init

.gitignore

@@ -0,0 +1,169 @@
+hf_token.txt
+hf_download/
+results/
+*.csv
+*.onnx
+
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+#   For a library or package, you might want to ignore these files since the code is
+#   intended to run in multiple environments; otherwise, check them in:
+# .python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# poetry
+#   Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#   https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
+#poetry.lock
+
+# pdm
+#   Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
+#pdm.lock
+#   pdm stores project-wide configurations in .pdm.toml, but it is recommended to not include it
+#   in version control.
+#   https://pdm.fming.dev/latest/usage/project/#working-with-version-control
+.pdm.toml
+.pdm-python
+.pdm-build/
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+venv*/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# PyCharm
+#  JetBrains specific template is maintained in a separate JetBrains.gitignore that can
+#  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
+#  and can be added to the global gitignore or merged into this file.  For a more nuclear
+#  option (not recommended) you can uncomment the following to ignore the entire idea folder.
+.idea/

LICENSE

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

NOTICE

@@ -0,0 +1,15 @@
+NOTICE
+
+This repository contains files that have been copied or modified from other sources:
+
+1. diffusers_helper/*.py # expect diffusers_helper/attention.py
+   Source: https://github.com/lllyasviel/Paints-UNDO/tree/main/diffusers_helper
+   License: Apache-2.0
+   Modifications: No
+
+2. app.py
+   Source: https://github.com/lllyasviel/Paints-UNDO/tree/main/gradio_app.py
+   License: Apache-2.0
+   Modifications: Modified UI for sketch gen only. Remove vid gen utils. Add clip concat utils.
+
+Please see the individual files for their specific licensing terms.

