I’ve noticed over the last year or so that Topaz has become very aggressive with their advertisement and search-engine optimization. After seeing this in my social media feed a few days ago
I decided to make this thread. Here are some of the better alternatives to TVAI.
And here are some others.
(Free ↓)
(Free ↓)
(Free ↓)
(Free, in alpha ↓)
Add more if you’d like.
Have you tried any of these?
Personally tried
Nero = pure crap
Unifab = almost good but some scene are for no reason completely wrecj, unusbale
AVCLabs = far less good than TVAI
Fai2fuX = crap, may be a bit good for only animes
Videoproc = best behind TVAI, but anyway behind clearly
Testing Audials Vision, may be weak results as H.265 unable in free emo … what a stupid choice
Yeah, we’ve been keeping an eye on Audials Vision’s progress for a while now, especially there Ai model to enhance audio up to HiRez 24bit-98!
Think this app is worth a try
Ans super cheap, currently 60$ !
They used stars as models names, what a coincidence ^^
Ahh… cool. Thank you, Alpheratz!
Seeing the buying frame, it is a german app.
Love fine tweak setting
HDR Rendering 10 days on RTX 3070, with Tensor max quality 
Videoproc has potential, but still a long way to go.
Well I also thought that but the GUI is so poor and step ahead are as rare as Topaz’s
That Audials Vision souncds great with almost what I expect, but a kid interface much less pro than TVAI
Only one model is IA and if you don’t choose it, you’ll spend 45 days with a RTX 3070 to 1080p → 4K, thats insane
@gene-8240 I’ve used all of them except Unsqueeze (because I don’t use Mac), enhancr, and chaiNNer.
@Alpheratz One thing I like about TVAI is the degree of manipulation that it allows. TVAI has shown me what a great program ffmpeg is 
. However, the average user doesn’t care about these things and most just want a simple program that works. Some of the other programs either already do a better job at this or are on track to reach that point.
@vincentabel42 Unifab also has AI audio upscaling, but there are standalone applications that do that as well. I have been doing audio processing for a very long time, so I don’t really need AI assistance with that.
I have been doing audio processing for a very long time, so I don’t really need AI assistance with that
Hello.
Cool. I’ve kinda been getting my feet wet (in swamps, though… ha-ha) a little bit here & there lately with Dolby Render, DTS:X suites; from your enthusiast take, what is your go-to one-stop-shop for all your audio enhancement needs ?
I also tried all of them, my verdict is sadly, no one reach the level of TVAI, some have even mor ebugs that this last one … So …
Only HDR handling is better in 2 apps, that are Unifab and Audials Vision, but interfaces are uggly and TVAI’s Hyperion is progressing, I hope
I didn’t like the results generated on my videos (Video8 720x576) testing"
AVCLabs
HitPaw
UniFab
VideoProc
Winxvideo
All of them destroyed details just to much.
VideoProc/Winxvideo (they use the same technology). were actually pretty well for Photo upscaling, especially on Adult Pictures. Topaz software is very bad for adult images/content.
Denoise (hqdn3d / nlmeans) preprocessing and deinterlace (bwdif) functions are still reliant on underlying FFmpeg.
The intel filters and model are a pain to compile.
Here’s an academic paper which describes the process.
Hopefully, a kind soul will soon compile and host a binary for Windows, macOS & Linux for easier installation.
sr super-resolution filter (and model). Results were disappointing compared to TVAI, using what was an old, off-the-shelf, pre-trained model.FFmpeg documentation steers users away from the sr filter and to use the dnn backend in preference.
To get full functionality (such as async execution), please use the dnn_processing filter.
So both sr and dnn_processing are just tools to use a model, rather than the models themselves. The training or selection of model becomes the critical component.
Pixop is worth a mention too https://www.pixop.com/. This is a pro-tool used by studios and broadcasters.
While not upscaling, Cinnafilm Tachyon (deinterlace) and Dark Energy (denoise), Wormhole (retiming) are also worth a callout. This is a pro-tool used by studios and broadcasters.
Neat Video is high end denoise.
Moscow State University also runs a comparision of super-resolution techniques. Video Super-Resolution Benchmark
This is a great thread - thankyou to OP for putting it together.
My go-to is Stereo Tool, but as the name indicates, it only does stereo processing. The learning curve is as high as can be on it. The results I have been able to get out of it are exactly what I want though.
Ok… ok… ok…
Great to know, FS! Thank you! I’m looking into this now. Thanks again.
Is there an Application that is able to fix the compression of Podcasts?
This is Topaz VIDEO AI thread 
Tested DVDFab audio enhancer : a pure failure for movie purpose, any chip that upscale DTS→DT-X or DD→DTHD will do far better job
Acrovid Neofootage…also for convert in HDR
This is a database of open-source AI upscaling models:
There are hundreds of models for all types of sources, including animation, VHS and film. The database is maintained by the developers of chaiNNer, and most of the models on the site are supported within chaiNNer.
