There are SO many comments about the TVAI 3 user interface being ‘unusable’ or ‘not as good as 2’ and several with the ‘I’m going back to 2’ thrown in. We basically forced them to change the UI. It was clear we wanted something better—or maybe just the UI of VEAI 2 back.
Now with all the same comments, but about TVAI 4 UI, we are going to force them to completely change the UI again.
I’m not sure what I want to vote for. VEAI 2 was nice. TVAI 3 was very similar to 4. I mainly just want a UI that has a zoomable timeline (With the scroll wheel. Check out GoldWave for a perfect example) and the ability to create as many sections as I need for a video. Each section needs to be able to have multiple AI models assigned to it—and those models need to be able to have different settings. Overall output resolution should be the same, but I should be able to choose if the section will process with the X2 or X4 version of the AI model.
My basic work-flow example would be this. I have a video with several scenes, but let’s just look at the first two. First one is dark indoors with lots of grain, the next is light outside with lots of trees. I would make a selection of scene one, set Nyx with high denoise and Proteus with specific settings that look great for that scene. I would then select then next scene. For it I would maybe turn off Nyx or turn the denoise down. I would still want to use Proteus, but with very different settings to make sure the trees don’t get fake nasty details added to them. Anyway, I would keep defining settings for sections of the video like that until I have it all covered.
I WOULD NEED TO SAVE AND LOAD THOSE SETTINGS because this could take days.
(I should probably add this explanation to this idea topic The title of it does not give the same idea, but it talks about it.
All of that explanation is because I think the UI of TVAI 4 is going in the right direction for me to get the work-flow I want. In the meantime I made a monster script that reads setting files and runs the CLI based on them. It does not allow different models for different sections though—or even more than two sections.