app.py

@@ -0,0 +1,392 @@
+'''
+Modified from https://github.com/lllyasviel/Paints-UNDO/blob/main/gradio_app.py
+'''
+import functools
+
+import spaces
+import gradio as gr
+import numpy as np
+import cv2
+import torch
+
+from PIL import Image
+from diffusers import AutoencoderKL, UNet2DConditionModel
+from diffusers.models.attention_processor import AttnProcessor2_0
+from transformers import CLIPTextModel, CLIPTokenizer
+from imgutils.metrics import lpips_difference
+from imgutils.tagging import get_wd14_tags
+
+from diffusers_helper.code_cond import unet_add_coded_conds
+from diffusers_helper.cat_cond import unet_add_concat_conds
+from diffusers_helper.k_diffusion import KDiffusionSampler
+from diffusers_helper.attention import AttnProcessor2_0_xformers, XFORMERS_AVAIL
+
+from lineart_models import MangaLineExtraction, LineartAnimeDetector, LineartDetector
+
+
+def resize_and_center_crop(
+    image, target_width, target_height=None, interpolation=cv2.INTER_AREA
+):
+    original_height, original_width = image.shape[:2]
+    if target_height is None:
+        aspect_ratio = original_width / original_height
+        target_pixel_count = target_width * target_width
+        target_height = (target_pixel_count / aspect_ratio) ** 0.5
+        target_width = target_height * aspect_ratio
+    target_height = int(target_height)
+    target_width = int(target_width)
+    print(
+        f"original_height={original_height}, "
+        f"original_width={original_width}, "
+        f"target_height={target_height}, "
+        f"target_width={target_width}"
+    )
+    k = max(target_height / original_height, target_width / original_width)
+    new_width = int(round(original_width * k))
+    new_height = int(round(original_height * k))
+    resized_image = cv2.resize(
+        image, (new_width, new_height), interpolation=interpolation
+    )
+    x_start = (new_width - target_width) // 2
+    y_start = (new_height - target_height) // 2
+    cropped_image = resized_image[
+        y_start : y_start + target_height, x_start : x_start + target_width
+    ]
+    return cropped_image
+
+
+class ModifiedUNet(UNet2DConditionModel):
+    @classmethod
+    def from_config(cls, *args, **kwargs):
+        m = super().from_config(*args, **kwargs)
+        unet_add_concat_conds(unet=m, new_channels=4)
+        unet_add_coded_conds(unet=m, added_number_count=1)
+        return m
+
+
+DEVICE = "cuda"
+torch._dynamo.config.cache_size_limit = 256
+
+
+lineart_models = []
+
+lineart_model = MangaLineExtraction("cuda", "./hf_download")
+lineart_model.load_model()
+lineart_model.model.to(device=DEVICE).eval()
+lineart_models.append(lineart_model)
+
+lineart_model = LineartAnimeDetector()
+lineart_model.model.to(device=DEVICE).eval()
+lineart_models.append(lineart_model)
+
+lineart_model = LineartDetector()
+lineart_model.model.to(device=DEVICE).eval()
+lineart_models.append(lineart_model)
+
+
+model_name = "lllyasviel/paints_undo_single_frame"
+tokenizer: CLIPTokenizer = CLIPTokenizer.from_pretrained(
+    model_name, subfolder="tokenizer"
+)
+text_encoder: CLIPTextModel = (
+    CLIPTextModel.from_pretrained(
+        model_name,
+        subfolder="text_encoder",
+    )
+    .to(dtype=torch.float16, device=DEVICE)
+    .eval()
+)
+vae: AutoencoderKL = (
+    AutoencoderKL.from_pretrained(
+        model_name,
+        subfolder="vae",
+    )
+    .to(dtype=torch.bfloat16, device=DEVICE)
+    .eval()
+)
+unet: ModifiedUNet = (
+    ModifiedUNet.from_pretrained(
+        model_name,
+        subfolder="unet",
+    )
+    .to(dtype=torch.float16, device=DEVICE)
+    .eval()
+)
+
+if XFORMERS_AVAIL:
+    unet.set_attn_processor(AttnProcessor2_0_xformers())
+    vae.set_attn_processor(AttnProcessor2_0_xformers())
+else:
+    unet.set_attn_processor(AttnProcessor2_0())
+    vae.set_attn_processor(AttnProcessor2_0())
+
+# text_encoder = torch.compile(text_encoder, backend="eager", dynamic=True)
+# vae = torch.compile(vae, backend="eager", dynamic=True)
+# unet = torch.compile(unet, mode="reduce-overhead", dynamic=True)
+# for model in lineart_models:
+#     model.model = torch.compile(model.model, backend="eager", dynamic=True)
+k_sampler = KDiffusionSampler(
+    unet=unet,
+    timesteps=1000,
+    linear_start=0.00085,
+    linear_end=0.020,
+    linear=True,
+)
+
+
+@spaces.GPU
+@torch.inference_mode()
+def encode_cropped_prompt_77tokens(txt: str):
+    cond_ids = tokenizer(
+        txt,
+        padding="max_length",
+        max_length=tokenizer.model_max_length,
+        truncation=True,
+        return_tensors="pt",
+    ).input_ids.to(device=text_encoder.device)
+    text_cond = text_encoder(cond_ids, attention_mask=None).last_hidden_state
+    return text_cond
+
+
+@spaces.GPU
+@torch.inference_mode()
+def encode_cropped_prompt(txt: str, max_length=225):
+    cond_ids = tokenizer(
+        txt,
+        padding="max_length",
+        max_length=max_length + 2,
+        truncation=True,
+        return_tensors="pt",
