What I would like to see for these Topaz AI solutions is an option to update the models more frequently. If new features and bug fixes are being worked on it can take quite a bit of time to program in code and test it before releasing it to the public, I think updating the AI models can give us customers something to look forward to.
Another feature I would like to see is the ability to train our own AI model within the software so we can fine tune the AI model to fit our needs more.
A feature that I hope is already being worked on is giving us control to choose different settings for output. Color, blur, sharpening, codecs, bitrates etc. Right now Video Enhance AI is just a kind of a easier method to use the AI models for video, no more needing to extract every frame from a video to have Gigiapixel AI to render out.
AMD GPU support would be a great addition since I dont believe Nvidiaâs RT cores are being used.
I also see that Video Enhance uses different amount of VRAM after opening and closing the application in Windows 10. Maybe we should have a way to choose the batch size if we see the application not utilizing more of the GPUâs VRAM. For example on my GTX 1080 with 8GB of VRAM after opening Video Enhance AI it will use 4GB to 6GB of VRAM randomly, even if the settings are exactly the same, every time I close and reopen the application it will use a different amount of VRAM. At first I thought that the software is doing what it thinks is best to optimize the quality and speed, however I do see that rendering times are much shorter when more VRAM is being used while quality is the same.
A nice peace of mind option that seems to be missing is the check for updates option within the âhelpâ tab. Kinda odd to have a update button to appear only when either the application has been up and running for hours and to see it during a very long render.
Does anyone know if this software and Gigapixel AI use a similar AI model that neural-enhance uses?
There should be encode settings for advanced users. The default and only codec so far (H.263) is too outdated, and could cause blocks & color banding in many areas. I hope there could be at least H.264 with variable cfr (quality) settings, or, let advanced users use their own FFMPEG encode settings.
Outputing as PNG and encoding through image sequence takes too much extra disc space and time.
h.264 is more compatible. Also, the h.263 encode it does is at a very high bitrate. I doubt you would see any kind of artifacts (blocks, banding, etc.). I do agree there needs to be an âadvancedâ section where users can adjust some of the parameters for the encode.
I can assure you that the source is very good and doesnât contain such problem at all. No blocks, no color banding, little noise.
There are other examples as well, if you look closely. Yes, itâs at a very high bitrate, but the quality isnât exactly what youâd expect in such bitrate. My own encodes usually donât have visible blocks & color banding, and can be achieved in a much lower bitrate. Iâm confident to say, that with the proper codec and encode settings, this can be improved by a lot.
does this happens on every setting (LQ, HQ and CG)? And what are the specs on the original video (res, codec, bitrate)? In my experience VEAI usually lessens the amount of blocks. I always use HQ because it seems to me to be to have better âerrors to image quality enhancementâ ratio, but it may differs depending on settings and original video quality
I would like to have to option to resize the original video and let the program sample the smaller video I want because I notice that for some lowtier videos lowering it makes everything greater!
Anyways its like having a whole CGI team at your home this so amazing I cant believe this software is real ! Holy shit my brain explodes by this software!
No, I havenât tested the same source again, and I assume itâs the compressionâs fault. If itâs not, then it would be a bigger problem.
In my latest PNG output test, I found that the output has small blocks near a few edges where there werenât any blocks (very smooth). This is very upsetting as well.
I want to echo the call for better encoding options. The H263 codec is giving me playback issues on my Vero4K running OSMC (e.g. Kodi.) It plays properly on VLC player on my Windows PC but the Vero4K has crazy bands that make it unwatchable.
The Vero forums recommended I encode my video with H264, but I currently donât have that option with Video Enhance AI.
Iâm guessing they went with h.263 because they think it might be âmore compatibleâ with older players. h.264 is pretty much the defacto standard today. I have equipment from 2010 that handles h.264 just fine. They really should change it, or give users an âadvancedâ box with some options for video encoding.
Reencode with AVC/h264 or HEVC/h265 with handbrake. I sugest going for a âTrial and errorâ with a 10 second clip first with different bitrates, but as for myself I got good results with h265@6,5mbps or h264@8mbps for 1440p VEAI video files. Deppending on image quality and resolution you may want to use higher or lower bitrates
I finished a lot of small old Vhs/S-Vhs footage. The results from clip to clip is very various.
I testet each kind of pre-rendert down- or upscaling in order to deal with trial on error.
