scripts/inference.py

# Copyright 2024 Adobe. All rights reserved.
import argparse, os, sys, glob
# sys.path.append('.')
import cv2
import torch
import numpy as np
from omegaconf import OmegaConf
from PIL import Image
from tqdm import tqdm, trange
from imwatermark import WatermarkEncoder
from itertools import islice
from einops import rearrange
from torchvision.utils import make_grid
import time
from pytorch_lightning import seed_everything
from torch import autocast
from contextlib import contextmanager, nullcontext
import torchvision
from ldm.util import instantiate_from_config
from ldm.models.diffusion.ddim import DDIMSampler
from ldm.models.diffusion.plms import PLMSSampler

from diffusers.pipelines.stable_diffusion.safety_checker import StableDiffusionSafetyChecker
from transformers import AutoFeatureExtractor
import clip
from torchvision.transforms import Resize
import json
wm = "Paint-by-Example"
wm_encoder = WatermarkEncoder()
wm_encoder.set_watermark('bytes', wm.encode('utf-8'))
safety_model_id = "CompVis/stable-diffusion-safety-checker"
safety_feature_extractor = AutoFeatureExtractor.from_pretrained(safety_model_id)
safety_checker = StableDiffusionSafetyChecker.from_pretrained(safety_model_id)

def chunk(it, size):
    it = iter(it)
    return iter(lambda: tuple(islice(it, size)), ())

def get_tensor_clip(normalize=True, toTensor=True):
    transform_list = []
    if toTensor:
        transform_list += [torchvision.transforms.ToTensor()]

    if normalize:
        transform_list += [torchvision.transforms.Normalize((0.48145466, 0.4578275, 0.40821073),
                                                (0.26862954, 0.26130258, 0.27577711))]
    return torchvision.transforms.Compose(transform_list)

def numpy_to_pil(images):
    """
    Convert a numpy image or a batch of images to a PIL image.
    """
    if images.ndim == 3:
        images = images[None, ...]
    images = (images * 255).round().astype("uint8")
    pil_images = [Image.fromarray(image) for image in images]

    return pil_images


def load_model_from_config(config, ckpt, verbose=False):
    print(f"Loading model from {ckpt}")
    pl_sd = torch.load(ckpt, map_location="cpu")
    if "global_step" in pl_sd:
        print(f"Global Step: {pl_sd['global_step']}")
    sd = pl_sd["state_dict"]
    model = instantiate_from_config(config.model)
    # print('NOTE: NO CHECKPOINT IS LOADED')
    m, u = model.load_state_dict(sd, strict=False)
    if len(m) > 0 and verbose:
        print("missing keys:")
        print(m)
    if len(u) > 0 and verbose:
        print("unexpected keys:")
        print(u)

    model.cuda()
    model.eval()
    return model


def put_watermark(img, wm_encoder=None):
    if wm_encoder is not None:
        img = cv2.cvtColor(np.array(img), cv2.COLOR_RGB2BGR)
        img = wm_encoder.encode(img, 'dwtDct')
        img = Image.fromarray(img[:, :, ::-1])
    return img


def load_replacement(x):
    try:
        hwc = x.shape
        y = Image.open("assets/rick.jpeg").convert("RGB").resize((hwc[1], hwc[0]))
        y = (np.array(y)/255.0).astype(x.dtype)
        assert y.shape == x.shape
        return y
    except Exception:
        return x


def check_safety(x_image):
    safety_checker_input = safety_feature_extractor(numpy_to_pil(x_image), return_tensors="pt")
    x_checked_image, has_nsfw_concept = safety_checker(images=x_image, clip_input=safety_checker_input.pixel_values)
    assert x_checked_image.shape[0] == len(has_nsfw_concept)
    for i in range(len(has_nsfw_concept)):
        if has_nsfw_concept[i]:
            x_checked_image[i] = load_replacement(x_checked_image[i])
    return x_checked_image, has_nsfw_concept

def get_tensor(normalize=True, toTensor=True):
    transform_list = []
    if toTensor:
        transform_list += [torchvision.transforms.ToTensor()]

