Continued video / audio sample / audio file
Hello, TenStrip! First, I want to thank you for this model and for the incredibly cool workflow. The quality and candor of the videos I received from your workflow are extraordinary compared to typical workflows.
Secondly, I'd like to ask you if you might ever modify your workflow with any of the following features (I've done some of this with simple workflows and basic models, but I don't understand how to incorporate them into your workflow):
Video continuation: Several input frames are specified (to clarify movement and context), and a new video can be "continued" using these frames, ultimately creating a very long video.
Last image: This simulates "continued" videos, so you can at least know what frame the previous video started from and continue from there. This is also useful for looping videos.
Input audio file for full voiceover. To voice the entire audio in a video being created (for example, a "talk-along" video).
An input audio file as a voice sample. So that the voice resembles the voice from the input file.
An input audio file with instructions on which sections of the voiceover to take and how to insert them into the video, indicating when to speak them.
p.s.
Thank you so much for your work. This is really cool!
It's more or less a normal workflow, ignore the guider and the anchor nodes and implement all that as normal. The only thing to figure out is wiring the prompt guidance and masked audio latents. All the anchor nodes have no real use for video extension so the only thing unique in the workflow that other ones don't have is just the STG Guider and sigmas which are the main backbone that increase quality, delete all the other stuff and just add in those capabilities normally.