File size: 7,006 Bytes
4efc065 744d735 4efc065 744d735 4efc065 744d735 4efc065 56e3148 0687973 4400ef7 bf36194 01f7239 bf36194 744d735 4efc065 4400ef7 6d4ad23 4efc065 6d4ad23 33d1a30 55ca821 4efc065 6d4ad23 86c6ead 6d4ad23 86c6ead 0129d6f 6d4ad23 86c6ead 6d4ad23 744d735 0129d6f 4efc065 79bb82a 14da090 01f7239 e849c2a 01f7239 4efc065 ed1bd2d 4efc065 0687973 14da090 dee1946 14da090 0687973 56e3148 14da090 744d735 4efc065 6d4ad23 0129d6f 91b2943 a678ae0 0129d6f 91b2943 0129d6f 9d60b6e 0129d6f 79bb82a 0129d6f 81e129b 0129d6f 91b2943 0129d6f 3f9c34e 0129d6f 81e129b 0129d6f 91b2943 4efc065 6d4ad23 744d735 7c87b47 744d735 7c87b47 744d735 4efc065 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 |
from typing import Dict, List, Any
import torch
from diffusers import DPMSolverMultistepScheduler, StableDiffusionInpaintPipeline, AutoPipelineForInpainting, AutoPipelineForImage2Image, StableDiffusionXLImg2ImgPipeline, StableDiffusionControlNetInpaintPipeline, ControlNetModel
from PIL import Image
import base64
from io import BytesIO
import numpy as np
# set device
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
if device.type != 'cuda':
raise ValueError("need to run on GPU")
def make_inpaint_condition(image, image_mask):
image = np.array(image.convert("RGB")).astype(np.float32) / 255.0
image_mask = np.array(image_mask.convert("L")).astype(np.float32) / 255.0
assert image.shape[0:1] == image_mask.shape[0:1], "image and image_mask must have the same image size"
image[image_mask > 0.5] = -1.0 # set as masked pixel
image = np.expand_dims(image, 0).transpose(0, 3, 1, 2)
image = torch.from_numpy(image)
return image
class EndpointHandler():
def __init__(self, path=""):
#self.fast_pipe = AutoPipelineForInpainting.from_pretrained("diffusers/stable-diffusion-xl-1.0-inpainting-0.1", torch_dtype=torch.float16, variant="fp16").to("cuda")
#self.generator = torch.Generator(device="cuda").manual_seed(0)
# self.smooth_pipe = StableDiffusionXLImg2ImgPipeline.from_pretrained(
# "stabilityai/stable-diffusion-xl-refiner-1.0", torch_dtype=torch.float16, variant="fp16", use_safetensors=True
# )
# self.smooth_pipe.to("cuda")
self.controlnet = ControlNetModel.from_pretrained(
"lllyasviel/control_v11p_sd15_inpaint", torch_dtype=torch.float16
)
self.pipe = StableDiffusionControlNetInpaintPipeline.from_pretrained(
"runwayml/stable-diffusion-v1-5", controlnet=controlnet, torch_dtype=torch.float16
)
self.pipe.scheduler = DDIMScheduler.from_config(pipe.scheduler.config)
self.pipe.enable_model_cpu_offload()
self.pipe.enable_xformers_memory_efficient_attention()
"""
# load StableDiffusionInpaintPipeline pipeline
self.pipe = AutoPipelineForInpainting.from_pretrained(
"runwayml/stable-diffusion-inpainting",
revision="fp16",
torch_dtype=torch.float16,
)
# use DPMSolverMultistepScheduler
self.pipe.scheduler = DPMSolverMultistepScheduler.from_config(self.pipe.scheduler.config)
self.pipe.enable_model_cpu_offload()
self.pipe.enable_xformers_memory_efficient_attention()
# move to device
#self.pipe = self.pipe.to(device)
self.pipe2 = AutoPipelineForInpainting.from_pretrained("stabilityai/stable-diffusion-xl-refiner-1.0", torch_dtype=torch.float16, variant="fp16", use_safetensors=True)
#self.pipe2.enable_model_cpu_offload()
self.pipe2.enable_xformers_memory_efficient_attention()
self.pipe2.to("cuda")
self.pipe3 = AutoPipelineForImage2Image.from_pipe(self.pipe2)
#self.pipe3.enable_model_cpu_offload()
self.pipe3.enable_xformers_memory_efficient_attention()
"""
def __call__(self, data: Any) -> List[List[Dict[str, float]]]:
"""
:param data: A dictionary contains `inputs` and optional `image` field.
:return: A dictionary with `image` field contains image in base64.
"""
encoded_image = data.pop("image", None)
encoded_mask_image = data.pop("mask_image", None)
prompt = data.pop("prompt", "")
negative_prompt = data.pop("negative_prompt", "")
method = data.pop("method", "slow")
strength = data.pop("strength", 0.2)
guidance_scale = data.pop("guidance_scale", 8.0)
num_inference_steps = data.pop("num_inference_steps", 20)
"""
if(method == "smooth"):
if encoded_image is not None:
image = self.decode_base64_image(encoded_image)
out = self.smooth_pipe(prompt, image=image).images[0]
return out
"""
# process image
if encoded_image is not None and encoded_mask_image is not None:
image = self.decode_base64_image(encoded_image).convert("RGB")
mask_image = self.decode_base64_image(encoded_mask_image).convert("RGB")
else:
image = None
mask_image = None
"""
if(method == "fast"):
image = self.fast_pipe(
prompt=prompt,
negative_prompt=negative_prompt,
image=image,
mask_image=mask_image,
guidance_scale=guidance_scale,
num_inference_steps=num_inference_steps, # steps between 15 and 30 work well for us
strength=strength, # make sure to use `strength` below 1.0
generator=self.generator,
).images[0]
return image
"""
#pipe = AutoPipelineForInpainting.from_pretrained("diffusers/stable-diffusion-xl-1.0-inpainting-0.1", torch_dtype=torch.float16, variant="fp16").to("cuda")
"""
# run inference pipeline
out = self.pipe(prompt=prompt, negative_prompt=negative_prompt, image=image, mask_image=mask_image)
print("1st pipeline part successful!")
image = out.images[0].resize((1024, 1024))
print("image resizing successful!")
image = self.pipe2(
prompt=prompt,
negative_prompt=negative_prompt,
image=image,
mask_image=mask_image,
guidance_scale=guidance_scale, #8.0
num_inference_steps=num_inference_steps, #100
strength=strength, #0.2
output_type="latent", # let's keep in latent to save some VRAM
).images[0]
print("2nd pipeline part successful!")
image2 = self.pipe3(
prompt=prompt,
image=image,
guidance_scale=guidance_scale, #8.0
num_inference_steps=num_inference_steps, #100
strength=strength, #0.2
).images[0]
print("3rd pipeline part successful!")
# return first generate PIL image
return image2
"""
control_image = make_inpaint_condition(image, mask_image)
# generate image
image = pipe(
prompt=prompt,
negative_prompt=negative_prompt,
num_inference_steps=num_inference_steps,
eta=1.0,
image=image,
mask_image=mask_image,
control_image=control_image,
guidance_scale,
strength=strength
).images[0]
return image
# helper to decode input image
def decode_base64_image(self, image_string):
base64_image = base64.b64decode(image_string)
buffer = BytesIO(base64_image)
image = Image.open(buffer)
return image
|