    if normalize:
        transform_list += [torchvision.transforms.Normalize((0.5, 0.5, 0.5),
                                                (0.5, 0.5, 0.5))]
    transform_list += [
        torchvision.transforms.Resize(512),
        torchvision.transforms.CenterCrop(512)
    ]
    return torchvision.transforms.Compose(transform_list)

def get_tensor_clip(normalize=True, toTensor=True):
    transform_list = []
    if toTensor:
        transform_list += [torchvision.transforms.ToTensor()]

    if normalize:
        transform_list += [torchvision.transforms.Normalize((0.48145466, 0.4578275, 0.40821073),
                                                (0.26862954, 0.26130258, 0.27577711))]
    return torchvision.transforms.Compose(transform_list)


def main():
    parser = argparse.ArgumentParser()

    parser.add_argument(
        "--outdir",
        type=str,
        nargs="?",
        help="dir to write results to",
        default="outputs/txt2img-samples"
    )
    parser.add_argument(
        "--skip_grid",
        action='store_true',
        help="do not save a grid, only individual samples. Helpful when evaluating lots of samples",
    )
    parser.add_argument(
        "--skip_save",
        action='store_true',
        help="do not save individual samples. For speed measurements.",
    )
    parser.add_argument(
        "--ddim_steps",
        type=int,
        default=50,
        help="number of ddim sampling steps",
    )
    parser.add_argument(
        "--plms",
        action='store_true',
        help="use plms sampling",
    )
    parser.add_argument(
        "--fixed_code",
        action='store_true',
        help="if enabled, uses the same starting code across samples ",
    )
    parser.add_argument(
        "--ddim_eta",
        type=float,
        default=0.0,
        help="ddim eta (eta=0.0 corresponds to deterministic sampling",
    )
    parser.add_argument(
        "--n_iter",
        type=int,
        default=2,
        help="sample this often",
    )
    parser.add_argument(
        "--H",
        type=int,
        default=512,
        help="image height, in pixel space",
    )
    parser.add_argument(
        "--W",
        type=int,
        default=512,
        help="image width, in pixel space",
    )
    parser.add_argument(
        "--n_imgs",
        type=int,
        default=100,
        help="image width, in pixel space",
    )
    parser.add_argument(
        "--C",
        type=int,
        default=4,
        help="latent channels",
    )
    parser.add_argument(
        "--f",
        type=int,
        default=8,
        help="downsampling factor",
    )
    parser.add_argument(
        "--n_samples",
        type=int,
        default=1,
        help="how many samples to produce for each given reference image. A.k.a. batch size",
    )
    parser.add_argument(
        "--n_rows",
        type=int,
        default=0,
        help="rows in the grid (default: n_samples)",
    )
    parser.add_argument(
        "--scale",
        type=float,
        default=1,
        help="unconditional guidance scale: eps = eps(x, empty) + scale * (eps(x, cond) - eps(x, empty))",
    )
    parser.add_argument(
        "--config",
        type=str,
        default="",
        help="path to config which constructs model",
    )
    parser.add_argument(
        "--ckpt",
        type=str,
        default="",
        help="path to checkpoint of model",
    )
    parser.add_argument(
        "--seed",
        type=int,
        default=42,
        help="the seed (for reproducible sampling)",
    )
    parser.add_argument(
        "--precision",
        type=str,
        help="evaluate at this precision",
        choices=["full", "autocast"],
        default="autocast"
    )
    parser.add_argument(
        "--image_path",
        type=str,
        help="evaluate at this precision",
        default=""
    )
    parser.add_argument(
        "--mask_path",
        type=str,
        help="evaluate at this precision",
        default=""
    )
    parser.add_argument(
        "--reference_path",
        type=str,
        help="evaluate at this precision",
        default=""
    )
    opt = parser.parse_args()


    seed_everything(opt.seed)

    config = OmegaConf.load(f"{opt.config}")
    model = load_model_from_config(config, f"{opt.ckpt}")

    device = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")
    model = model.to(device)

