Welcome to the new era of video content customization — reduce the price 10x, save the time of your team and forget about specific VFX skills with Dowell.
We trained neural networks to automation transfer faces from video to video with a high degree of realism.
You give us 2 videos — ‘donor’ with the face, which you need to transfer, and ‘receiver’ with the scene, where you want to put ‘donor’ face.
Mimic, poses, lightening and size of the donor's face should be just like on ‘receiver’'s face.
Our model is learning 24 hours and can create 50 videos in parallel. We are developing the algorithm which will transfer multiple faces in real-time right now.
1. we take 2 videos — ‘donor’ and ‘receiver’
2. apply face detection for both of them
3. remove the artifacts
4. train the algorithm
5. improve the final result
Working with celebrities
We know how brands adore working with celebs. They are brilliant, exude success and increase brand awareness. But also we're getting inflated budget estimates, missed deadlines and offensive performance (the star doesn’t mean an artist).
Make a 2 minutes video with your celeb on a green screen and shoot a spot with a professional stand-in. For everything else, there's Dowell.