I confess, while I’m a great editor, as a videographer I have some flaws. For example, on occasion my focus can be slightly off. I’ve tried using Topaz Video in the hopes of saving my footage but it really does nothing extraordinary. I’ve tried all the preset and parameters. And sharpen just makes it so much worse.
But… if I export the clips as a TIFF sequence and bring them into Topaz photo, choose recover face and then batch process the images with that setting applied to all, my footage is saved. The difference is night and day.
Why does Topaz Video not recover these details the same way Topaz Photo does and just do it to a clip? This is the same company.
This is due to how the models see the frames, the Photo AI model only looks at the single image without context of the other frames around it. Video AI models take the other frames and tries to retain the motion and adjust accordingly.
When doing the adjustments like this in Photo AI you can get differences from frame to frame that when played back later do not line up.
The research team is still working on developing the models and tools in Video AI to improve their results when it comes to adjusting focus and understanding depth of field.
