extras.py

# A script that enables merging SD 1.x based models to SD 2.x based models. Replace the same file of WebUI, which should be located at the modules folder.

from __future__ import annotations
import math
import os
import sys
import traceback
import shutil

import numpy as np
from PIL import Image

import torch
import tqdm

from typing import Callable, List, OrderedDict, Tuple
from functools import partial
from dataclasses import dataclass

from modules import processing, shared, images, devices, sd_models, sd_samplers, sd_vae
from modules.shared import opts
import modules.gfpgan_model
from modules.ui import plaintext_to_html
import modules.codeformer_model
import gradio as gr
import safetensors.torch

class LruCache(OrderedDict):
    @dataclass(frozen=True)
    class Key:
        image_hash: int
        info_hash: int
        args_hash: int

    @dataclass
    class Value:
        image: Image.Image
        info: str

    def __init__(self, max_size: int = 5, *args, **kwargs):
        super().__init__(*args, **kwargs)
        self._max_size = max_size

    def get(self, key: LruCache.Key) -> LruCache.Value:
        ret = super().get(key)
        if ret is not None:
            self.move_to_end(key)  # Move to end of eviction list
        return ret

    def put(self, key: LruCache.Key, value: LruCache.Value) -> None:
        self[key] = value
        while len(self) > self._max_size:
            self.popitem(last=False)


cached_images: LruCache = LruCache(max_size=5)


def run_extras(extras_mode, resize_mode, image, image_folder, input_dir, output_dir, show_extras_results, gfpgan_visibility, codeformer_visibility, codeformer_weight, upscaling_resize, upscaling_resize_w, upscaling_resize_h, upscaling_crop, extras_upscaler_1, extras_upscaler_2, extras_upscaler_2_visibility, upscale_first: bool, save_output: bool = True):
    devices.torch_gc()

    shared.state.begin()
    shared.state.job = 'extras'

    imageArr = []
    # Also keep track of original file names
    imageNameArr = []
    outputs = []

    if extras_mode == 1:
        #convert file to pillow image
        for img in image_folder:
            image = Image.open(img)
            imageArr.append(image)
            imageNameArr.append(os.path.splitext(img.orig_name)[0])
    elif extras_mode == 2:
        assert not shared.cmd_opts.hide_ui_dir_config, '--hide-ui-dir-config option must be disabled'

        if input_dir == '':
            return outputs, "Please select an input directory.", ''
        image_list = shared.listfiles(input_dir)
        for img in image_list:
            try:
                image = Image.open(img)
            except Exception:
                continue
            imageArr.append(image)
            imageNameArr.append(img)
    else:
        imageArr.append(image)
        imageNameArr.append(None)

    if extras_mode == 2 and output_dir != '':
        outpath = output_dir
    else:
        outpath = opts.outdir_samples or opts.outdir_extras_samples

    # Extra operation definitions

    def run_gfpgan(image: Image.Image, info: str) -> Tuple[Image.Image, str]:
        shared.state.job = 'extras-gfpgan'
        restored_img = modules.gfpgan_model.gfpgan_fix_faces(np.array(image, dtype=np.uint8))
        res = Image.fromarray(restored_img)

        if gfpgan_visibility < 1.0:
            res = Image.blend(image, res, gfpgan_visibility)

        info += f"GFPGAN visibility:{round(gfpgan_visibility, 2)}\n"
        return (res, info)

    def run_codeformer(image: Image.Image, info: str) -> Tuple[Image.Image, str]:
        shared.state.job = 'extras-codeformer'
        restored_img = modules.codeformer_model.codeformer.restore(np.array(image, dtype=np.uint8), w=codeformer_weight)
        res = Image.fromarray(restored_img)

        if codeformer_visibility < 1.0:
            res = Image.blend(image, res, codeformer_visibility)

        info += f"CodeFormer w: {round(codeformer_weight, 2)}, CodeFormer visibility:{round(codeformer_visibility, 2)}\n"
        return (res, info)

    def upscale(image, scaler_index, resize, mode, resize_w, resize_h, crop):
        shared.state.job = 'extras-upscale'
        upscaler = shared.sd_upscalers[scaler_index]
        res = upscaler.scaler.upscale(image, resize, upscaler.data_path)
        if mode == 1 and crop:
            cropped = Image.new("RGB", (resize_w, resize_h))
            cropped.paste(res, box=(resize_w // 2 - res.width // 2, resize_h // 2 - res.height // 2))
            res = cropped
        return res

