|
import spaces |
|
import random |
|
import torch |
|
import cv2 |
|
import gradio as gr |
|
import numpy as np |
|
from huggingface_hub import snapshot_download |
|
from transformers import CLIPVisionModelWithProjection,CLIPImageProcessor |
|
from diffusers.utils import load_image |
|
from kolors.pipelines.pipeline_controlnet_xl_kolors_img2img import StableDiffusionXLControlNetImg2ImgPipeline |
|
from kolors.models.modeling_chatglm import ChatGLMModel |
|
from kolors.models.tokenization_chatglm import ChatGLMTokenizer |
|
from kolors.models.controlnet import ControlNetModel |
|
from diffusers import AutoencoderKL |
|
from kolors.models.unet_2d_condition import UNet2DConditionModel |
|
from diffusers import EulerDiscreteScheduler |
|
from PIL import Image |
|
from annotator.midas import MidasDetector |
|
from annotator.dwpose import DWposeDetector |
|
from annotator.util import resize_image, HWC3 |
|
|
|
|
|
device = "cuda" |
|
ckpt_dir = snapshot_download(repo_id="Kwai-Kolors/Kolors") |
|
ckpt_dir_depth = snapshot_download(repo_id="Kwai-Kolors/Kolors-ControlNet-Depth") |
|
ckpt_dir_canny = snapshot_download(repo_id="Kwai-Kolors/Kolors-ControlNet-Canny") |
|
ckpt_dir_pose = snapshot_download(repo_id="Kwai-Kolors/Kolors-ControlNet-Pose") |
|
|
|
text_encoder = ChatGLMModel.from_pretrained(f'{ckpt_dir}/text_encoder', torch_dtype=torch.float16).half().to(device) |
|
tokenizer = ChatGLMTokenizer.from_pretrained(f'{ckpt_dir}/text_encoder') |
|
vae = AutoencoderKL.from_pretrained(f"{ckpt_dir}/vae", revision=None).half().to(device) |
|
scheduler = EulerDiscreteScheduler.from_pretrained(f"{ckpt_dir}/scheduler") |
|
unet = UNet2DConditionModel.from_pretrained(f"{ckpt_dir}/unet", revision=None).half().to(device) |
|
controlnet_depth = ControlNetModel.from_pretrained(f"{ckpt_dir_depth}", revision=None).half().to(device) |
|
controlnet_canny = ControlNetModel.from_pretrained(f"{ckpt_dir_canny}", revision=None).half().to(device) |
|
controlnet_pose = ControlNetModel.from_pretrained(f"{ckpt_dir_pose}", revision=None).half().to(device) |
|
|
|
pipe_depth = StableDiffusionXLControlNetImg2ImgPipeline( |
|
vae=vae, |
|
controlnet = controlnet_depth, |
|
text_encoder=text_encoder, |
|
tokenizer=tokenizer, |
|
unet=unet, |
|
scheduler=scheduler, |
|
force_zeros_for_empty_prompt=False |
|
) |
|
|
|
pipe_canny = StableDiffusionXLControlNetImg2ImgPipeline( |
|
vae=vae, |
|
controlnet = controlnet_canny, |
|
text_encoder=text_encoder, |
|
tokenizer=tokenizer, |
|
unet=unet, |
|
scheduler=scheduler, |
|
force_zeros_for_empty_prompt=False |
|
) |
|
|
|
pipe_pose = StableDiffusionXLControlNetImg2ImgPipeline( |
|
vae=vae, |
|
controlnet = controlnet_pose, |
|
text_encoder=text_encoder, |
|
tokenizer=tokenizer, |
|
unet=unet, |
|
scheduler=scheduler, |
|
force_zeros_for_empty_prompt=False |
|
) |
|
|
|
@spaces.GPU |
|
def process_canny_condition(image, canny_threods=[100,200]): |