    if opt.plms:
        sampler = PLMSSampler(model)
    else:
        sampler = DDIMSampler(model)

    os.makedirs(opt.outdir, exist_ok=True)
    outpath = opt.outdir


    batch_size = opt.n_samples
    n_rows = opt.n_rows if opt.n_rows > 0 else batch_size

    sample_path = os.path.join(outpath, "source")
    result_path = os.path.join(outpath, "results")
    grid_path=os.path.join(outpath, "grid")
    os.makedirs(sample_path, exist_ok=True)
    os.makedirs(result_path, exist_ok=True)
    os.makedirs(grid_path, exist_ok=True)


    start_code = None
    if opt.fixed_code:
        start_code = torch.randn([opt.n_samples, opt.C, opt.H // opt.f, opt.W // opt.f], device=device)

    precision_scope = autocast if opt.precision=="autocast" else nullcontext
    
    # split_path = ''
    # with open(split_path) as f:
    #     sample_paths = json.load(f)
    
    # np.random.seed(opt.seed)
    # np.random.shuffle(sample_paths)
    
    # print(sample_paths[0])
    # raise ValueError
    
    with torch.no_grad():
        with precision_scope("cuda"):
            for i in range(1):
                with model.ema_scope():
                    filename=os.path.basename(opt.image_path)
                    img_p = Image.open(opt.image_path).convert("RGB")
                    image_tensor = get_tensor()(img_p)
                    image_tensor = image_tensor.unsqueeze(0)
                    ref_p = Image.open(opt.reference_path).convert("RGB")
                    width, height = ref_p.size   # Get dimensions
                    new_width = min(width, height)
                    new_height = new_width

                    left = (width - new_width)/2
                    top = (height - new_height)/2
                    right = (width + new_width)/2
                    bottom = (height + new_height)/2

                    # Crop the center of the image
                    ref_p = ref_p.crop((left, top, right, bottom))
                    ref_p = ref_p.resize((224,224))
                    ref_tensor=get_tensor_clip()(ref_p)
                    ref_tensor = ref_tensor.unsqueeze(0)
                    mask=Image.open(opt.mask_path).convert("L")
                    mask = mask.crop((left, top, right, bottom))
                    mask = np.array(mask)[None,None]
                    mask = mask.astype(np.float32)/255.0
                    mask[mask < 0.5] = 0
                    mask[mask >= 0.5] = 1
                    mask_tensor = torch.from_numpy(mask)
                    inpaint_image = image_tensor#*mask_tensor
                    # mask_tensor = torch.ones_like(inpaint_image)
                    # mask_tensor = mask_tensor[:, :1] # TODO PLEASE COMMENT OUT SOON
                    print('inpaint image size', inpaint_image.shape)
                    test_model_kwargs={}
                    test_model_kwargs['inpaint_mask']=mask_tensor.to(device)
                    test_model_kwargs['inpaint_image']=inpaint_image.to(device)
                    ref_tensor=ref_tensor.to(device)
                    uc = None
                    if opt.scale != 1.0:
                        uc = model.learnable_vector
                    c = model.get_learned_conditioning(ref_tensor.to(torch.float16))
                    c = model.proj_out(c)
                    inpaint_mask=test_model_kwargs['inpaint_mask']
                    z_inpaint = model.encode_first_stage(test_model_kwargs['inpaint_image'])
                    z_inpaint = model.get_first_stage_encoding(z_inpaint).detach()
                    test_model_kwargs['inpaint_image']=z_inpaint
                    test_model_kwargs['inpaint_mask']=Resize([z_inpaint.shape[-2],z_inpaint.shape[-1]])(test_model_kwargs['inpaint_mask'])

                    shape = [opt.C, opt.H // opt.f, opt.W // opt.f]
                    
