Portfolio Fictional player videos created with Stable Diffusion

Remove this Banner Ad

Hey guys,

Just wanted to share something I was mucking around with this evening. I made some fictional North players from clips of Boomer Harvey using Stable Diffusion, then mapped them back to the original videos using EbSynth.

Much more work could have been done to make the video smoother, but I was really just testing if it could be done at all.

 
This is really interesting, I’ve often wondered about the process of using AI image generation in the brainstorming process and actually plan on posting about this too, but in a similar vein to this I wonder if multiple different technologies can be used to make this indistinguishable from reality. Something like deepfaking a real player face onto the picture and manually adding in the actual logos onto the jumper then tracking the paths they should move across. It’s truly fascinating though, but perhaps equally frightening.
 
This is really interesting, I’ve often wondered about the process of using AI image generation in the brainstorming process and actually plan on posting about this too, but in a similar vein to this I wonder if multiple different technologies can be used to make this indistinguishable from reality. Something like deepfaking a real player face onto the picture and manually adding in the actual logos onto the jumper then tracking the paths they should move across. It’s truly fascinating though, but perhaps equally frightening.

You could easily do some masking and restore the logos. Or even just correct the logos in Photoshop before processing in EbSynth.

I knocked this one up this morning. These are far from polished products, but it's cool to see what's possible.


 

Log in to remove this ad.

Remove this Banner Ad

Back
Top