    def run_prepare_crop(image: Image.Image, info: str) -> Tuple[Image.Image, str]:
        # Actual crop happens in run_upscalers_blend, this just sets upscaling_resize and adds info text
        nonlocal upscaling_resize
        if resize_mode == 1:
            upscaling_resize = max(upscaling_resize_w/image.width, upscaling_resize_h/image.height)
            crop_info = " (crop)" if upscaling_crop else ""
            info += f"Resize to: {upscaling_resize_w:g}x{upscaling_resize_h:g}{crop_info}\n"
        return (image, info)

    @dataclass
    class UpscaleParams:
        upscaler_idx: int
        blend_alpha: float

    def run_upscalers_blend(params: List[UpscaleParams], image: Image.Image, info: str) -> Tuple[Image.Image, str]:
        blended_result: Image.Image = None
        image_hash: str = hash(np.array(image.getdata()).tobytes())
        for upscaler in params:
            upscale_args = (upscaler.upscaler_idx, upscaling_resize, resize_mode,
                            upscaling_resize_w, upscaling_resize_h, upscaling_crop)
            cache_key = LruCache.Key(image_hash=image_hash,
                                     info_hash=hash(info),
                                     args_hash=hash(upscale_args))
            cached_entry = cached_images.get(cache_key)
            if cached_entry is None:
                res = upscale(image, *upscale_args)
                info += f"Upscale: {round(upscaling_resize, 3)}, visibility: {upscaler.blend_alpha}, model:{shared.sd_upscalers[upscaler.upscaler_idx].name}\n"
                cached_images.put(cache_key, LruCache.Value(image=res, info=info))
            else:
                res, info = cached_entry.image, cached_entry.info

            if blended_result is None:
                blended_result = res
            else:
                blended_result = Image.blend(blended_result, res, upscaler.blend_alpha)
        return (blended_result, info)

    # Build a list of operations to run
    facefix_ops: List[Callable] = []
    facefix_ops += [run_gfpgan] if gfpgan_visibility > 0 else []
    facefix_ops += [run_codeformer] if codeformer_visibility > 0 else []

    upscale_ops: List[Callable] = []
    upscale_ops += [run_prepare_crop] if resize_mode == 1 else []

    if upscaling_resize != 0:
        step_params: List[UpscaleParams] = []
        step_params.append(UpscaleParams(upscaler_idx=extras_upscaler_1, blend_alpha=1.0))
        if extras_upscaler_2 != 0 and extras_upscaler_2_visibility > 0:
            step_params.append(UpscaleParams(upscaler_idx=extras_upscaler_2, blend_alpha=extras_upscaler_2_visibility))

        upscale_ops.append(partial(run_upscalers_blend, step_params))

    extras_ops: List[Callable] = (upscale_ops + facefix_ops) if upscale_first else (facefix_ops + upscale_ops)

    for image, image_name in zip(imageArr, imageNameArr):
        if image is None:
            return outputs, "Please select an input image.", ''

        shared.state.textinfo = f'Processing image {image_name}'
        
        existing_pnginfo = image.info or {}

        image = image.convert("RGB")
        info = ""
        # Run each operation on each image
        for op in extras_ops:
            image, info = op(image, info)

        if opts.use_original_name_batch and image_name is not None:
            basename = os.path.splitext(os.path.basename(image_name))[0]
        else:
            basename = ''

        if opts.enable_pnginfo: # append info before save
            image.info = existing_pnginfo
            image.info["extras"] = info

        if save_output:
            # Add upscaler name as a suffix.
            suffix = f"-{shared.sd_upscalers[extras_upscaler_1].name}" if shared.opts.use_upscaler_name_as_suffix else ""
            # Add second upscaler if applicable.
            if suffix and extras_upscaler_2 and extras_upscaler_2_visibility:
                suffix += f"-{shared.sd_upscalers[extras_upscaler_2].name}"

            images.save_image(image, path=outpath, basename=basename, seed=None, prompt=None, extension=opts.samples_format, info=info, short_filename=True,
                            no_prompt=True, grid=False, pnginfo_section_name="extras", existing_info=existing_pnginfo, forced_filename=None, suffix=suffix)

