from @levelsio | by @levelsio



over 1 year ago

โ€ขView on Twitter

๐Ÿง  Trying to train my own model with Dreambooth today You seed it with 5 to 30 pics and it will modify Stable Diffusion to be more about what you seeded it with, kinda like steering the model I think Will update on progress in this thread!

โœ… Trained my own model It worked! At first glance it seems like there's A LOT less anomalies and artefacts as the re-trained (on interior) model knows better what we want (interior)

More from @levelsioReply on Twitter

Page created with TweetHunter

Write your own