|
np_image = image.copy() |
|
np_image = cv2.Canny(np_image, canny_threods[0], canny_threods[1]) |
|
np_image = np_image[:, :, None] |
|
np_image = np.concatenate([np_image, np_image, np_image], axis=2) |
|
np_image = HWC3(np_image) |
|
return Image.fromarray(np_image) |
|
|
|
model_midas = MidasDetector() |
|
@spaces.GPU |
|
def process_depth_condition_midas(img, res = 1024): |
|
h,w,_ = img.shape |
|
img = resize_image(HWC3(img), res) |
|
result = HWC3(model_midas(img)) |
|
result = cv2.resize(result, (w,h)) |
|
return Image.fromarray(result) |
|
|
|
model_dwpose = DWposeDetector() |
|
@spaces.GPU |
|
def process_dwpose_condition(image, res=1024): |
|
h,w,_ = image.shape |
|
img = resize_image(HWC3(image), res) |
|
out_res, out_img = model_dwpose(image) |
|
result = HWC3(out_img) |
|
result = cv2.resize( result, (w,h) ) |
|
return Image.fromarray(result) |
|
|
|
MAX_SEED = np.iinfo(np.int32).max |
|
MAX_IMAGE_SIZE = 1024 |
|
|
|
@spaces.GPU |
|
def infer_depth(prompt, |
|
image = None, |
|
negative_prompt = "nsfwοΌθΈι¨ι΄ε½±οΌδ½εθΎ¨ηοΌjpegδΌͺε½±γ樑η³γη³η³οΌι»θΈοΌιθΉη―", |
|
seed = 397886929, |
|
randomize_seed = False, |
|
guidance_scale = 6.0, |
|
num_inference_steps = 50, |
|
controlnet_conditioning_scale = 0.7, |
|
control_guidance_end = 0.9, |
|
strength = 1.0 |
|
): |
|
if randomize_seed: |
|
seed = random.randint(0, MAX_SEED) |
|
generator = torch.Generator().manual_seed(seed) |
|
init_image = resize_image(image, MAX_IMAGE_SIZE) |
|
pipe = pipe_depth.to("cuda") |
|
condi_img = process_depth_condition_midas( np.array(init_image), MAX_IMAGE_SIZE) |
|
image = pipe( |
|
prompt= prompt , |
|
image = init_image, |
|
controlnet_conditioning_scale = controlnet_conditioning_scale, |
|
control_guidance_end = control_guidance_end, |
|
strength= strength , |
|
control_image = condi_img, |
|
negative_prompt= negative_prompt , |
|
num_inference_steps= num_inference_steps, |
|
guidance_scale= guidance_scale, |
|
num_images_per_prompt=1, |
|
generator=generator, |
|
).images[0] |
|
return [condi_img, image], seed |
|
|
|
@spaces.GPU |
|
def infer_canny(prompt, |
|
image = None, |
|
negative_prompt = "nsfwοΌθΈι¨ι΄ε½±οΌδ½εθΎ¨ηοΌjpegδΌͺε½±γ樑η³γη³η³οΌι»θΈοΌιθΉη―", |
|
seed = 397886929, |
|
randomize_seed = False, |
|
guidance_scale = 6.0, |
|
num_inference_steps = 50, |
|
controlnet_conditioning_scale = 0.7, |
|
control_guidance_end = 0.9, |
|
strength = 1.0 |
|
): |
|
if randomize_seed: |
|
seed = random.randint(0, MAX_SEED) |
|
generator = torch.Generator().manual_seed(seed) |
|
init_image = resize_image(image, MAX_IMAGE_SIZE) |
|
pipe = pipe_canny.to("cuda") |
|
condi_img = process_canny_condition(np.array(init_image)) |
|
image = pipe( |
|
prompt= prompt , |
|
image = init_image, |
|
controlnet_conditioning_scale = controlnet_conditioning_scale, |
|
control_guidance_end = control_guidance_end, |
|
strength= strength , |
