I've never created a LoRA in such a rushed manner before. I primarily did it to create a funny profile picture for my social media accounts for the New Year. Perhaps someone else might be interested in doing the same. I believe I've made it adaptable to any character - you'll need to specify that they're "raising a cup of champagne" and that there are "fireworks in the background", and it should work.
If you like this LoRA, don"t hesitate to:
publishing your creations directly in the model's gallery
I've got a ko-fi link too
I created THIS DISCORD SERVER to propose future models, keep you updated on upcoming training, gather your feedback, and facilitate member exchanges. A brief introduction will be required to unlock access to the appropriate channels.
Thank you for your support.
Description
512x512px training
FAQ
Comments (2)
Hi ! I've tried it a few times, but it still looks like Leonardo Di Caprio.
As explained in the description, this is a hastily made model. I didn't use the variety of different images that such a model would need to gain flexibility. Therefore, I suggest lowering the setting of this LoRA around 0.8 and increasing any LoRA of the character you wish to incorporate instead. Then, at the HiResFix stage, I remove Gatsby's LoRA with a denoising over 0.4 (for me, this method works; I'll provide an example when I have time). If there are more generations of this type shared below, I could retrain this model. But really, I didn't aim for high quality here, sorry.