So when I am applying Proteus manually, and testing by preview and I decide I need further tweaking of the settings, it opens a new output, and I need to close the video on the previous output. Is that the way it is supposed to work, or am I doing something wrong?
What do you mean?
I think you’re supposed to be able to just click on the completed previews in the list and view them. That way you can compare them somewhat quickly.
You keep spelling it Handbreak. It’s Handbrake.
3 Likes
Look at everything else!
What Screen Shot you look at?
I see no difference…
so of course, we can see that the image is compressed with artemis low! we can see that the edges are smooth, and there is a loss of detail anyway…
I use vegas pro too, and I confirm, the preview is slow when it starts to have effects, the technique consists in selecting a small area of a few seconds and doing “shift+b”, it will make a pre-render to read the segment without lag 
1 Like
It is hard to look at any details, when images uploaded to this forum are showing compression artifacts.
I’m curious what is up with the odd input frame rate of 28.444?
What is causing that?
Your screenshots are terrible. We don’t need to see your entire desktop. Can you not maximize Video AI and then capture the entire screen?
And most of the time nobody here has any clue what you are trying to point out about the images.
6 Likes
In the end I think that I found a good compromise for my settings in Proteus:
Set “Parameters” to “Manual” and sampled “Estimate” at about 10 to 15 different frames across the whole Video. This confirmed me that Parameters vary a lot thru the whole video (off course that’s a nobrainer), and in consequence one single set would not perform best for all frames. Therefore I concluded to set “Parameters” to “Relative to Auto” and adjusted the individual setting carefully a little bit (according the results found by “Estimate”) . For the moment that’s the best compromise for upscaling my TV-Mediathek videos.
This is called Variable framerate. Very few videos does a real Constant frame rate. VEAI certainly display the real framerate. so you can see a 30.371 to 28.88 instead of 29.97 etc…
ii’s displayed in mediainfo. video are variable or constant birate. information is written in it on the full details page.
For example every video with a constant bitrate remised in Avidemux is converted to variable bitrate. the bitrate does not usually go above or below 1 fps of variable.
Ok - I’ve researched it. Definitely a problem maker in a lot of video editors (if they support it). In the end they just change it to CFR anyway.
Best to convert to a CFR prior to using.
guys , i am so shocked this was released with all these issues, i paid for the update last week, and now i kinda almost regret paying to update. The issues need to be fixed, speed is defiantly slower, and you can definitely see more artifacts. I dont understand why a rtx 4090 isnt that much faster than the 3090. Makes no sense, some addressing needs to be done for the optimization for nvidia cards working with VEAI. Might as well stuck with a 3090 , its mind baffling especially since the 4090 is supposed to be twice as fast as the 3090 but you do not see it in terms of performance with VEAI and also with the fact version 2.6 seems to unfortunately still be faster at conversion. I usually like the updates on VEAI, but the bugs, speed, and artifacts made me wonder how this was signed off on as a full version release.
7 Likes
The Topaz Video AI 3.0.0 upgrade seems to have resulted in an product similar in nature to the recent merger of Denoise AI, Sharpen AI, and GigaPixel AI into the single Topaz Photo AI program, in that some functionality and refinement was lost in translation. Although the newly combined Photo AI is convenient and often amazing, there are times when one finds better control and refined results using the prior three individual programs; not always, but in a case-by-case basis.
I like the direction Video AI 3.0 is going, and hopefully, their programmers can refine this into a do-all workhorse. Obviously, we are not there yet as evidenced by the reports in this forum. From the comments here, it seems there is a need for additional user inputs to refine and control the results and better handling of diverse video sources and refinement needs.
I still feel like on a blind date with video ai during processing a video where I am still “Waiting for (the very first) preview”. Maybe wait for a hot fix or the next version because even testing without a control image is impossible for me. 
1 Like
Great experience using this software so far, however I’ve been unable to find a way to delete a user defined preset or to otherwise modify or overwrite one if I want to tweak its settings.
Under “Edit” you can find these options.
If you want to modify/change one of your existing presets simply select it, change the settings you want to edit and then under “Edit” click “Modify Preset”
1 Like
The new deinterlace finally looks fantastic. In v2.6.4 it looks terrible, but now in v3 it looks like the video was never interlaced to begin with. Brilliant!
(This on PAL-SD from S-VHS tape digitized to .ts mpeg2 with no upscaling (just deinterlacing))
