gen_compare_v11/gen_reference_image_for_inpaint.py

# from https://github.com/lllyasviel/ControlNet/blob/main/gradio_canny2image.py

import einops
import numpy as np
import torch
from PIL import Image
import sys
import os
import yaml

CONTROL_NET_PATH = '/home/takuma/Documents/co/ControlNet-v1-1-nightly/'
CONTROL_NET_MODEL_PATH = '../../ControlNet-v1-1'
sys.path.append(CONTROL_NET_PATH)

from share import *
from pytorch_lightning import seed_everything
from cldm.model import create_model, load_state_dict
from cldm.ddim_hacked import DDIMSampler
from diffusers.utils import load_image

test_prompt = "best quality, extremely detailed"
test_negative_prompt = "lowres, bad anatomy, worst quality, low quality"

@torch.no_grad()
def generate(prompt, n_prompt, seed, control, ddim_steps=20, eta=0.0, scale=9.0, H=512, W=512, strength = 1.0, guess_mode=False):
    seed_everything(seed)

    cond = {"c_concat": [control], "c_crossattn": [model.get_learned_conditioning([prompt] * num_samples)]}
    un_cond = {"c_concat": None if guess_mode else [control], "c_crossattn": [model.get_learned_conditioning([n_prompt] * num_samples)]}
    shape = (4, H // 8, W // 8)
    
    model.control_scales = [strength * (0.825 ** float(12 - i)) for i in range(13)] if guess_mode else ([strength] * 13)  # Magic number. IDK why. Perhaps because 0.825**12<0.01 but 0.826**12>0.01
    latent = torch.randn((1,) + shape, device="cpu", generator=torch.Generator(device="cpu").manual_seed(seed)).cuda()
    samples, intermediates = ddim_sampler.sample(ddim_steps, num_samples,
                                                    shape, cond, x_T=latent,
                                                    verbose=False, eta=eta,
                                                    unconditional_guidance_scale=scale,
                                                    unconditional_conditioning=un_cond)
    x_samples = model.decode_first_stage(samples)
    x_samples = (einops.rearrange(x_samples, 'b c h w -> b h w c') * 127.5 + 127.5).cpu().numpy().clip(0, 255).astype(np.uint8)
 
    return Image.fromarray(x_samples[0])

def control_images(control_image_folder, model_name):
    with open('./control_images.yaml', 'r') as f:
        d = yaml.safe_load(f)
    filenames = d[model_name]
    return [Image.open(f'{control_image_folder}/{fn}').convert("RGB") for fn in filenames]

def make_image_condition(image, image_mask=None):
    image = np.array(image.convert("RGB")).astype(np.float32) / 255.0
    if image_mask is not None:
        image_mask = np.array(image_mask.convert("L"))
        assert (
            image.shape[0:1] == image_mask.shape[0:1]
        ), "image and image_mask must have the same image size"
        image[image_mask < 128] = -1.0  # set as masked pixel
    image = np.expand_dims(image, 0).transpose(0, 3, 1, 2)
    image = torch.from_numpy(image)
    return image

if __name__ == '__main__':
    model_name = "p_sd15_inpaint"

    original_image_folder = "./control_images/"
    control_image_folder = './control_images/converted/'
    output_image_folder = './output_images/ref/'
    os.makedirs(output_image_folder, exist_ok=True)

    if model_name == 'p_sd15s2_lineart_anime':
        base_model_file = 'anything-v3-full.safetensors'
    else:
        base_model_file = 'v1-5-pruned.ckpt'

    num_samples = 1
    model = create_model(f'{CONTROL_NET_MODEL_PATH}/control_v11{model_name}.yaml').cpu()
    model.load_state_dict(load_state_dict(f'{CONTROL_NET_PATH}/models/{base_model_file}', location='cuda'), strict=False)
    model.load_state_dict(load_state_dict(f'{CONTROL_NET_MODEL_PATH}/control_v11{model_name}.pth', location='cuda'), strict=False)
    model = model.cuda()
    ddim_sampler = DDIMSampler(model)

    original_image_filenames = [
        "pexels-sound-on-3760767_512x512.png",
        "vermeer_512x512.png",
        "bird_512x512.png",
    ]

    inpaint_image_conditions = [
        make_image_condition(
            Image.open(f"{original_image_folder}{fn}"),
            Image.open(f"{original_image_folder}mask_512x512.png"),
        )
        for fn in original_image_filenames
    ]

    for i, control in enumerate(inpaint_image_conditions):
        # control = np.array(control)[:,:,::-1].copy()
        # control = torch.from_numpy(control).float().cuda() / 255.0
        # control = torch.stack([control for _ in range(num_samples)], dim=0)
        # control = einops.rearrange(control, 'b h w c -> b c h w').clone()
        control = control.cuda()
        for seed in range(4):
            image = generate(test_prompt, test_negative_prompt, seed=seed, control=control)
            image.save(f'{output_image_folder}output_{model_name}_{i}_{seed}.png')