--- base_model: krea/Krea-2-Raw tags: - text-to-image - diffusers - lora - krea2 - template:sd-lora license: apache-2.0 instance_prompt: "MIDRI" widget: - text: "A futuristic cybernetic owl perched on a neon-lit skyscraper ledge in a rainy Tokyo alley, MIDRI cinematic lighting, 8k resolution." output: url: sample_0.png - text: "An ornate gold pocket watch floating inside a swirling cosmic nebula of purple and gold stardust, MIDRI ethereal atmosphere, surrealism." output: url: sample_1.png - text: "A rustic sourdough loaf resting on a flour-dusted wooden table in a sun-drenched Provencal cottage kitchen, MIDRI hyper-realistic textures, soft bokeh." output: url: sample_2.png --- # Krea 2 LoRA — funseshon/midri A DreamBooth-LoRA for **Krea 2**, trained on **Krea 2 RAW** and shown on **Krea 2 Turbo**. The samples below were generated with this LoRA on Turbo (8 steps). ## Trigger Use the token `MIDRI` to invoke the concept. ## Samples ![sample](./sample_0.png) > *"A futuristic cybernetic owl perched on a neon-lit skyscraper ledge in a rainy Tokyo alley, MIDRI cinematic lighting, 8k resolution."* ![sample](./sample_1.png) > *"An ornate gold pocket watch floating inside a swirling cosmic nebula of purple and gold stardust, MIDRI ethereal atmosphere, surrealism."* ![sample](./sample_2.png) > *"A rustic sourdough loaf resting on a flour-dusted wooden table in a sun-drenched Provencal cottage kitchen, MIDRI hyper-realistic textures, soft bokeh."* ## Use it with diffusers ```py import torch from diffusers import Krea2Pipeline pipe = Krea2Pipeline.from_pretrained("krea/Krea-2-Turbo", torch_dtype=torch.bfloat16).to("cuda") pipe.load_lora_weights("funseshon/midri") image = pipe("A futuristic cybernetic owl perched on a neon-lit skyscraper ledge in a rainy Tokyo alley, MIDRI cinematic lighting, 8k resolution.", num_inference_steps=8, guidance_scale=0.0).images[0] image.save("output.png") ```