|
control_image = condi_img, |
|
negative_prompt= negative_prompt , |
|
num_inference_steps= num_inference_steps, |
|
guidance_scale= guidance_scale, |
|
num_images_per_prompt=1, |
|
generator=generator, |
|
).images[0] |
|
return [condi_img, image], seed |
|
|
|
@spaces.GPU |
|
def infer_pose(prompt, |
|
image = None, |
|
negative_prompt = "nsfwοΌθΈι¨ι΄ε½±οΌδ½εθΎ¨ηοΌjpegδΌͺε½±γ樑η³γη³η³οΌι»θΈοΌιθΉη―", |
|
seed = 66, |
|
randomize_seed = False, |
|
guidance_scale = 6.0, |
|
num_inference_steps = 50, |
|
controlnet_conditioning_scale = 0.7, |
|
control_guidance_end = 0.9, |
|
strength = 1.0 |
|
): |
|
if randomize_seed: |
|
seed = random.randint(0, MAX_SEED) |
|
generator = torch.Generator().manual_seed(seed) |
|
init_image = resize_image(image, MAX_IMAGE_SIZE) |
|
pipe = pipe_pose.to("cuda") |
|
condi_img = process_dwpose_condition(np.array(init_image), MAX_IMAGE_SIZE) |
|
image = pipe( |
|
prompt= prompt , |
|
image = init_image, |
|
controlnet_conditioning_scale = controlnet_conditioning_scale, |
|
control_guidance_end = control_guidance_end, |
|
strength= strength , |
|
control_image = condi_img, |
|
negative_prompt= negative_prompt , |
|
num_inference_steps= num_inference_steps, |
|
guidance_scale= guidance_scale, |
|
num_images_per_prompt=1, |
|
generator=generator, |
|
).images[0] |
|
return [condi_img, image], seed |
|
|
|
|
|
canny_examples = [ |
|
["μλ¦λ€μ΄ μλ
, κ³ νμ§, λ§€μ° μ λͺ
, μμν μμ, μ΄κ³ ν΄μλ, μ΅μμ νμ§, 8k, κ³ νμ§, 4K", |
|
"image/woman_1.png"], |
|
["νλ
ΈλΌλ§, μ»΅ μμ μμμλ κ·μ¬μ΄ ν° κ°μμ§, μΉ΄λ©λΌλ₯Ό λ°λΌλ³΄λ, μ λλ©μ΄μ
μ€νμΌ, 3D λ λλ§, μ₯ν
μΈ λ λ", |
|
"image/dog.png"] |
|
] |
|
|
|
depth_examples = [ |
|
["μ μΉ΄μ΄ λ§μ½ν μ€νμΌ, νλΆν μκ°, μ΄λ‘ μ
μΈ λ₯Ό μ
μ μ¬μ±μ΄ λ€νμ μ μλ, μλ¦λ€μ΄ νκ²½, λ§κ³ λ°μ, μΌλ£©μ§ λΉκ³Ό κ·Έλ¦Όμ, μ΅κ³ μ νμ§, μ΄μΈλ°, 8K νμ§", |
|
"image/woman_2.png"], |
|
["νλ €ν μμμ μμ μ, κ³ νμ§, λ§€μ° μ λͺ
, μμν μμ, μ΄κ³ ν΄μλ, μ΅μμ νμ§, 8k, κ³ νμ§, 4K", |
|
"image/bird.png"] |
|
] |
|
|
|
pose_examples = [ |
|
["보λΌμ νΌν μ¬λ¦¬λΈ λλ μ€λ₯Ό μ
κ³ μκ΄κ³Ό ν°μ λ μ΄μ€ μ₯κ°μ λ μλ
κ° μ μμΌλ‘ μΌκ΅΄μ κ°μΈκ³ μλ, κ³ νμ§, λ§€μ° μ λͺ
, μμν μμ, μ΄κ³ ν΄μλ, μ΅μμ νμ§, 8k, κ³ νμ§, 4K", |
|
"image/woman_3.png"], |
|
["κ²μμ μ€ν¬μΈ μ¬ν·κ³Ό ν°μ μ΄λλ₯Ό μ
κ³ λͺ©κ±Έμ΄λ₯Ό ν μ¬μ±μ΄ 거리μ μ μλ, λ°°κ²½μ λΉ¨κ° κ±΄λ¬Όκ³Ό λ
Ήμ λ무, κ³ νμ§, λ§€μ° μ λͺ
, μμν μμ, μ΄κ³ ν΄μλ, μ΅μμ νμ§, 8k, κ³ νμ§, 4K", |
|
"image/woman_4.png"] |
|
] |
|
|
|
css = """ |
|
footer { |
|
visibility: hidden; |
|
} |
|
""" |
|
|
|
|
|
def load_description(fp): |
|
with open(fp, 'r', encoding='utf-8') as f: |
|
content = f.read() |
|
return content |
|
|
|
with gr.Blocks(theme="Nymbo/Nymbo_Theme", css=css) as Kolors: |
|
with gr.Row(): |
|
with gr.Column(elem_id="col-left"): |
|
with gr.Row(): |
|
prompt = gr.Textbox( |
|
label="ν둬ννΈ", |
|
placeholder="ν둬ννΈλ₯Ό μ
λ ₯νμΈμ", |
|
lines=2 |
|
) |
|
with gr.Row(): |
|
image = gr.Image(label="μ΄λ―Έμ§", type="pil") |
