Spaces:
Runtime error
Runtime error
QRCode pipeline
Browse files- qr-code.png +0 -0
- server/pipelines/controlnetLoraSD15QRCode.py +239 -0
qr-code.png
ADDED
|
server/pipelines/controlnetLoraSD15QRCode.py
ADDED
|
@@ -0,0 +1,239 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from diffusers import (
|
| 2 |
+
StableDiffusionControlNetImg2ImgPipeline,
|
| 3 |
+
ControlNetModel,
|
| 4 |
+
LCMScheduler,
|
| 5 |
+
AutoencoderTiny,
|
| 6 |
+
)
|
| 7 |
+
from compel import Compel
|
| 8 |
+
import torch
|
| 9 |
+
|
| 10 |
+
try:
|
| 11 |
+
import intel_extension_for_pytorch as ipex # type: ignore
|
| 12 |
+
except:
|
| 13 |
+
pass
|
| 14 |
+
|
| 15 |
+
import psutil
|
| 16 |
+
from config import Args
|
| 17 |
+
from pydantic import BaseModel, Field
|
| 18 |
+
from PIL import Image
|
| 19 |
+
import math
|
| 20 |
+
|
| 21 |
+
taesd_model = "madebyollin/taesd"
|
| 22 |
+
controlnet_model = "monster-labs/control_v1p_sd15_qrcode_monster"
|
| 23 |
+
base_model = "nitrosocke/mo-di-diffusion"
|
| 24 |
+
lcm_lora_id = "latent-consistency/lcm-lora-sdv1-5"
|
| 25 |
+
default_prompt = "abstract art of a men with curly hair by Pablo Picasso"
|
| 26 |
+
page_content = """
|
| 27 |
+
<h1 class="text-3xl font-bold">Real-Time Latent Consistency Model SDv1.5</h1>
|
| 28 |
+
<h3 class="text-xl font-bold">LCM + LoRA + Controlnet + QRCode</h3>
|
| 29 |
+
<p class="text-sm">
|
| 30 |
+
This demo showcases
|
| 31 |
+
<a
|
| 32 |
+
href="https://huggingface.co/blog/lcm_lora"
|
| 33 |
+
target="_blank"
|
| 34 |
+
class="text-blue-500 underline hover:no-underline">LCM LoRA</a>
|
| 35 |
+
+ ControlNet + Image to Imasge pipeline using
|
| 36 |
+
<a
|
| 37 |
+
href="https://huggingface.co/docs/diffusers/main/en/using-diffusers/lcm#performing-inference-with-lcm"
|
| 38 |
+
target="_blank"
|
| 39 |
+
class="text-blue-500 underline hover:no-underline">Diffusers</a
|
| 40 |
+
> with a MJPEG stream server.
|
| 41 |
+
</p>
|
| 42 |
+
<p class="text-sm text-gray-500">
|
| 43 |
+
Change the prompt to generate different images, accepts <a
|
| 44 |
+
href="https://github.com/damian0815/compel/blob/main/doc/syntax.md"
|
| 45 |
+
target="_blank"
|
| 46 |
+
class="text-blue-500 underline hover:no-underline">Compel</a
|
| 47 |
+
> syntax.
