Video Enhance AI v1.8.0

[Maveno (Mike)]

I am doing great work with MSI 3060 Ti Trio X 8G. It is an amazing card. Ultra silent, turbo fast.
It eats my previous 2080 RTX for a lunch in Topaz.

[Chase]

@Maveno

I wish to ask the developer once again.
Please provide us all, ALL newest models you have so we can download them. I do not care if the data is 10 GB or 200 GB. I just need the best results from Topaz Video Enhance AI and I can download it. I am sitting on 1 GBit/s symetric fiber connection 128 MB/s speed.

I’m pretty sure the dev’s have OK’d this before, if not, then just delete this post.

These are all the live models currently:

artemis-aa-9.json
artemis-hq-6.json
artemis-hq-7.json
artemis-hq-8.json
artemis-hq-9.json
artemis-lq-6.json
artemis-lq-7.json
artemis-lq-8.json
artemis-lq-9.json
artemis-mq-9.json
gaia-cg-5.json
gaia-cg-6.json
gaia-hq-5.json
gaia-hq-6.json
theia-hide-3.json
theia-hide-4.json
theia-hide-5.json
theia-hide-6.json
theia-hifi-3.json
theia-hifi-4.json

Here’s a download list. Your models folder is by default in C:\Users\Admin\AppData\Roaming\Topaz Labs LLC\Video Enhance AI\models.

Take a look in there and see what models VEAI is using:

.ml.tz = Apple Core ML (useless to Windows Users)
.ov.tz = Intel Open VINO (for CPU speed boost on non GPU machines)
.fp16…ox.tz = Open Neural Network Exchange Half Floating Point Precision
.fp32…ox.tz = Open Neural Network Exchange Single Floating Point Precision

Some video cards may use FP32, others use FP16. Note that FP32 is twice is big as FP16. (Twice the download size.) The default models VEAI installs with locally are 288x288 fp16 ox (Open Neural Network Exchange).

You can paste these lists into a download manager, or JDownloader. Model files and JSONs go into the above models folder.

Apple CoreML: http://veai-models.topazlabs.com/aaa-v9-fnet-fp16-256x352-1x-ml.tzhttp://veai- - Pastebin.com
Intel OpenVINO: http://veai-models.topazlabs.com/aaa-v9-fnet-fp16-256x352-1x-ov.tzhttp://veai- - Pastebin.com
Open Neural Network Exchange Half Float: http://veai-models.topazlabs.com/aaa-v9-fnet-fp16-256x352-1x-ox.tzhttp://veai- - Pastebin.com
Open Neural Network Exchange Single Float: http://veai-models.topazlabs.com/aaa-v9-fnet-fp32-256x352-1x-ox.tzhttp://veai- - Pastebin.com

For convenience, here are the JSONs… I strongly recommend you do NOT overwrite your existing ones: JSONs.zip - Google Drive

The following models will likely fail to download, as they are not present on the download server: http://veai-models.topazlabs.com/thd-v6-fnet-fp16-ov.tzhttp://veai-models.topa - Pastebin.com

I’ll repeat my own notes. Understand these are from my own personal testing. Things may perform differently for you.

Here’s my personal notes that I’ve taken regarding the published (i.e., live on the public server) models so far:

 
 
Non functioning models:

Theia Detail v4
Produces black frames

Theia Detail v6
Produces “Unable to load” error message

Gaia-CG v5
Produces results visually indistinguishable from regular resize (spline, lanczos, etc.)

 
 
Models visually indistinguishable from one another:

Gaia-HQ v5 and Gaia-HQ v6

Artemis-MQ v9 and Artemis-HQ v9

 
 
Models with missing / incomplete model files on the download server:

Theia Detail v4, v5, and v6:
Missing OpenVINO files, breaks entire libraries

Theia Detail v5:
Missing entire library files for Apple Core ML, breaks entire library

Theia Fidelity v3:
Missing OpenVINO files, breaks entire library

Artemis HQ v6:
Missing Open Neural Network Exchange Single Float (fp32) files, breaks 1x scale mode

Artemis LQ v6:
Missing Apple Core ML files, affects 1x and 4x scale mode, missing one block size file for 1x and one for 4x, library should still function

 
 
Notes / differences on other models:

 
Artemis LQ v6, 7, 8, and 9:

Artemis LQ 6, 7, and 8 are mostly the same. 7 removes a little more noise than 6; 8 is slightly sharper in some areas and less noisy (at the cost of a little detail) than 7.

