import gradio as gr
import numpy as np
import random
from diffusers import StableDiffusionXLPipeline, EulerAncestralDiscreteScheduler
import torch
from transformers import AutoTokenizer, AutoModelForCausalLM, GenerationConfig
import spaces
import os
from huggingface_hub import hf_hub_download
from safetensors.torch import load_file
device = "cuda"
token=os.environ["TOKEN"]
model_id="aipicasso/emix-0-5"
scheduler = EulerAncestralDiscreteScheduler.from_pretrained(model_id,subfolder="scheduler",token=token)
pipe = StableDiffusionXLPipeline.from_pretrained(model_id, scheduler=scheduler, torch_dtype=torch.bfloat16,token=token)
negative_ti_file = hf_hub_download(repo_id="Aikimi/unaestheticXL_Negative_TI", filename="unaestheticXLv31.safetensors")
state_dict = load_file(negative_ti_file)
pipe.load_textual_inversion(state_dict["clip_g"], token="unaestheticXLv31", text_encoder=pipe.text_encoder_2, tokenizer=pipe.tokenizer_2)
pipe.load_textual_inversion(state_dict["clip_l"], token="unaestheticXLv31", text_encoder=pipe.text_encoder, tokenizer=pipe.tokenizer)
pipe = pipe.to(device)
MODEL_NAME = "p1atdev/dart-v1-sft"
tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME, trust_remote_code=True) # trust_remote_code is required for tokenizer
model = AutoModelForCausalLM.from_pretrained(MODEL_NAME, torch_dtype=torch.bfloat16)
model=model.to(device)
MAX_SEED = np.iinfo(np.int32).max
MAX_IMAGE_SIZE = 2048
@spaces.GPU
def infer(seed, randomize_seed, width, height, guidance_scale, num_inference_steps):
prompt = "<|bos|>rating:sfw, rating:generaloriginal<|long|>1girl<|input_end|>"
inputs = tokenizer(prompt, return_tensors="pt").input_ids.to(device)
with torch.no_grad():
outputs = model.generate(inputs, generation_config=model.generation_config)
prompt=tokenizer.decode(outputs[0], skip_special_tokens=True).split("original, ")[1]
negative_prompt="unaestheticXLv31, 3d, photo, realism"
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, prompt
css="""
#col-container {
margin: 0 auto;
max-width: 520px;
}
"""
with gr.Blocks(css=css) as demo:
with gr.Column(elem_id="col-container"):
gr.Markdown(f"""
# 著作権のない画像
## Anime image with No copyright
Generateボタンを押し、画像を生成してください。この画像がいくらきれいであろうと著作権は誰にもありません。この画像は時刻を入力とした自然現象によって作られたものです。美しいとは何でしょうか。
""")
with gr.Row():
run_button = gr.Button("Generate", scale=0)
result = gr.Image(label="Result", show_label=False)
generated_prompt = gr.Textbox(label="Generated prompt", show_label=False, interactive=False)
with gr.Accordion("Advanced Settings", open=False):
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=64,
value=1024,
)
height = gr.Slider(
label="Height",
minimum=512,
maximum=MAX_IMAGE_SIZE,
step=64,
value=1024,
)
with gr.Row():
guidance_scale = gr.Slider(
label="Guidance scale",
minimum=1.0,
maximum=10.0,
step=0.1,
value=7.5,
)
num_inference_steps = gr.Slider(
label="Number of inference steps",
minimum=1,
maximum=30,
step=1,
value=20,
)
run_button.click(
fn = infer,
inputs = [seed, randomize_seed, width, height, guidance_scale, num_inference_steps],
outputs = [result,generated_prompt]
)
demo.queue().launch()