|
| 48 |
+
</p>
|
| 49 |
+
"""
|
| 50 |
+
|
| 51 |
+
|
| 52 |
+
class Pipeline:
|
| 53 |
+
class Info(BaseModel):
|
| 54 |
+
name: str = "controlnet+loras+sd15"
|
| 55 |
+
title: str = "LCM + LoRA + Controlnet"
|
| 56 |
+
description: str = "Generates an image from a text prompt"
|
| 57 |
+
input_mode: str = "image"
|
| 58 |
+
page_content: str = page_content
|
| 59 |
+
|
| 60 |
+
class InputParams(BaseModel):
|
| 61 |
+
prompt: str = Field(
|
| 62 |
+
default_prompt,
|
| 63 |
+
title="Prompt",
|
| 64 |
+
field="textarea",
|
| 65 |
+
id="prompt",
|
| 66 |
+
)
|
| 67 |
+
seed: int = Field(
|
| 68 |
+
2159232, min=0, title="Seed", field="seed", hide=True, id="seed"
|
| 69 |
+
)
|
| 70 |
+
steps: int = Field(
|
| 71 |
+
5, min=1, max=15, title="Steps", field="range", hide=True, id="steps"
|
| 72 |
+
)
|
| 73 |
+
width: int = Field(
|
| 74 |
+
512, min=2, max=15, title="Width", disabled=True, hide=True, id="width"
|
| 75 |
+
)
|
| 76 |
+
height: int = Field(
|
| 77 |
+
512, min=2, max=15, title="Height", disabled=True, hide=True, id="height"
|
| 78 |
+
)
|
| 79 |
+
guidance_scale: float = Field(
|
| 80 |
+
1.0,
|
| 81 |
+
min=0,
|
| 82 |
+
max=2,
|
| 83 |
+
step=0.001,
|
| 84 |
+
title="Guidance Scale",
|
| 85 |
+
field="range",
|
| 86 |
+
hide=True,
|
| 87 |
+
id="guidance_scale",
|
| 88 |
+
)
|
| 89 |
+
strength: float = Field(
|
| 90 |
+
0.6,
|
| 91 |
+
min=0.25,
|
| 92 |
+
max=1.0,
|
| 93 |
+
step=0.001,
|
| 94 |
+
title="Strength",
|
| 95 |
+
field="range",
|
| 96 |
+
hide=True,
|
| 97 |
+
id="strength",
|
| 98 |
+
)
|
| 99 |
+
controlnet_scale: float = Field(
|
| 100 |
+
1.0,
|
| 101 |
+
min=0,
|
| 102 |
+
max=1.0,
|
| 103 |
+
step=0.001,
|
| 104 |
+
title="Controlnet Scale",
|
| 105 |
+
field="range",
|
| 106 |
+
hide=True,
|
| 107 |
+
id="controlnet_scale",
|
| 108 |
+
)
|
| 109 |
+
controlnet_start: float = Field(
|
| 110 |
+
0.0,
|
| 111 |
+
min=0,
|
| 112 |
+
max=1.0,
|
| 113 |
+
step=0.001,
|
| 114 |
+
title="Controlnet Start",
|
| 115 |
+
field="range",
|
| 116 |
+
hide=True,
|
| 117 |
+
id="controlnet_start",
|
| 118 |
+
)
|
| 119 |
+
controlnet_end: float = Field(
|
| 120 |
+
1.0,
|
| 121 |
+
min=0,
|
| 122 |
+
max=1.0,
|
| 123 |
+
step=0.001,
|
| 124 |
+
title="Controlnet End",
|
| 125 |
+
field="range",
|
| 126 |
+
hide=True,
|
| 127 |
+
id="controlnet_end",
|
| 128 |
+
)
|
| 129 |
+
blend: float = Field(
|
| 130 |
+
0.1,
|
| 131 |
+
min=0.0,
|
| 132 |
+
max=1.0,
|
| 133 |
+
step=0.001,
|
| 134 |
+
title="Blend",
|
| 135 |
+
field="range",
|
| 136 |
+
hide=True,
|
| 137 |
+
id="blend",
|
| 138 |
+
)
|
| 139 |
+
|
| 140 |
+
def __init__(self, args: Args, device: torch.device, torch_dtype: torch.dtype):
|
| 141 |
+
controlnet_qrcode = ControlNetModel.from_pretrained(
|
| 142 |
+
controlnet_model, torch_dtype=torch_dtype, subfolder="v2"
|
| 143 |
+
).to(device)
|
| 144 |
+
|
| 145 |
+
if args.safety_checker:
|
| 146 |
+
self.pipe = StableDiffusionControlNetImg2ImgPipeline.from_pretrained(
|
| 147 |
+
base_model,
|
| 148 |
+
controlnet=controlnet_qrcode,
|
| 149 |
+
)
|
| 150 |
+
else:
|
| 151 |
+
self.pipe = StableDiffusionControlNetImg2ImgPipeline.from_pretrained(
|
| 152 |
+
base_model,
|
| 153 |
+
safety_checker=None,
|
| 154 |