                    
                    # compute context here
                    
                    # ref_latent = model.encode_first_stage(img maybe)
                    # contexts = context_unet.compute_context()
                    
                    samples_ddim, _ = sampler.sample(S=opt.ddim_steps,
                                                        conditioning=c,
                                                        batch_size=opt.n_samples,
                                                        shape=shape,
                                                        verbose=False,
                                                        unconditional_guidance_scale=opt.scale,
                                                        unconditional_conditioning=uc,
                                                        eta=opt.ddim_eta,
                                                        x_T=start_code,
                                                        test_model_kwargs=test_model_kwargs)

                    x_samples_ddim = model.decode_first_stage(samples_ddim)
                    x_samples_ddim = torch.clamp((x_samples_ddim + 1.0) / 2.0, min=0.0, max=1.0)
                    x_samples_ddim = x_samples_ddim.cpu().permute(0, 2, 3, 1).numpy()

                    # x_checked_image, has_nsfw_concept = check_safety(x_samples_ddim)
                    x_checked_image=x_samples_ddim
                    x_checked_image_torch = torch.from_numpy(x_checked_image).permute(0, 3, 1, 2)

                    def un_norm(x):
                        return (x+1.0)/2.0
                    def un_norm_clip(x):
                        x[0,:,:] = x[0,:,:] * 0.26862954 + 0.48145466
                        x[1,:,:] = x[1,:,:] * 0.26130258 + 0.4578275
                        x[2,:,:] = x[2,:,:] * 0.27577711 + 0.40821073
                        return x

                    if not opt.skip_save:
                        for i,x_sample in enumerate(x_checked_image_torch):
                            

                            all_img=[]
                            all_img.append(un_norm(image_tensor[i]).cpu())
                            all_img.append(un_norm(inpaint_image[i]).cpu())
                            ref_img=ref_tensor
                            ref_img=Resize([opt.H, opt.W])(ref_img)
                            all_img.append(un_norm_clip(ref_img[i]).cpu())
                            all_img.append(x_sample)
                            grid = torch.stack(all_img, 0)
                            grid = make_grid(grid)
                            grid = 255. * rearrange(grid, 'c h w -> h w c').cpu().numpy()
                            img = Image.fromarray(grid.astype(np.uint8))
                            img = put_watermark(img, wm_encoder)
                            img.save(os.path.join(grid_path, 'grid-'+filename[:-4]+'_'+str(opt.seed)+f'_{i}.png'))
                            


                            x_sample = 255. * rearrange(x_sample.cpu().numpy(), 'c h w -> h w c')
                            img = Image.fromarray(x_sample.astype(np.uint8))
                            img = put_watermark(img, wm_encoder)
                            img.save(os.path.join(result_path, filename[:-4]+'_'+str(opt.seed)+f"_{i}.png"))
                            
                            mask_save=255.*rearrange(un_norm(inpaint_mask[i]).cpu(), 'c h w -> h w c').numpy()
                            mask_save= cv2.cvtColor(mask_save,cv2.COLOR_GRAY2RGB)
                            mask_save = Image.fromarray(mask_save.astype(np.uint8))
                            mask_save.save(os.path.join(sample_path, filename[:-4]+'_'+str(opt.seed)+f"_mask_{i}.png"))
                            GT_img=255.*rearrange(all_img[0], 'c h w -> h w c').numpy()
                            GT_img = Image.fromarray(GT_img.astype(np.uint8))
                            GT_img.save(os.path.join(sample_path, filename[:-4]+'_'+str(opt.seed)+f"_GT_{i}.png"))
                            inpaint_img=255.*rearrange(all_img[1], 'c h w -> h w c').numpy()
                            inpaint_img = Image.fromarray(inpaint_img.astype(np.uint8))
                            inpaint_img.save(os.path.join(sample_path, filename[:-4]+'_'+str(opt.seed)+f"_inpaint_{i}.png"))
                            ref_img=255.*rearrange(all_img[2], 'c h w -> h w c').numpy()
                            ref_img = Image.fromarray(ref_img.astype(np.uint8))
                            ref_img.save(os.path.join(sample_path, filename[:-4]+'_'+str(opt.seed)+f"_ref_{i}.png"))

    print(f"Your samples are ready and waiting for you here: \n{outpath} \n"
          f" \nEnjoy.")


if __name__ == "__main__":
    main()