AI to bring old interviews to life???

BS Dave

BS Dave

BS Dave

Soldato
Joined
19 Oct 2004
Posts
5,010
Location
London
So i work for a somewhere we have a lot of archive interviews (audio) and in a lot of cases, pictures of people before/during/after the interview, but no video.

I was wondering what programs etc I would use to make the photos come alive and speak their part of the interview?

AI can do this now right? This is just for research at this point. The other thing is the interviewees would be famous. And i think some AIs don't let you create images of them, is that correct?

If anyone can help me out it would be really appreciated.

Thanks
 
BS Dave said:
So i work for a somewhere we have a lot of archive interviews (audio) and in a lot of cases, pictures of people before/during/after the interview, but no video.

I was wondering what programs etc I would use to make the photos come alive and speak their part of the interview?

AI can do this now right? This is just for research at this point. The other thing is the interviewees would be famous. And i think some AIs don't let you create images of them, is that correct?

If anyone can help me out it would be really appreciated.

Thanks
Click to expand...
I've not used any video generation tools, but Sora 2 is I believe the best option at the moment, and can take images as an input, so probably worth looking at.
 
To avoid uncanny valley, especially with recognisable subjects, you might need to feed the model with quite a lot of content - even given today's AI capabilities.

Also, this will be expensive. VERY expensive. VERY VERY expensive. It is one thing to create novel 10 second clips; and a whole different ball game if you are looking to generate videos of length. Did I mention how expensive this would be? :cry:
 
SharkyUK said:
To avoid uncanny valley, especially with recognisable subjects, you might need to feed the model with quite a lot of content - even given today's AI capabilities.

Also, this will be expensive. VERY expensive. VERY VERY expensive. It is one thing to create novel 10 second clips; and a whole different ball game if you are looking to generate videos of length. Did I mention how expensive this would be? :cry:
Click to expand...
OK cool. SO it could make something that could live on social, but that could be used to throw to a longer audio piece. Thanks
 
Creating longer segments with things like lip syncing takes a fair bit of hands on and a LOT of computation power currently.

I've not used it but ElevenLabs is usually one of the better bets in this area in terms of non-complicated approaches.

This kind of thing is possible see https://www.youtube.com/@Matt_Reconstructs_History for example but I don't know their workflow.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom