Though not talking about slow motion, I explain a work flow that would accommodate this ability in the main TVAI 4 topic.
I’m not going to copy paste it here, so there’s the link.
Running more than one model with a single process, selecting frames, and assigning different models to each frame or set of frames
The AI should be allowed to help the user to automatically identify and cut the video into different scene, and let the user choose the appropriate ai model in the different scene
1 Like
These ideas overlap nicely.
what I do in some cases is making two complete enhance run with different settings/models, then I Import both file in my Video cutter program. Depends on scenes I took upper or lower video track or you can play with opacity for video segments to get a mix of both.
1 Like
This may be what I’m thinking. I want to assign different filters to different sets of in & out points.
USE CASE: VHS video. IrisHQ is perfect for close-ups, but disastrous for mid-shots. I want to use Artemis in mid-shots on same video.
I believe there are two more idea topics that boil down to this idea. This is put more concisely.
I tend to cut movies into different scenes using Premiere Pro and process them independently, and then reassemble them in Premiere Pro. That also gives me the option of color correcting and enhancing audio (Premiere Pro Beta) before I process the movies in TVAI, or post process in Premiere Pro. It’s a lot more work, but I’ve found it’s the only way to get around using a single filter, which is most likely not one-size fits all, in TVAI.
Changing the speed of a video also changes the sound. Unless you’re ok with your sound changing, that means the app needs code to adjust the audio pitch in sync with the video.
Best process: use frame interpolation to increase framerate, then export the scene to a video editor that has speed ramping with the option to maintain audio pitch. Another function available in freeware that doesn’t need to be added to TVAI.
I figure if you want to change the speed of the clip, you already have a plan of what to do with the audio outside of TVAI.
It would be fine if TVAI just keeps the audio the same or removes it. I’ve never bothered to look what it does with audio now if you run a slow motion model on it. In VEAI 2.6.4 it would remove the audio.
That’s what I’ve done the few times I’ve used slow motion: use the slo-mo over the main video, keeping the original sound. But selecting one portion of a longer video and changing its speed is something someone would want to do while processing the entire video, and then what would they do about the sound?
I’ve made a topic addressing this issue.
So does that mean the slo-mo clip runs out of sound before it ends?
Yes. Since TVAI doesn’t have the feature to slow down the audio, choosing the “Copy” or “Convert” Audio Modes during export will add the sound at a normal speed while the video itself is slowed down - making the sound finish before the video does. Choosing the “None” Audio Mode will remove the sound completely.
Interesting. The one time I used slow motion I knew I wasn’t going to use the sound from the clip and set the enhancement to no sound. Can’t imagine why anyone would want a slow motion scene with out of sync sound that ends too soon.
I feel like that’s a fair compromise. You can take that audio and slow it down in an audio editor later, if that’s what you really wanted to do. The work of getting you the right clip of audio for that has already been done by TVAI at that point.
I keep a copy of the freeware videoproc editor because it has speed ramping with sound adjustment. It’s not great for anything else, but feed it a 60-120 fps clip and it’ll work wonders on slow motion with sound.
It’s too time-consuming to cut it yourself. The AI should be doing it.
And if you run a video and discover that one scene in the middle didn’t come out well, how time consuming will it be to redo the entire video?
I found another idea that should be merged into this one.
Minus the color tweaking. That’s a separate idea.