Artemis LQ 9 appears to be a completely new model, with much improved results. It generates a lot more fine “AI” detail, but also generates some “dither” like noise similar to Gaia CG v6. Liking it a lot!

 
Artemis HQ v6, 7, 8, and 9:

Artemis HQ is similar to LQ in the sense that versions 6, 7 and 8 seem to be variations of the same model, while 9 is a completely new model. 6 has the weakest effect in terms of focus / denoise. 7 produces much sharper results but also can enhance noise. 8 seems to try to get things even sharper and in focus and also tries to denoise better than 7, but still leaves noise. 9 looks nothing like 6, 7, or 8. It completely tries to denoise the entire frame, with absolutely no grain or “dither” like patterns. It increases focus, but doesn’t tend to generate any artificial detail. Looks less like an “AI” upscale and more like it could have been done with traditional filters, but is easier doing it with VEAI.

 
Artemis AA 9:

As the description says, it doesn’t upscale, just anti-alias. It looks softer than a regular lanczos resize. Seems to be for special use cases. Not really sure what this mode is doing in a machine learning type of software, because there are plenty of AviSynth / Vapoursynth anti-aliasing filters available already that do a very good job. Not necessarily saying it’s bad, just VEAI doesn’t seem to be the “appropriate” place for this. Just my opinion.

 
Artemis MQ 9:

At first I thought this was the model from version 1.2.0 that has been re-introduced, but it appears to produce the exact same results as Artemis HQ 9. PNG file size of result is exactly the same. I double checked my JSON files and model folder, it’s definitely loading the MQ files and HQ files for each mode.

 
Gaia HQ 5 & 6

These versions are visually the same. I haven’t experimented too much yet with them, so I’m not sure how “on par” they are with VEAI version 1.6.1 Gaia-HQ. (Yet.)

 
Gaia CG 5 & 6

I have so far not found any material that GaiaCG 5 does anything with that looks different than a regular lanczos resize. GaiaCG 6 looks much closer to VEAI 1.6.1 Gaia-CG. In 1.6.1, there seems to be a slightly higher level of what looks like a small amount of post-blurring and post-sharpening (not that it actually does that, just what the result looks like). This can lessen the “dither” types of patterns that it often creates. GaiaCG 6 doesn’t seem to have this “post blur” effect and is capable of creating more subtle, finer detail than Gaia CG in VEAI 1.6.1. But that means that the dither-type pattern is going to be more visible at times.

 
Theia Fidelity 3 & 4

These two are very close. Need to look close to see visible differences. 3 leaves a little more noise, but can sometimes preserve details which aren’t noise that 4 will remove. The difference is visibly subtle. I would probably go with Artemis HQ/LQ v9 before I went with this. But it can depend of the video, whether it’s animated, etc.

 
Theia Detail 3 & 5

Theia Detail 4 produces black frames while 6 produces an error message that the model doesn’t load. 3 and 5 look very close visually, but 5 can bring out more detail in small areas. You’ll have to look close to see it, but when you find it in still frame a/b compare, it’s there. Probably not noticeable on video playback. Version 5 is only available with ONNX acceleration as it’s missing crucial model files on the download server for both OpenVINO and Apple Core ML currently.

 

[johnnystar]

Your suggested upscale percentages are interesting, given that VEAI only enhances at 3 sizes - 100%, 200%, 400%, and all other in-between sizes are interpolated with conventional Lanczos scaling. Perhaps it has to do with some degree of image detail loss with the Lanczos 100% to 105% boost.

[cgvids]

I can’t get the GaiaCG 6 to show up, is there a step I’m missing?

[rbfx4x]

I think I can echo the sentiments about 1.8.0 being the best version and that Artemis V9 is magic.
However it can have some weird effects and sometimes turns poor quality sections of video into almost cell shaded cartoons.
This frame was a good example of where the AI struggles. Strange sharpening effects with other parts blurred almost like a photo of a miniature model of the place. Artemis MQ 480p DVD to 1080p.

Edit: Ryzen 7 3700X, 32Gb, RTX 2070 btw

[Chase]

Maybe gaia-cg-6.json ? It goes in the same folder as the models.

[lee.keels]

Wrong version? Wrong forum? We’re in the 1.8.1 now.

[gsbsdh]

You can run it twice if first time is 100% Denoise Deblock to clean up video then run A-HQ to do the upscale. Results are good.

[Artur_Alekseev (arthuralexeev)]

Yes. Quality of jpeg compression is set to 51%. But should be set to 97% as stated in explanation text. Surely it is a bug.