You won't be able to use my script without modifying all the paths etc. to your system. Also, you would need to adjust the reserved memory.
I would recommend using my script, better learn how to use Hybrid by reproducing the script.
Use the Vapoursynth Script View to compare what you have against my script.
Cu Selur
Hi,
Thanks again for your help! Honestly, the TIVTC + (edge masked) BasicVSR++ + CCD combo gives the best result I’ve seen so far.
The problem is, I have no idea how to reproduce it myself because I don’t know how to use the script. So I was wondering: is it possible to do all of this directly in Hybrid — like set up TIVTC + (edge masked) BasicVSR++ + CCD — without going through an external script?
Thanks in advance for your advice!
Sure, assuming Hybrid is installed with the torch-add-on (required for BasicVSR++).
Here's how to I configured TIVTC + (edge masked) BasicVSR++ + CCD:
- started Hybrid
- loaded Source
- enabled "Filtering->(De-)Interlace->Telecine->Overwrite input scan type to" and set it to "telecine" (optionally set Deinterlacer to 'QTGMC (preset="fast")' and enabled "GPU")
- enabled "Filtering->Vapoursynth->Misc->UI->Show 'Masked'-controls" (needed for the masking)
- enabled "Filtering->Vapoursynth->Other->BasicVSR++" (also enabled FP16; only do this if you GPU supports it)
- enabled "Filtering->Vapoursynth->Other->BasicVSR++->Masked", set "Mask" to "EdgeMask (Bloated)" and set "Expand/Shrink" to 3
- enabled "Filtering->Vapoursynth->Denoise->CCD" and set "Threshold" to 60
Side note: The scripts I post are always created with Hybrid and if I add I custom code snipped I also post about it. It's usually better to reproduce the script manually, since the scripts I post take inpt account the amount of VRAM and RAM my system has. And doing it yourself is usually better, since it allows you to adjust the parameters to your liking.
Cu Selur
Thanks so much for the detailed explanation! I really appreciate you taking the time to write it all out.
I understand now how you configured TIVTC + (edge masked) BasicVSR++ + CCD in Hybrid, and I see why reproducing it manually is better — especially considering VRAM and RAM differences, and to tweak parameters to my liking.
I’ll give it a try on my system tomorrow, and if I run into any issues, I hope I can still count on you for your precious advice ^^
Thanks again for sharing your setup and for all the guidance!
I'll be afk. tomorrow for most of the day and will likely not be back home and online before this time tomorrow.
I'll check if you posted something tomorrow when I'm back.
Cu Selur
Hi again
,
I tried your configuration, but I’m getting this error:
"Aborting '2026-04-08@10_11_23_2710_02_video' due to: One or multiple files were not created properly!"
This happens specifically when I encode using VapourSynth.
My system configuration is:
CPU: Intel Core i7-4790K @ 4.00GHz
GPU: NVIDIA GeForce RTX 3070 Ti
RAM: 32 GB DDR3
So performance shouldn’t be an issue.
Also, I can’t find the option:
BasicVSR++ -> Masked, then set Mask to "EdgeMask (Bloated)" and Expand/Shrink to 3.
I’ve attached a screenshot to show what I see on my side.[
attachment=3546]
Do you have any idea what could be causing this or what I should check first?
Thanks!
About the Crash: deinstall Hybrid completely, Install latest dev with latest torch addon. Create a DebugOutput If that does not fix the problem.
About masked: you forgot to enable to show the masked controls or you did Not scroll to the right
Cu Selur
Yes! 😄🎉 I finally got it working perfectly! Thanks again for your help 🙏💻
However, the encoding is extremely slow on my setup (i7-4790K + RTX 3070 Ti + 32 GB RAM). 😅
Are there any options I can check or specific settings to speed up the encoding without sacrificing too much quality? 🚀⚡
Also, I’m noticing some rainbow arc artifacts in the output 🌈. Is there any way to fix this, or is it something that can’t be resolved?
Thanks again for all your advice! 💛
Don't think there is anything you can dadfitionally do to speed things up.
Aside from upgrading the GPU.
About the rainbows: using BasicVSR++ without the masking might help, but will probably remive sone details. Other that that, I'll need a sample of the source.
Cu Selur