app.py

import gradio as gr
import numpy as np
import random

import spaces
from diffusers import DiffusionPipeline
from transformers import T5EncoderModel, CLIPTextModelWithProjection
import torch

device = "cuda" if torch.cuda.is_available() else "cpu"
text_encoder_repo = "silveroxides/CLIP_L_Fur"
text_encoder_2_repo = "silveroxides/SeaArtFurryCLIP_G"
text_encoder_3_repo = "silveroxides/t5xxl_flan_enc"
model_repo_id = "stabilityai/stable-diffusion-3.5-large-turbo"

if torch.cuda.is_available():
 torch_dtype = torch.bfloat16
else:
 torch_dtype = torch.float32
text_encoder = CLIPTextModelWithProjection.from_pretrained(text_encoder_repo, torch_dtype=torch_dtype)
text_encoder_2 = CLIPTextModelWithProjection.from_pretrained(text_encoder_2_repo, torch_dtype=torch_dtype)
text_encoder_3 = T5EncoderModel.from_pretrained(text_encoder_3_repo, torch_dtype=torch_dtype)
pipe = DiffusionPipeline.from_pretrained(model_repo_id, text_encoder=text_encoder, text_encoder_2=text_encoder_2, text_encoder_3=text_encoder_3, torch_dtype=torch_dtype)
pipe = pipe.to(device)

MAX_SEED = np.iinfo(np.int32).max
MAX_IMAGE_SIZE = 1216

@spaces.GPU
def infer(
 prompt,
 negative_prompt="",
 seed=42,
 randomize_seed=False,
 width=1024,
 height=1024,
 guidance_scale=0.0,
 num_inference_steps=4,
 progress=gr.Progress(track_tqdm=True),
):
 if randomize_seed:
 seed = random.randint(0, MAX_SEED)

 generator = torch.Generator().manual_seed(seed)

 image = pipe(
 prompt=prompt,
 negative_prompt=negative_prompt,
 guidance_scale=guidance_scale,
 num_inference_steps=num_inference_steps,
 width=width,
 height=height,
 generator=generator,
 ).images[0]

 return image, seed


examples = [
 "A capybara wearing a suit holding a sign that reads Hello World",
]

css = """
#col-container {
 margin: 0 auto;
 max-width: 640px;
}
"""

with gr.Blocks(css=css) as demo:
 with gr.Column(elem_id="col-container"):
 gr.Markdown(" # [Stable Diffusion 3.5 Large Turbo (8B)](https://huggingface.co/stabilityai/stable-diffusion-3.5-large-turbo)")
 gr.Markdown("Space for testing alternative text encoders with SD 3.5 L Turbo")
 with gr.Row():
 prompt = gr.Textbox(
 label="Prompt",
 show_label=False,
 max_lines=4,
 lines=4,
 placeholder="Enter your prompt",
 container=False,
 )

 run_button = gr.Button("Run", scale=0, variant="primary")

 result = gr.Image(label="Result", show_label=False)

 with gr.Accordion("Advanced Settings", open=False):
 negative_prompt = gr.Textbox(
 label="Negative prompt",
 max_lines=2,
 lines=2,
 placeholder="Enter a negative prompt",
 visible=True,
 )

 seed = gr.Slider(
 label="Seed",
 minimum=0,
 maximum=MAX_SEED,
 step=1,
 value=0,
 )

 randomize_seed = gr.Checkbox(label="Randomize seed", value=True)

 with gr.Row():
 width = gr.Slider(
 label="Width",
 minimum=512,
 maximum=MAX_IMAGE_SIZE,
 step=32,
 value=1024, 
 )

 height = gr.Slider(
 label="Height",
 minimum=512,
 maximum=MAX_IMAGE_SIZE,
 step=32,
 value=1024,
 )

 with gr.Row():
 guidance_scale = gr.Slider(
 label="Guidance scale",
 minimum=0.0,
 maximum=7.5,
 step=0.1,
 value=0.0,
 )

 num_inference_steps = gr.Slider(
 label="Number of inference steps",
 minimum=1,
 maximum=50,
 step=1,
 value=4, 
 )

 gr.Examples(examples=examples, inputs=[prompt], outputs=[result, seed], fn=infer, cache_examples=True, cache_mode="lazy")
 gr.on(
 triggers=[run_button.click, prompt.submit],
 fn=infer,
 inputs=[
 prompt,
 negative_prompt,
 seed,
 randomize_seed,
 width,
 height,
 guidance_scale,
 num_inference_steps,
 ],
 outputs=[result, seed],
 )

if __name__ == "__main__":
 demo.launch()