Awesome turbo LoRA!

#1
by Job0115 - opened

awesome turbo LoRA, hands down. It also fixed the blur issue that was present in 8-steps-dmd2-lora.safetensors by Lodestone. Thanks!

This model is indeed very good; it works exceptionally well for image-to-image tasks, such as style transfer. However, when it comes to text-to-image generation, the model tends not to follow the prompts very closely.

This model is indeed very good; it works exceptionally well for image-to-image tasks, such as style transfer. However, when it comes to text-to-image generation, the model tends not to follow the prompts very closely.

can you post an example that reproduces the prompt following issue so I can take a look ?

This model is indeed very good; it works exceptionally well for image-to-image tasks, such as style transfer. However, when it comes to text-to-image generation, the model tends not to follow the prompts very closely.

can you post an example that reproduces the prompt following issue so I can take a look ?

Sorry, maybe I concluded too early. The main prompt I used during testing was to have a man cover his chest, but the image result always produced a woman covering her breasts. It wasn't until I removed several prompts related to female features that the male images could be generated properly.

There is now a V2 update.

Sign up or log in to comment