        if extras_mode != 2 or show_extras_results :
            outputs.append(image)

    devices.torch_gc()

    return outputs, plaintext_to_html(info), ''

def clear_cache():
    cached_images.clear()


def run_pnginfo(image):
    if image is None:
        return '', '', ''

    geninfo, items = images.read_info_from_image(image)
    items = {**{'parameters': geninfo}, **items}

    info = ''
    for key, text in items.items():
        info += f"""
<div>
<p><b>{plaintext_to_html(str(key))}</b></p>
<p>{plaintext_to_html(str(text))}</p>
</div>
""".strip()+"\n"

    if len(info) == 0:
        message = "Nothing found in the image."
        info = f"<div><p>{message}<p></div>"

    return '', geninfo, info


def create_config(ckpt_result, config_source, a, b, c):
    def config(x):
        res = sd_models.find_checkpoint_config(x) if x else None
        return res if res != shared.sd_default_config else None

    if config_source == 0:
        cfg = config(a) or config(b) or config(c)
    elif config_source == 1:
        cfg = config(b)
    elif config_source == 2:
        cfg = config(c)
    else:
        cfg = None

    if cfg is None:
        return

    filename, _ = os.path.splitext(ckpt_result)
    checkpoint_filename = filename + ".yaml"

    print("Copying config:")
    print("   from:", cfg)
    print("     to:", checkpoint_filename)
    shutil.copyfile(cfg, checkpoint_filename)


checkpoint_dict_skip_on_merge = ["cond_stage_model.transformer.text_model.embeddings.position_ids"]


def to_half(tensor, enable):
    if enable and tensor.dtype == torch.float:
        return tensor.half()

    return tensor


def run_modelmerger(id_task, primary_model_name, secondary_model_name, tertiary_model_name, interp_method, multiplier, save_as_half, custom_name, checkpoint_format, config_source, bake_in_vae):
    shared.state.begin()
    shared.state.job = 'model-merge'

    def fail(message):
        shared.state.textinfo = message
        shared.state.end()
        return [*[gr.update() for _ in range(4)], message]

    def weighted_sum(theta0, theta1, alpha):
        return ((1 - alpha) * theta0) + (alpha * theta1)

    def get_difference(theta1, theta2):
        return theta1 - theta2

    def add_difference(theta0, theta1_2_diff, alpha):
        return theta0 + (alpha * theta1_2_diff)

    def filename_weighted_sum():
        a = primary_model_info.model_name
        b = secondary_model_info.model_name
        Ma = round(1 - multiplier, 2)
        Mb = round(multiplier, 2)

        return f"{Ma}({a}) + {Mb}({b})"

    def filename_add_difference():
        a = primary_model_info.model_name
        b = secondary_model_info.model_name
        c = tertiary_model_info.model_name
        M = round(multiplier, 2)

        return f"{a} + {M}({b} - {c})"

    def filename_nothing():
        return primary_model_info.model_name

    theta_funcs = {
        "Weighted sum": (filename_weighted_sum, None, weighted_sum),
        "Add difference": (filename_add_difference, get_difference, add_difference),
        "No interpolation": (filename_nothing, None, None),
    }
    filename_generator, theta_func1, theta_func2 = theta_funcs[interp_method]
    shared.state.job_count = (1 if theta_func1 else 0) + (1 if theta_func2 else 0)

    if not primary_model_name:
        return fail("Failed: Merging requires a primary model.")

    primary_model_info = sd_models.checkpoints_list[primary_model_name]

    if theta_func2 and not secondary_model_name:
        return fail("Failed: Merging requires a secondary model.")

    secondary_model_info = sd_models.checkpoints_list[secondary_model_name] if theta_func2 else None

    if theta_func1 and not tertiary_model_name:
        return fail(f"Failed: Interpolation method ({interp_method}) requires a tertiary model.")

