I refeed my profile picture to AI 175 times and the results were weird.
To get things started, here's my last video. It's a very simple compilation of AI regenerated pictures based on my profile picture. If you want to read the prompts that lead to all these generations, they are pretty funny as well:
I'm going to start making some more high effort content in the future. Just starting small to see if there's any interest in my channel still
Sign up for OddsRabbit
Join communities like /c/ContentCreators and never miss posts like this. Every signup donates a meal for a child in need. Learn more
Comments
Wow, the post on snowpire is extremely thorough. How long did it take you for the video, and then for the post itself?
If you're ever thinking of potential future content ideas, I think this but experimented across 2-3 different AI models would be interesting to see.
Thanks. Most of the time was spent on programming. The website was generated by python with the images and LLM prompts getting appended as they were generated. For the video I also stitched together all the generated images using FFMPEG. I also did do an alternative run using a different image model, Mixtral, with some very interesting results, but thought that maybe two separate sites would be a it too much. However I can make a video where I compare the models and talk more about the process. It's something I sort of want to do, and was thinking about whether I should have done it at the end of this video. Maybe that does become my next post!
I think it could be a very fascinating and viral idea. A video split in half, side-by-side comparison of the ai models. You can make it a series and compare all the different models