On the end It depends on the elements/content on each clip:
footage which were made e.g. in/from towns/cityâs with mostly buildings in the picture brings the best
results; sometimes near perfect.
footage in a mix of buidlings and green (Trees, grass) shows first big problems: to all green
plants VEAI adds smaller or bigger structures like a ball-net >>> not usable!
footage with only landscape content shows real big problemes with new added smaller and bigger
structures >>> absolutely not usable on the end
In order to get rid of the new added structures, I pre-rendert the landscape footage with the plugin Neat Video. This tool reduce picture video noise very impressive. I very carefully reduced the noise.
Afterwards I used VEAI: yes, the added structures on landscape content is nearlly gone, but the picture is now âwashedâ and too smooth without sharpness. Also here I tested all resolution sizes without any improvements.
So VEAI needs definitly more and better AI templates, which can recognize in the footage scenes with e.g. more buildings and/or more landscape and green plants.
OK, it tested the new 1.2.0 version.
The rendering process with my GTX is faster.
Again: I tested my common old VHS/S-VHS video clips with different pre-rendered resolutions:
640x480, 720x576, 720p and 1080p.
I see on each new named modus - I tested each new (named) modus in VEAI - no improvements on landscape footage. I use since many weeks in testing landscape footage several clips von Monument Valley. You have there a good mix from big stones, table stones and green (plants). Even on the table stones and much more on green plants VEAI adds the always named small point structures like a net.
The 3 variations of Artemis modus are not usable for low resolution, noisy and blocking footage.
The improvement results are always noisy, blocking and has no clear edges like e.g. the CG modus.
I tested alternative other landscape footage like Yosemite, Sequioa and so on. Same as with above named footage: The only modus, which brings a real improvement is the CG modus. But here again the added point net structures on e.g. greens/green plants/fields are not helpful.
I get a small improvement - also with the versions before -, if I very slightly denoise my footage with Neat Video and us it afterwards in VEAI. But also here on some clips I get the point net structures added.
It seems the common thing I hear around different areas of the net about this program is âLQ (now renamed Artemis) is no good.â So it sounds like the 1.2.0 update didnât change things much. When I tested it I could see what people meant in that it seemed to try to create detail that didnât exist in the source, making it look more like artifacts. Hopefully this could maybe get better in the future. But I still think weâre a ways off from feeding in crappy EP mode VHS and getting footage that looks like it was shot in modern HD. But with the way computers and technology is progressing, I really do think we will get to that point some time in the future.
Thanks, thatâs what Iâve started doing. The other issue I run into is that the mp4 produced by Video Enhance AI has the audio horribly out of sync. Iâm not sure but I think it may be because my source is 24fps and the program seems to default to 29.97 fps, and I donât see an option to adjust the fps.
That the idea/concept of VEAI from the scratch by Topaz could work pretty good is already known.
For example: I have old bad footage of Las Vegas from 1992 via S-VHS. Itâs very noisy, has no clear edges/lines and has common blockings. Overall: the picture is really âdirtyâ. Here VEAI works really âmagicâ. The results are really really impressive. Nobody including myself could believe the comparison before and after. No remarkable noise, clear edges and lines, deblocked and sharper picture. Great Great âmagicâ
Also impressive are results of the night/dark footage of Las Vegas. Really impressive!
Same mostly with footage e.g. from San Francisco (also from 1992). BUT: footage taken e.g. in parks of San Francisco shows the problem: greens/plants/grass/trees became the named additional structures.
Perhaps it need much much more AI learnings, which depends on landscapes, especially on handling of plants/trees/grass. Perhaps itâs on the end not possibleâŠI donât know. But I hope very strong, that Topaz will work hard on it, cause using VEAI only for âtown/cityâ footage canât be the business case.
From my side I always stress adjustment sliders. Perhaps it need on the end only one to adjust the added fishnet/pointnet structure.
Have you checked fps before and after you render with VEAI? Iâve only had the out-of-sync issue when I render a video that was previously rendered in VEAI. In other words, I take the original footage, run it through VEAI at 100%, then run it once more and it will have the out-of-sync issue.
Using Mediainfo, I see the fps before and after are the same. Theyâre 23.976. Both the original footage and the VEAI rendered clip so I doubt thatâs the problem. I have not tried it in version 1.2.0. I will wait for the new release next week and some release notes.
