txt2img.py

import argparse
import os
from itertools import islice

import cv2
import numpy as np
import torch
from einops import rearrange
from omegaconf import OmegaConf
from PIL import Image
from torchvision.utils import make_grid
from tqdm import tqdm, trange

try:
    from lightning.pytorch import seed_everything
except:
    from pytorch_lightning import seed_everything

from contextlib import nullcontext

from imwatermark import WatermarkEncoder
from ldm.models.diffusion.ddim import DDIMSampler
from ldm.models.diffusion.dpm_solver import DPMSolverSampler
from ldm.models.diffusion.plms import PLMSSampler
from ldm.util import instantiate_from_config
from torch import autocast
from utils import replace_module

torch.set_grad_enabled(False)


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


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)
    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.eval()
    return model


def parse_args():
    parser = argparse.ArgumentParser()
    parser.add_argument(
        "--prompt",
        type=str,
        nargs="?",
        default="a professional photograph of an astronaut riding a triceratops",
        help="the prompt to render",
    )
    parser.add_argument(
        "--outdir", type=str, nargs="?", help="dir to write results to", default="outputs/txt2img-samples"
    )
    parser.add_argument(
        "--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(
        "--dpm",
        action="store_true",
        help="use DPM (2) sampler",
    )
    parser.add_argument(
        "--fixed_code",
        action="store_true",
        help="if enabled, uses the same starting code across all 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=3,
        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(
        "--C",
        type=int,
        default=4,
        help="latent channels",
    )
    parser.add_argument(
        "--f",
        type=int,
        default=8,
        help="downsampling factor, most often 8 or 16",
    )
    parser.add_argument(
        "--n_samples",
        type=int,
        default=3,
        help="how many samples to produce for each given prompt. 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=9.0,
        help="unconditional guidance scale: eps = eps(x, empty) + scale * (eps(x, cond) - eps(x, empty))",
    )
    parser.add_argument(
        "--from-file",
        type=str,
        help="if specified, load prompts from this file, separated by newlines",
    )
    parser.add_argument(
        "--config",
        type=str,
        default="configs/stable-diffusion/v2-inference.yaml",
        help="path to config which constructs model",
    )
    parser.add_argument(
        "--ckpt",
        type=str,
        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(
        "--repeat",
        type=int,
        default=1,
        help="repeat each prompt in file this often",
    )
    parser.add_argument(
        "--use_int8",
        type=bool,
        default=False,
        help="use int8 for inference",
    )
    opt = parser.parse_args()
    return opt


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 main(opt):
    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)

    # quantize model
    if opt.use_int8:
        model = replace_module(model)
        # # to compute the model size
        # getModelSize(model)

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

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

    print("Creating invisible watermark encoder (see https://github.com/ShieldMnt/invisible-watermark)...")
    wm = "SDV2"
    wm_encoder = WatermarkEncoder()
    wm_encoder.set_watermark("bytes", wm.encode("utf-8"))

    batch_size = opt.n_samples
    n_rows = opt.n_rows if opt.n_rows > 0 else batch_size
    if not opt.from_file:
        prompt = opt.prompt
        assert prompt is not None
        data = [batch_size * [prompt]]

    else:
        print(f"reading prompts from {opt.from_file}")
        with open(opt.from_file, "r") as f:
            data = f.read().splitlines()
            data = [p for p in data for i in range(opt.repeat)]
            data = list(chunk(data, batch_size))

    sample_path = os.path.join(outpath, "samples")
    os.makedirs(sample_path, exist_ok=True)
    sample_count = 0
    base_count = len(os.listdir(sample_path))
    grid_count = len(os.listdir(outpath)) - 1

    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
    with torch.no_grad(), precision_scope("cuda"), model.ema_scope():
        all_samples = list()
        for n in trange(opt.n_iter, desc="Sampling"):
            for prompts in tqdm(data, desc="data"):
                uc = None
                if opt.scale != 1.0:
                    uc = model.get_learned_conditioning(batch_size * [""])
                if isinstance(prompts, tuple):
                    prompts = list(prompts)
                c = model.get_learned_conditioning(prompts)
                shape = [opt.C, opt.H // opt.f, opt.W // opt.f]
                samples, _ = sampler.sample(
                    S=opt.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,
                )

                x_samples = model.decode_first_stage(samples)
                x_samples = torch.clamp((x_samples + 1.0) / 2.0, min=0.0, max=1.0)

                for x_sample in x_samples:
                    x_sample = 255.0 * 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(sample_path, f"{base_count:05}.png"))
                    base_count += 1
                    sample_count += 1

                all_samples.append(x_samples)

        # additionally, save as grid
        grid = torch.stack(all_samples, 0)
        grid = rearrange(grid, "n b c h w -> (n b) c h w")
        grid = make_grid(grid, nrow=n_rows)

        # to image
        grid = 255.0 * rearrange(grid, "c h w -> h w c").cpu().numpy()
        grid = Image.fromarray(grid.astype(np.uint8))
        grid = put_watermark(grid, wm_encoder)
        grid.save(os.path.join(outpath, f"grid-{grid_count:04}.png"))
        grid_count += 1

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


if __name__ == "__main__":
    opt = parse_args()
    main(opt)
    # # to compute the mem allocated
    # print(torch.cuda.max_memory_allocated() / 1024 / 1024)