for 24>144 movies conversion, usually GPU usage looks like RIFE x4 with 6x real FLUIDITY rate!

On videos with LSFG3 only, looks pretty normal, conversion from 24FPS is less accurate, but smooth enough!


Do you do it with lossless scaling?

2,158 Reply by Vequa

  • Thanks: 3
  • Thanks for the post: 1

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

RickyAstle98 wrote:

Rename the 4.25 heavy to 4.30 for example, worked for me without replacing vsmlrt.py a long time ago!

I have no idea why or how but this actually fixed it. Thanks!

2,159 Reply by RickyAstle98 (edited by RickyAstle98 17-01-2025 14:01:27)

  • Thanks: 78
  • Thanks for the post: 2

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

raider10 wrote:
RickyAstle98 wrote:
flowreen91 wrote:

Guys, can anyone check if they made any significant improvements since last time by testing the 20x frame generation on 4k videos?
https://www.reddit.com/r/nvidia/comment … n_30_with/

LSFG3 vs LSFG2 on any multipliers - smoother and better! Quality increased, same fast transitions now look more accurate, not just a fluid!

Im using RIFE x3 + LSFG x2 for 24>144 movies conversion, usually GPU usage looks like RIFE x4 with 6x real FLUIDITY rate!

On videos with LSFG3 only, looks pretty normal, conversion from 24FPS is less accurate, but smooth enough!


Do you do it with lossless scaling?

As you can see, I did RIFE x3 (1st action) on movie, then (LSFG3 x2) ontop!

The LSFG3 x3 alone: looks choppier and less accurate than RIFE x3 mode, takes 8 times less resources!

Thats better solution for 4K videos for example! For HFR monitor owners SPECIFICALLY!

Usage: when your GPU able to do 3x RIFE with 4K input, theres enough GPU headroom to RIFE x2 + LSFG3 x3 ontop without 100% usage!

  • Thanks: 210
  • Thanks for the post: 1

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

RickyAstle98 wrote:
raider10 wrote:
RickyAstle98 wrote:

LSFG3 vs LSFG2 on any multipliers - smoother and better! Quality increased, same fast transitions now look more accurate, not just a fluid!

Im using RIFE x3 + LSFG x2 for 24>144 movies conversion, usually GPU usage looks like RIFE x4 with 6x real FLUIDITY rate!

On videos with LSFG3 only, looks pretty normal, conversion from 24FPS is less accurate, but smooth enough!


Do you do it with lossless scaling?

As you can see, I did RIFE x3 (1st action) on movie, then (LSFG3 x2) ontop!
I'm probably missing something obvious but how are you applying it? I only see a Steam implementation, nothing else.

The LSFG3 x3 alone: looks choppier and less accurate than RIFE x3 mode, takes 8 times less resources!

Thats better solution for 4K videos for example! For HFR monitor owners SPECIFICALLY! Usage: when your GPU able to do 3x RIFE with 4K input, theres enough GPU headroom to RIFE x2 + LSFG3 x3 ontop without 100% usage!

2,161 Reply by RickyAstle98 (edited by RickyAstle98 17-01-2025 14:25:13)

  • Thanks: 78

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

dawkinscm wrote:
RickyAstle98 wrote:
raider10 wrote:

Do you do it with lossless scaling?

As you can see, I did RIFE x3 (1st action) on movie, then (LSFG3 x2) ontop!
I'm probably missing something obvious but how are you applying it? I only see a Steam implementation, nothing else.

The LSFG3 x3 alone: looks choppier and less accurate than RIFE x3 mode, takes 8 times less resources!

Thats better solution for 4K videos for example! For HFR monitor owners SPECIFICALLY!

Usage: when your GPU able to do 3x RIFE with 4K input, theres enough GPU headroom to RIFE x2 + LSFG3 x3 ontop without 100% usage!

For 144Hz users (1080p output / 4070 power levels): RIFE x3 + LSFG x2 = looks very smooth (24>144 / GPU usage 70% ~120W)

For 240Hz users (1080p output / 4070 power levels): RIFE x2.5 + LSFG3 x4 = looks very smooth and accurate (24>240 / GPU usage 80% ~130W)

For 360Hz users (1080p output / 4070 power levels): RIFE x3 + LSFG3 x5 = looks very smooth and accurate (24>360)

All tests done with 4.4v2 model (old) for GPU usage local measurement, newest models almost 2 times heavier, the multiplier decision, gentlemens, is up to you!

  • Thanks: 210
  • Thanks for the post: 1

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

RickyAstle98 wrote:
dawkinscm wrote:
RickyAstle98 wrote:

As you can see, I did RIFE x3 (1st action) on movie, then (LSFG3 x2) ontop!
I'm probably missing something obvious but how are you applying it? I only see a Steam implementation, nothing else.

The LSFG3 x3 alone: looks choppier and less accurate than RIFE x3 mode, takes 8 times less resources!

Thats better solution for 4K videos for example! For HFR monitor owners SPECIFICALLY!

Usage: when your GPU able to do 3x RIFE with 4K input, theres enough GPU headroom to RIFE x2 + LSFG3 x3 ontop without 100% usage!

For 144Hz users (1080p output / 4070 power levels): RIFE x3 + LSFG x2 = looks very smooth (24>144 / GPU usage 70% ~120W)

For 240Hz users (1080p output / 4070 power levels): RIFE x2.5 + LSFG3 x4 = looks very smooth and accurate (24>240 / GPU usage 80% ~130W)

For 360Hz users (1080p output / 4070 power levels): RIFE x3 + LSFG3 x5 = looks very smooth and accurate (24>360)

All tests done with 4.4v2 model (old) for GPU usage local measurement, newest models almost 2 times heavier, the multiplier decision, gentlemens, is up to you!

