Spaces:
Running
on
Zero
Running
on
Zero
File size: 5,122 Bytes
d0852ef 1486d19 d0852ef d74c6d6 d0852ef 1486d19 d74c6d6 d0852ef 1486d19 d0852ef 1486d19 d0852ef 1486d19 d0852ef 1486d19 d0852ef d74c6d6 d0852ef 1486d19 d0852ef 1486d19 d0852ef 1486d19 d0852ef 1486d19 d0852ef 1486d19 d0852ef 1486d19 d0852ef 1486d19 d0852ef 1486d19 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 |
import spaces
import gradio as gr
import numpy as np
import random
from diffusers import DiffusionPipeline
import torch
import random
from diffusers import (
ControlNetModel,
DiffusionPipeline,
StableDiffusionControlNetPipeline,
StableDiffusionXLControlNetPipeline,
UniPCMultistepScheduler,
EulerDiscreteScheduler,
AutoencoderKL
)
from transformers import DPTFeatureExtractor, DPTForDepthEstimation, DPTImageProcessor
from transformers import CLIPImageProcessor
from diffusers.utils import load_image
from gradio_imageslider import ImageSlider
device = "cuda"
base_model_id = "SG161222/RealVisXL_V4.0"
controlnet_model_id = "diffusers/controlnet-depth-sdxl-1.0"
vae_model_id = "madebyollin/sdxl-vae-fp16-fix"
if torch.cuda.is_available():
# load pipe
controlnet = ControlNetModel.from_pretrained(controlnet_model_id, variant="fp16", use_safetensors=True, torch_dtype=torch.float16)
vae = AutoencoderKL.from_pretrained(vae_model_id, torch_dtype=torch.float16)
pipe = StableDiffusionXLControlNetPipeline.from_pretrained(
base_model_id,
controlnet=controlnet,
vae=vae,
variant="fp16",
use_safetensors=True,
torch_dtype=torch.float16,
)
pipe.scheduler = EulerDiscreteScheduler.from_config(pipe.scheduler.config)
pipe.to(device)
depth_estimator = DPTForDepthEstimation.from_pretrained("Intel/dpt-hybrid-midas").to("cuda")
feature_extractor = DPTImageProcessor.from_pretrained("Intel/dpt-hybrid-midas")
MAX_SEED = np.iinfo(np.int32).max
MAX_IMAGE_SIZE = 1024
USE_TORCH_COMPILE = 0
ENABLE_CPU_OFFLOAD = 0
def randomize_seed_fn(seed: int, randomize_seed: bool) -> int:
if randomize_seed:
seed = random.randint(0, MAX_SEED)
return seed
def get_depth_map(image):
image = feature_extractor(images=image, return_tensors="pt").pixel_values.to("cuda")
with torch.no_grad(), torch.autocast("cuda"):
depth_map = depth_estimator(image).predicted_depth
depth_map = torch.nn.functional.interpolate(
depth_map.unsqueeze(1),
size=(1024, 1024),
mode="bicubic",
align_corners=False,
)
depth_min = torch.amin(depth_map, dim=[1, 2, 3], keepdim=True)
depth_max = torch.amax(depth_map, dim=[1, 2, 3], keepdim=True)
depth_map = (depth_map - depth_min) / (depth_max - depth_min)
image = torch.cat([depth_map] * 3, dim=1)
image = image.permute(0, 2, 3, 1).cpu().numpy()[0]
image = Image.fromarray((image * 255.0).clip(0, 255).astype(np.uint8))
return image
@spaces.GPU(enable_queue=True)
def process(orginal_image, image_url, prompt, n_prompt, num_steps, guidance_scale, control_strength, seed):
if image_url:
orginal_image = load_image(image_url)
width = 1024
height = 1024
depth_image = get_depth_map(orginal_image.resize((1024, 1024)))
generator = torch.Generator().manual_seed(seed)
generated_image = self.pipe(
prompt=prompt,
negative_prompt=n_prompt,
width=width,
height=height,
guidance_scale=guidance_scale,
num_inference_steps=num_steps,
strength=control_strength,
generator=generator,
image=depth_image,
).images[0]
return [[depth_image, generated_image], "ok"]
with gr.Blocks() as demo:
with gr.Row():
with gr.Column():
image = gr.Image()
image_url = gr.Textbox(label="Image Url", placeholder="Enter image URL here (optional)")
prompt = gr.Textbox(label="Prompt")
run_button = gr.Button("Run")
with gr.Accordion("Advanced options", open=True):
num_steps = gr.Slider(label="Number of steps", minimum=1, maximum=100, value=30, step=1)
guidance_scale = gr.Slider(label="Guidance scale", minimum=0.1, maximum=30.0, value=7.5, step=0.1)
control_strength = gr.Slider(label="Control Strength", minimum=0.1, maximum=4.0, value=0.8, step=0.1)
seed = gr.Slider(label="Seed", minimum=0, maximum=MAX_SEED, step=1, value=0)
randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
n_prompt = gr.Textbox(
label="Negative prompt",
value="longbody, lowres, bad anatomy, bad hands, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality",
)
with gr.Column():
result = ImageSlider(label="Generate image", type="pil", slider_color="pink")
logs = gr.Textbox(label="logs")
inputs = [
image,
image_url,
prompt,
n_prompt,
num_steps,
guidance_scale,
control_strength,
seed
]
run_button.click(
fn=randomize_seed_fn,
inputs=[seed, randomize_seed],
outputs=seed,
queue=False,
api_name=False,
).then(
fn=process,
inputs=inputs,
outputs=[result, logs],
api_name=False
)
demo.queue().launch() |