|
with gr.Accordion("κ³ κΈ μ€μ ", open=False): |
|
negative_prompt = gr.Textbox( |
|
label="λ€κ±°ν°λΈ ν둬ννΈ", |
|
placeholder="λ€κ±°ν°λΈ ν둬ννΈλ₯Ό μ
λ ₯νμΈμ", |
|
visible=True, |
|
value="nsfw, μΌκ΅΄ κ·Έλ¦Όμ, μ ν΄μλ, jpeg μν°ν©νΈ, νλ¦Ών¨, μ΄μ
ν¨, κ²μ μΌκ΅΄, λ€μ¨ μ‘°λͺ
" |
|
) |
|
seed = gr.Slider( |
|
label="μλ", |
|
minimum=0, |
|
maximum=MAX_SEED, |
|
step=1, |
|
value=0, |
|
) |
|
randomize_seed = gr.Checkbox(label="μλ 무μμν", value=True) |
|
with gr.Row(): |
|
guidance_scale = gr.Slider( |
|
label="κ°μ΄λμ€ μ€μΌμΌ", |
|
minimum=0.0, |
|
maximum=10.0, |
|
step=0.1, |
|
value=6.0, |
|
) |
|
num_inference_steps = gr.Slider( |
|
label="μΆλ‘ λ¨κ³ μ", |
|
minimum=10, |
|
maximum=50, |
|
step=1, |
|
value=30, |
|
) |
|
with gr.Row(): |
|
controlnet_conditioning_scale = gr.Slider( |
|
label="컨νΈλ‘€λ· 컨λμ
λ μ€μΌμΌ", |
|
minimum=0.0, |
|
maximum=1.0, |
|
step=0.1, |
|
value=0.7, |
|
) |
|
control_guidance_end = gr.Slider( |
|
label="컨νΈλ‘€ κ°μ΄λμ€ μ’
λ£", |
|
minimum=0.0, |
|
maximum=1.0, |
|
step=0.1, |
|
value=0.9, |
|
) |
|
with gr.Row(): |
|
strength = gr.Slider( |
|
label="κ°λ", |
|
minimum=0.0, |
|
maximum=1.0, |
|
step=0.1, |
|
value=1.0, |
|
) |
|
with gr.Row(): |
|
canny_button = gr.Button("μΊλ", elem_id="button") |
|
depth_button = gr.Button("κΉμ΄", elem_id="button") |
|
pose_button = gr.Button("ν¬μ¦", elem_id="button") |
|
|
|
with gr.Column(elem_id="col-right"): |
|
result = gr.Gallery(label="κ²°κ³Ό", show_label=False, columns=2) |
|
seed_used = gr.Number(label="μ¬μ©λ μλ") |
|
|
|
|
|
|
|
with gr.Row(): |
|
gr.Examples( |
|
fn = infer_canny, |
|
examples = canny_examples, |
|
inputs = [prompt, image], |
|
outputs = [result, seed_used], |
|
label = "Canny" |
|
) |
|
with gr.Row(): |
|
gr.Examples( |
|
fn = infer_depth, |
|
examples = depth_examples, |
|
inputs = [prompt, image], |
|
outputs = [result, seed_used], |
|
label = "Depth" |
|
) |
|
|
|
with gr.Row(): |
|
gr.Examples( |
|
fn = infer_pose, |
|
examples = pose_examples, |
|
inputs = [prompt, image], |
|
outputs = [result, seed_used], |
|
label = "Pose" |
|
) |
|
|
|
canny_button.click( |
|
fn = infer_canny, |
|
inputs = [prompt, image, negative_prompt, seed, randomize_seed, guidance_scale, num_inference_steps, controlnet_conditioning_scale, control_guidance_end, strength], |
|
outputs = [result, seed_used] |
|
) |
|
|
|
depth_button.click( |
|
fn = infer_depth, |
|
inputs = [prompt, image, negative_prompt, seed, randomize_seed, guidance_scale, num_inference_steps, controlnet_conditioning_scale, control_guidance_end, strength], |
|
outputs = [result, seed_used] |
|
) |
|
|
|
pose_button.click( |
|
fn = infer_pose, |
|
inputs = [prompt, image, negative_prompt, seed, randomize_seed, guidance_scale, num_inference_steps, controlnet_conditioning_scale, control_guidance_end, strength], |
|
outputs = [result, seed_used] |
|
) |
|
|
|
Kolors.queue().launch(debug=True) |
|
|