    tertiary_model_info = sd_models.checkpoints_list[tertiary_model_name] if theta_func1 else None

    result_is_inpainting_model = False

    if theta_func2:
        shared.state.textinfo = f"Loading B"
        print(f"Loading {secondary_model_info.filename}...")
        theta_1 = sd_models.read_state_dict(secondary_model_info.filename, map_location='cpu')
    else:
        theta_1 = None

    if theta_func1:
        shared.state.textinfo = f"Loading C"
        print(f"Loading {tertiary_model_info.filename}...")
        theta_2 = sd_models.read_state_dict(tertiary_model_info.filename, map_location='cpu')

        shared.state.textinfo = 'Merging B and C'
        shared.state.sampling_steps = len(theta_1.keys())
        for key in tqdm.tqdm(theta_1.keys()):
            if key in checkpoint_dict_skip_on_merge:
                continue

            if 'model' in key:
                if key in theta_2:
                    t2 = theta_2.get(key, torch.zeros_like(theta_1[key]))
                    theta_1[key] = theta_func1(theta_1[key], t2)
                else:
                    theta_1[key] = torch.zeros_like(theta_1[key])

            shared.state.sampling_step += 1
        del theta_2

        shared.state.nextjob()

    shared.state.textinfo = f"Loading {primary_model_info.filename}..."
    print(f"Loading {primary_model_info.filename}...")
    theta_0 = sd_models.read_state_dict(primary_model_info.filename, map_location='cpu')

    print("Merging...")
    shared.state.textinfo = 'Merging A and B'
    shared.state.sampling_steps = len(theta_0.keys())
    for key in tqdm.tqdm(theta_0.keys()):
        if theta_1 and 'model' in key and key in theta_1:

            if key in checkpoint_dict_skip_on_merge:
                continue

            a = theta_0[key]
            b = theta_1[key]

            # this enables merging an inpainting model (A) with another one (B);
            # where normal model would have 4 channels, for latenst space, inpainting model would
            # have another 4 channels for unmasked picture's latent space, plus one channel for mask, for a total of 9
            # Note: I will keep the original code as a comment so that you can turn it back on whenever necessary.
            if a.shape != b.shape: #and a.shape[0:1] + a.shape[2:] == b.shape[0:1] + b.shape[2:]:
              theta_0[key] = a # just let it go as it is on size mismatch
                #if a.shape[1] == 4 and b.shape[1] == 9:
                    #raise RuntimeError("When merging inpainting model with a normal one, A must be the inpainting model.")

                #assert a.shape[1] == 9 and b.shape[1] == 4, f"Bad dimensions for merged layer {key}: A={a.shape}, B={b.shape}"

                #theta_0[key][:, 0:4, :, :] = theta_func2(a[:, 0:4, :, :], b, multiplier)
                #result_is_inpainting_model = True
            else:
              theta_0[key] = theta_func2(a, b, multiplier)

            theta_0[key] = to_half(theta_0[key], save_as_half)

        shared.state.sampling_step += 1

    del theta_1

    bake_in_vae_filename = sd_vae.vae_dict.get(bake_in_vae, None)
    if bake_in_vae_filename is not None:
        print(f"Baking in VAE from {bake_in_vae_filename}")
        shared.state.textinfo = 'Baking in VAE'
        vae_dict = sd_vae.load_vae_dict(bake_in_vae_filename, map_location='cpu')

        for key in vae_dict.keys():
            theta_0_key = 'first_stage_model.' + key
            if theta_0_key in theta_0:
                theta_0[theta_0_key] = to_half(vae_dict[key], save_as_half)

        del vae_dict

    if save_as_half and not theta_func2:
        for key in theta_0.keys():
            theta_0[key] = to_half(theta_0[key], save_as_half)

    ckpt_dir = shared.cmd_opts.ckpt_dir or sd_models.model_path

    filename = filename_generator() if custom_name == '' else custom_name
    filename += ".inpainting" if result_is_inpainting_model else ""
    filename += "." + checkpoint_format

    output_modelname = os.path.join(ckpt_dir, filename)

    shared.state.nextjob()
    shared.state.textinfo = "Saving"
    print(f"Saving to {output_modelname}...")

    _, extension = os.path.splitext(output_modelname)
    if extension.lower() == ".safetensors":
        safetensors.torch.save_file(theta_0, output_modelname, metadata={"format": "pt"})
    else:
        torch.save(theta_0, output_modelname)

    sd_models.list_models()

    create_config(output_modelname, config_source, primary_model_info, secondary_model_info, tertiary_model_info)

    print(f"Checkpoint saved to {output_modelname}.")
    shared.state.textinfo = "Checkpoint saved"
    shared.state.end()

    return [*[gr.Dropdown.update(choices=sd_models.checkpoint_tiles()) for _ in range(4)], "Checkpoint saved to " + output_modelname]