Every source is different but for one tv recording that had lots of colorful noise all over it, this worked well: Enhanced using nyx-3; mode: relative to auto; reduce noise at -70; reduce large grain at -70; revert compression at 0; sharpen at -20; dehalo at -30; anti-alias/deblur at 0; and recover original detail at 20. No upscaling on that project. It was just a cleaning. I think I decombed the file first in another application.
Thank You much appreciated
Every source is indeed different my TV 2008 SD such as X factor are fine using protheus v4
But the very ugly VHS tapes I got from 1993 were hardware encoded but not correct so I use Interlaced to progressive with extreme values to keep the details using Iris v3 Gaia and Aion
Update ;
sadly enough for my VHS digitized tapes the Nyxv3 settings inserted more artificial faces. It look as if Nyxv3 is not made for my sources…
In my endover to find something that did improve my topaz lh q iris v2 generated 55 mbit video i came along Nero AI Video upscaler and I found out that this did the trick for me I feeded my topaz video and tried all 4 models. There is a face model that got introduced in december 2024 and enhanced it magically in terms of clarity and detail. It also optimizes the settings based on your Nvidia card and uses 3D Video encode and Video decode simultanuously again this is geared to VHS digitixed material while topaz is geared towards native broadcast TV digitized content after 2008, but can restore with progressive interlaced with hardware encoded vhs tapes clarity details deblurring, but a treatment with Nero will give it another boost and very fast. Nero is german not Chinese so please do not blame me with video proc pro is Chinese hence not good it is pretty good for it;s main purpose but very slow very vry slow in video enhancing and not good for VHS digitized sources. It is also striking that only Nero AI Video Upscaler has been introducing new enhancing models that do work, at least for VHS digitized sources it is by the way not really cheap permanent licence is 180 Euro
Is there a reason why several of my upscaled videos, which have included an audio track in the source, have exported without/missing an audio track?
My Codec settings have a “Copy” for “Audio Mode”.
Very Good Result ![]()
All the Rainbowing & Chroma Noise is gone. VHS Recording ?
ok you compensate the bad thick lines parts NYX does with negative values of Deblur and Sharpen. I thought oing into negative with sharpen levler has no effect, I’ll have to try it out now, I was probably wrong.
Unfortunately further testing suggests the trick to prevent remapping of colors (to avoid red shifting to orange when starting with a BT.601 source) does not in fact work.
I have this same issue … and if the audio track does exist, it is corrupted and only half the audio track plays, and the other half of the audio track has no audio. Even though i have Video AI set to COPY audio,
I then have to REMUX the original audio back in with MKVmergeGUI. But this has been happening since v5 was released.
Very frustrating!
Guys, please bring back the loop preview. This is ridiculous. Why would you remove something so useful?
Actually, there is an installer already, just not announced.
If it hasn’t been announced then links to it must not be posted here.
Hello,
On the homepage, the download link is broken. This is redirects to the url,
- ${downloadDomain}/TopazVideoAI/6.0.0/TopazVideoAI-6.0.1.dmg
and it is wrong one. I just want to let you know the error.
Thanks,
Hello, the download url for Topaz Video AI 6 is broken, it points to the 6.0.1.dmg but the path is 6.0.0. It cannot be downloaded and got access denied error. I just want to let you know this. Thanks,
Why are you wasting time comparing an unreleased version with a beta version?
Aren’t there enough bugs in the released versions already?
Bugs in the beta version should be reported in the beta section of this forum.
Unreleased versions should only be available to staff members.
And:
In November 2024 @lhkjacky Jacky said:
"Both Topaz staff and developers have reiterated their responses numerous times. A placeholder file has been included there solely for testing purposes. Developers have consistently advised against downloading the unreleased version, emphasizing its lack of finalization and readiness for release. It could potentially be an outdated file renamed for testing update notifications, thus possibly harboring multiple bugs or unforeseen issues.
The inclination of many individuals to take the risk of installing unreleased software perplexes me. Those interested in assisting with debugging the program should consider joining the beta testing group and experimenting with those beta and alpha versions instead."
That is not an official release!
Stop posting fake news.
The ONLY place to look for new releases is right at the top of this page.
It says “Releases”.
If it’s not there, then it’s not official.
Why would you even bother installing something that isn’t finished yet?
Well…you could say that about the official releases too, I guess.
I believe you. I believe that this file exists and when you run the software, it will tell you that its version is 6.0.2.
But it hasn’t been officially released and maybe when the final v6.0.2 comes out, it might be different from the version you are talking about. Maybe there will be some last minute changes.
People who already installed this unofficial version may report bugs that no longer exist in the final release. And nobody understands why they are having problems no one else has.
Why not just wait for the official release. I’m sure you will still be able to report plenty of issues then. And then there will also be a Topaz Video AI 6.0.2 section in this forum to do so.
Okay, let’s keep paying to have buggy versions with each new release.
They probably realized their mistake. ![]()
It’s worse than I described. It’s not that the program calculates ETA incorrectly - it doesn’t work properly at all!
After exiting the program for the first time (without pressing pause) and restarting the program, the processing action was automatically resumed (I like it). After a few days of processing with pausing for night, I decided that something was taking too long and should have ended a long time ago, so I looked at his previous work and to my surprise it turned out that apart from the first part, all subsequent parts start with the same scene!!! So how was program supposed to finish his job???
I lost time and I lost money for energy – this is not eco.
I am reporting this to the authors of the program as a serious error.
For me, when I do stabilization in TVAI, the audio comes out totally white noise.
One more time for clarity:
Using stabilization in Topaz Video AI completely ruins the audio of the export.
I would say it does not work at all anymore

