| import gradio as gr | |
| import torch | |
| import numpy as np | |
| import modin.pandas as pd | |
| from PIL import Image | |
| from diffusers import DiffusionPipeline, StableDiffusionLatentUpscalePipeline | |
| device = "cuda" if torch.cuda.is_available() else "cpu" | |
| pipe = DiffusionPipeline.from_pretrained("prompthero/openjourney-v4", safety_checker=None) | |
| upscaler = StableDiffusionLatentUpscalePipeline.from_pretrained("stabilityai/sd-x2-latent-upscaler", torch_dtype=torch.float16, safety_checker=None) | |
| upscaler = upscaler.to(device) | |
| pipe = pipe.to(device) | |
| def genie (prompt, scale, steps, seed): | |
| generator = torch.Generator(device=device).manual_seed(seed) | |
| images = pipe(prompt, num_inference_steps=steps, guidance_scale=scale, generator=generator).images[0] | |
| low_res_latents = pipe(prompt, generator=generator, output_type="latent").images | |
| upscaled_image = upscaler(prompt=prompt, image=low_res_latents, num_inference_steps=20, guidance_scale=0, generator=generator).images[0] | |
| return (images, upscaled_image) | |
| gr.Interface(fn=genie, inputs=[gr.Textbox(label='What you want the AI to generate. 77 Token Limit.'), | |
| gr.Slider(1, maximum=15, value=10, step=.25), | |
| gr.Slider(1, maximum=50, value=25, step=1), | |
| gr.Slider(minimum=1, step=1, maximum=987654321, randomize=True)], | |
| outputs=('image', 'image', | |
| title = 'OpenJourney V4 CPU', | |
| description = "OJ V4 CPU. <b>WARNING:</b> Extremely Slow. 35s/Iteration. Expect 8-16mins an image for 15-30 iterations respectively. 50 iterations takes ~28mins.", | |
| article = "Code Monkey: <a href=\"https://huggingface.co./Manjushri\">Manjushri</a>")).launch(debug=True, max_threads=True) |