sd-textures-lora-ft

<!-- This model card has been generated automatically according to the information the training script had access to. You should probably proofread and complete it, then remove this comment. -->

LoRA text2image fine-tuning - adit-desai/sd-textures-lora-ft

These are LoRA adaption weights for stable-diffusion-v1-5/stable-diffusion-v1-5. The weights were fine-tuned on the adit-desai/textures_color_1k dataset. You can find some example images in the following.

Limitations

The prompts need to be structured in a particular way (the way they are structured in the dataset).

Ex 1. Prompts have to start with "a texture, ..."

Ex 2. Certain characteristics need to be written in a certain way. For cracks, you have to write "crack, cracks, cracked" (since this is how it is written in the dataset). There are various such examples. If you write just "cracks", it's likely that generated image isn't what you desire.

You also might need to run the same prompt multiple times or try different words to get an output you are satisfied with.

In the future, if I get time, I will work on improving the quality of prompts in the dataset to overcome above issues.