I checked the audio bitrate again but in opposite to the promise of the team the audio quality is still at terrible low 96 kbit/sec. 
2 Likes
Don’t think it matters but sure.
RTX 3080, Ryzen 5900x, 2TB NVME M.2 SSD, 4x16GB 3600Mhz RAM
Hello.
Hey what’s happened with GAIA-HQ AI model? Previous version i got good results with it. Newer version it’s worse.
This update completely broke Artemis Low Quality 200% for me. I run out of VRAM. At 100% it just crashes after a few minutes.
RTX3070 FE
3700x
32GB 3600MHz
Windows 10
Everything up to date
Are previous versions available for download somewhere?
Actually, no. Wrong. Gaming drivers are newer and updated much more frequently. Studio drivers are older versions that are considered more stable. There’s nothing to “turn on and off”. Not sure where you got this info, but it ain’t right.
3 Likes
I cannot help yoo since I do not use QTGMC, I have just got deinterlaced a few clips by others. Sorry. I wish I am not familiar with it, what a pity. Maybe in future I shall try to get to know how to use it.
did you really enable resizable bar? turn on it in mobo bios is step 1.
step 2 you have to flash your 3090 vbios.
if you don’t flash your vbios (3090 bios) resizable bar is still disable.
you understand that right?
you can easily check if resizable bar is enabled by going into 3090 control panel.
1 Like
All ready done. It shows Yes in Nvidia control panel. Still using it without any problem.
Try using v.2.1.1 and see if it fixes your issue. When running VEAI, the power consumption of GPU goes up and down, creating spikes. I recommend undervoltage your GPU for stability, speed and lower power consumption.
For anyone reading this with same problem, this won’t resolve the issue.
Can you post a clearer screenshot of your GPU power consumption while running VEAI? If it goes up and down, then it’s OK. CPU doesn’t matter. VEAI doesn’t run 100% CPU and GPU usage, because it uses tensore cores.
Is there anybody who uses Neat Video or DVO Clarity?
This is the script I’m using. It uses a 2-pass technique initially described to me by @dbasthemer in this thread: My Star Trek DVD Upscale Project - #3 by dbasthemer which is nothing short of miraculous compared to converting VFR sources to 24fps. imho. It then uses a QTGMC repair process based on that described by Joel Hruska’s “start trek DS9” project. It also converts to 12 bit (for better resolution during processing) and uses checkmate to help with dot crawl. There’s much discussion about what others have been doing in that other star trek thread.
I’ve been working on some tooling to automate all the tools involved in this process from end to end. It’s probably not quite ready for a layman to try but this is an easy way now to point you to my avisynth scripts:
To use this you’d have to substitute real file names for the variables at the top commented as “API” - this gets used by an automated workflow that rewrites it with the right file names for each usage.
1 Like
As of late, studio drivers are updated just as often.
I’ve used Neat Video a bit, usually to remove dirt or grain from older films.
In 2021, there have been six Game Ready driver releases, and four Studio driver releases. It’s about once a month or once every six weeks. This is pretty logical because the Game Ready driver eventually becomes the Studio driver unless problems are found with it.
I like what this script does to the source for Dione output. I can see in the process workflow you have Artemis Medium V13 currently as Topaz Method - try Dione DV 1 in Progressive mode. AM13 on the left Dione DV1 on the Right
I am not sure I I will carry on with the VFR encoding at the end so I might tweak it, at least for now, and it didn’t clean up the title sequence as much as my oriignal script did, but I am impressed enough with most of the faces - foreground and background - that I will do a whole episode to compare.
1 Like
No question this is a very compelling comparison. I’m going to give it a go. To my eyes the Dione version retains significantly more detail. I’m going to give a try on Stargate SG1 too and see how it stacks up to Gaia HQ; if it’s similar results but faster I’d be glad to switch.
I’m fully planning to redo everything whenever I get the automation part of all this finished anyway, I’ve already made so many changes since I started this. I’ve basically been tinkering/tweaking/re-watching 
from S1E01, I’m about halfway through S2 right now in rewatching and a few episodes farther along in recoding. So I have no problem putting everything back through the pipeline when I find a better way.
@virtutis So which model exactly are you using? Do I need to tweak the JSON for it to behave correctly for progressive sources?