+
controlnet=controlnet_qrcode,
|
| 155 |
+
)
|
| 156 |
+
|
| 157 |
+
self.control_image = Image.open(
|
| 158 |
+
"qr-code.png").convert("RGB").resize((512, 512))
|
| 159 |
+
|
| 160 |
+
self.pipe.scheduler = LCMScheduler.from_config(
|
| 161 |
+
self.pipe.scheduler.config)
|
| 162 |
+
self.pipe.set_progress_bar_config(disable=True)
|
| 163 |
+
if device.type != "mps":
|
| 164 |
+
self.pipe.unet.to(memory_format=torch.channels_last)
|
| 165 |
+
|
| 166 |
+
if args.taesd:
|
| 167 |
+
self.pipe.vae = AutoencoderTiny.from_pretrained(
|
| 168 |
+
taesd_model, torch_dtype=torch_dtype, use_safetensors=True
|
| 169 |
+
).to(device)
|
| 170 |
+
|
| 171 |
+
# Load LCM LoRA
|
| 172 |
+
self.pipe.load_lora_weights(lcm_lora_id, adapter_name="lcm")
|
| 173 |
+
self.pipe.to(device=device, dtype=torch_dtype).to(device)
|
| 174 |
+
if args.compel:
|
| 175 |
+
self.compel_proc = Compel(
|
| 176 |
+
tokenizer=self.pipe.tokenizer,
|
| 177 |
+
text_encoder=self.pipe.text_encoder,
|
| 178 |
+
truncate_long_prompts=False,
|
| 179 |
+
)
|
| 180 |
+
if args.torch_compile:
|
| 181 |
+
self.pipe.unet = torch.compile(
|
| 182 |
+
self.pipe.unet, mode="reduce-overhead", fullgraph=True
|
| 183 |
+
)
|
| 184 |
+
self.pipe.vae = torch.compile(
|
| 185 |
+
self.pipe.vae, mode="reduce-overhead", fullgraph=True
|
| 186 |
+
)
|
| 187 |
+
self.pipe(
|
| 188 |
+
prompt="warmup",
|
| 189 |
+
image=[Image.new("RGB", (512, 512))],
|
| 190 |
+
control_image=[Image.new("RGB", (512, 512))],
|
| 191 |
+
)
|
| 192 |
+
|
| 193 |
+
def predict(self, params: "Pipeline.InputParams") -> Image.Image:
|
| 194 |
+
generator = torch.manual_seed(params.seed)
|
| 195 |
+
|
| 196 |
+
prompt = f"modern disney style {params.prompt}"
|
| 197 |
+
prompt_embeds = None
|
| 198 |
+
prompt = params.prompt
|
| 199 |
+
if hasattr(self, "compel_proc"):
|
| 200 |
+
prompt_embeds = self.compel_proc(prompt)
|
| 201 |
+
prompt = None
|
| 202 |
+
|
| 203 |
+
steps = params.steps
|
| 204 |
+
strength = params.strength
|
| 205 |
+
if int(steps * strength) < 1:
|
| 206 |
+
steps = math.ceil(1 / max(0.10, strength))
|
| 207 |
+
|
| 208 |
+
blend_qr_image = Image.blend(
|
| 209 |
+
params.image,
|
| 210 |
+
self.control_image,
|
| 211 |
+
alpha=params.blend
|
| 212 |
+
)
|
| 213 |
+
results = self.pipe(
|
| 214 |
+
image=blend_qr_image,
|
| 215 |
+
control_image=self.control_image,
|
| 216 |
+
prompt=prompt,
|
| 217 |
+
prompt_embeds=prompt_embeds,
|
| 218 |
+
generator=generator,
|
| 219 |
+
strength=strength,
|
| 220 |
+
num_inference_steps=steps,
|
| 221 |
+
guidance_scale=params.guidance_scale,
|
| 222 |
+
width=params.width,
|
| 223 |
+
height=params.height,
|
| 224 |
+
output_type="pil",
|
| 225 |
+
controlnet_conditioning_scale=params.controlnet_scale,
|
| 226 |
+
control_guidance_start=params.controlnet_start,
|
| 227 |
+
control_guidance_end=params.controlnet_end,
|
| 228 |
+
)
|
| 229 |
+
|
| 230 |
+
nsfw_content_detected = (
|
| 231 |
+
results.nsfw_content_detected[0]
|
| 232 |
+
if "nsfw_content_detected" in results
|
| 233 |
+
else False
|
| 234 |
+
)
|
| 235 |
+
if nsfw_content_detected:
|
| 236 |
+
return None
|
| 237 |
+
result_image = results.images[0]
|
| 238 |
+
|
| 239 |
+
return result_image
|