Part of my message got lost. I was asking how are you actually running LSFG? I have not seen anything except for a steam package.

2,163 Reply by RickyAstle98 (edited by RickyAstle98 17-01-2025 14:42:33)

  • Thanks: 78
  • Thanks for the post: 1

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

Part of my message got lost. I was asking how are you actually running LSFG? I have not seen anything except for a steam package.

I will upload quick video YT guide, I cant record generated frames from LSFG overlay!

https://youtu.be/8cIXy2IuMAs

  • Thanks: 210
  • Thanks for the post: 1

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

RickyAstle98 wrote:

Part of my message got lost. I was asking how are you actually running LSFG? I have not seen anything except for a steam package.

I will upload quick video YT guide, I cant record generated frames from LSFG overlay!

https://youtu.be/8cIXy2IuMAs

Oh I see. It's part of the new Nvidia app.

2,165 Reply by RickyAstle98 (edited by RickyAstle98 17-01-2025 14:59:04)

  • Thanks: 78

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

dawkinscm wrote:
RickyAstle98 wrote:

Part of my message got lost. I was asking how are you actually running LSFG? I have not seen anything except for a steam package.

I will upload quick video YT guide, I cant record generated frames from LSFG overlay!

https://youtu.be/8cIXy2IuMAs

Oh I see. It's part of the new Nvidia app.

Third-party software, not part of new NVIDIA App actually!

2,166 Reply by Sach (edited by Sach 21-01-2025 17:50:59)

  • Sach
  • Novice
  • Offline
  • Thanks: 6
  • Thanks for the post: 1

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

Hi everyone
Do you know any way to use
MultiPassDedup https://github.com/routineLife1/MultiPassDedup
Or
Gmfss fortuna https://github.com/98mxr/GMFSS_Fortuna
With SVP through MPC-HC OR BE when VapouSynth is used in real time with Rife TRT
THANKSSSSSSS

2,167 Reply by dlr5668 (edited by dlr5668 21-01-2025 18:02:43)

  • Thanks: 233
  • Thanks for the post: 3

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

Sach wrote:

Hi everyone
Do you know any way to use
MultiPassDedup https://github.com/routineLife1/MultiPassDedup
Or
Gmfss fortuna https://github.com/98mxr/GMFSS_Fortuna
With SVP through MPC-HC OR BE when VapouSynth is used in real time with Rife TRT
THANKSSSSSSS

its 1/10 realtime without rife interpolation. For now use RIFE x2 and its good enough. If you have 144+ screen slap LSFG x3 for x6 total

2,168 Reply by odedg

  • Thanks: 1
  • Thanks for the post: 1

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

Hi,
What is expected uplift in RIFE from RTX 4090 to 5090 ?
Is it about 30% or more ?

  • Thanks: 210
  • Thanks for the post: 1

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

odedg wrote:

Hi,
What is expected uplift in RIFE from RTX 4090 to 5090 ?
Is it about 30% or more ?

The 30% number is generally for gaming, but 30% does track with the increase of Tensor cores and other hardware. So it could be 30%, it could be more, it could be less. The only certainty is that it will use more power.

2,170 Reply by narkohol (edited by narkohol 25-01-2025 19:17:40)

  • Thanks: 18
  • Thanks for the post: 1

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

vs-mlrt has been updated to 15.8 at https://github.com/AmusementClub/vs-mlrt/releases/

Besides support for RTX 50-series, are there any improvements we can expect on current GPUs by using this new file?

2,171 Reply by dawkinscm (edited by dawkinscm 26-01-2025 08:26:39)

  • Thanks: 210
  • Thanks for the post: 2

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

narkohol wrote:

vs-mlrt has been updated to 15.8 at https://github.com/AmusementClub/vs-mlrt/releases/

Besides support for RTX 50-series, are there any improvements we can expect on current GPUs by using this new file?

Testing might reveal something else, but there's nothing obvious in the change log except possible improvements for lower specced GPUs below RTX4070ti.

2,172 Reply by RickyAstle98 (edited by RickyAstle98 26-01-2025 09:35:12)

  • Thanks: 78
  • Thanks for the post: 3

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

narkohol wrote:

vs-mlrt has been updated to 15.8 at https://github.com/AmusementClub/vs-mlrt/releases/

Besides support for RTX 50-series, are there any improvements we can expect on current GPUs by using this new file?

No improvement in realtime performance across all models, almost the same transcoding speed (0.8% slower), but new heavy models runs 5% faster now!

RTX 4070

2,173 Reply by dawkinscm (edited by dawkinscm 26-01-2025 12:21:40)

  • Thanks: 210
  • Thanks for the post: 1

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

RickyAstle98 wrote:
narkohol wrote:

vs-mlrt has been updated to 15.8 at https://github.com/AmusementClub/vs-mlrt/releases/

Besides support for RTX 50-series, are there any improvements we can expect on current GPUs by using this new file?

No improvement in realtime performance across all models, almost the same transcoding speed (0.8% slower), but new heavy models runs 5% faster now!

RTX 4070

Going back to Lossless Scaling. The only app I see is a Steam Application which I wouldn't want to run. Where is the non steam version?

Forget it. I think I found it thanks smile

2,174 Reply by Hottea

  • Thanks: 4
  • Thanks for the post: 1

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

Do you guys think something like 5090 finally get us above 4k 60hz with newest models or there are another bottlenecks like DDR bandwidth like some ppl have been saying?

2,175 Reply by scb

  • scb
  • Member
  • Offline
  • Thanks: 9
  • Thanks for the post: 1

Re: New RIFE filter - 3x faster AI interpolation possible in SVP!!!

What is the lossless scaling algorithm vs SVP vs SVP RIFE, all like?