+    ).input_ids.to(device=text_encoder.device)
+    if max_length + 2 > tokenizer.model_max_length:
+        input_ids = cond_ids.squeeze(0)
+        id_list = list(range(1, max_length + 2 - tokenizer.model_max_length + 2, tokenizer.model_max_length - 2))
+        text_cond_list = []
+        for i in id_list:
+            ids_chunk = (
+                input_ids[0].unsqueeze(0),
+                input_ids[i : i + tokenizer.model_max_length - 2],
+                input_ids[-1].unsqueeze(0),
+            )
+            if torch.all(ids_chunk[1] == tokenizer.pad_token_id):
+                break
+            text_cond = text_encoder(torch.concat(ids_chunk).unsqueeze(0)).last_hidden_state
+            if text_cond_list == []:
+                text_cond_list.append(text_cond[:, :1])
+            text_cond_list.append(text_cond[:, 1:tokenizer.model_max_length - 1])
+        text_cond_list.append(text_cond[:, -1:])
+        text_cond = torch.concat(text_cond_list, dim=1)
+    else:
+        text_cond = text_encoder(
+            cond_ids, attention_mask=None
+        ).last_hidden_state
+    return text_cond.flatten(0, 1).unsqueeze(0)
+
+
+@spaces.GPU
+@torch.inference_mode()
+def pytorch2numpy(imgs):
+    results = []
+    for x in imgs:
+        y = x.movedim(0, -1)
+        y = y * 127.5 + 127.5
+        y = y.detach().float().cpu().numpy().clip(0, 255).astype(np.uint8)
+        results.append(y)
+    return results
+
+
+@spaces.GPU
+@torch.inference_mode()
+def numpy2pytorch(imgs):
+    h = torch.from_numpy(np.stack(imgs, axis=0)).float() / 127.5 - 1.0
+    h = h.movedim(-1, 1)
+    return h
+
+
+@spaces.GPU
+@torch.inference_mode()
+def interrogator_process(x):
+    img = Image.fromarray(x)
+    rating, features, chars = get_wd14_tags(img, general_threshold=0.25, no_underline=True)
+    result = ""
+    for char in chars:
+        result += char
+        result += ", "
+    for feature in features:
+        result += feature
+        result += ", "
+    result += max(rating, key=rating.get)
+    return result
+
+
+@spaces.GPU
+@torch.inference_mode()
+def process(
+    input_fg,
+    prompt,
+    input_undo_steps,
+    image_width,
+    seed,
+    steps,
+    n_prompt,
+    cfg,
+    num_sets,
+    progress=gr.Progress(),
+):
+    lineart_fg = input_fg
+    linearts = []
+    for model in lineart_models:
+        linearts.append(model(lineart_fg))
+    fg = resize_and_center_crop(input_fg, image_width)
+    for i, lineart in enumerate(linearts):
+        lineart = resize_and_center_crop(lineart, fg.shape[1], fg.shape[0])
+        linearts[i] = lineart
+
+    concat_conds = numpy2pytorch([fg]).to(device=vae.device, dtype=vae.dtype)
+    concat_conds = (
+        vae.encode(concat_conds).latent_dist.mode() * vae.config.scaling_factor
+    )
+
+    conds = encode_cropped_prompt(prompt)
+    unconds = encode_cropped_prompt_77tokens(n_prompt)
+    print(conds.shape, unconds.shape)
+    torch.cuda.empty_cache()
+
+    fs = torch.tensor(input_undo_steps).to(device=unet.device, dtype=torch.long)
+    initial_latents = torch.zeros_like(concat_conds)
+    concat_conds = concat_conds.to(device=unet.device, dtype=unet.dtype)
+    latents = []
+    rng = torch.Generator(device=DEVICE).manual_seed(int(seed))
+    latents = (
+        k_sampler(
+            initial_latent=initial_latents,
+            strength=1.0,
+            num_inference_steps=steps,
+            guidance_scale=cfg,
+            batch_size=len(input_undo_steps) * num_sets,
+            generator=rng,
+            prompt_embeds=conds,
+            negative_prompt_embeds=unconds,
+            cross_attention_kwargs={
+                "concat_conds": concat_conds,
+                "coded_conds": fs,
+            },
+            same_noise_in_batch=False,
+            progress_tqdm=functools.partial(
+                progress.tqdm, desc="Generating Key Frames"
+            ),
+        ).to(vae.dtype)
+        / vae.config.scaling_factor
+    )
+    torch.cuda.empty_cache()
+
+    pixels = torch.concat(
+        [vae.decode(latent.unsqueeze(0)).sample for latent in latents]
+    )
+    pixels = pytorch2numpy(pixels)
+    pixels_with_lpips = []
+    lineart_pils = [Image.fromarray(lineart) for lineart in linearts]
+    for pixel in pixels:
+        pixel_pil = Image.fromarray(pixel)
+        pixels_with_lpips.append(
+            (
+                sum(
+                    [
+                        lpips_difference(lineart_pil, pixel_pil)
+                        for lineart_pil in lineart_pils
+                    ]
+                ),
+                pixel,
+            )
+        )
+    pixels = np.stack(
+        [i[1] for i in sorted(pixels_with_lpips, key=lambda x: x[0])], axis=0
+    )
+    torch.cuda.empty_cache()
+
+    return pixels, np.stack(linearts)
+
+
+block = gr.Blocks().queue()
+with block:
+    gr.Markdown("# Sketch/Lineart extractor")
+
+    with gr.Row():
+        with gr.Column():
+            input_fg = gr.Image(
+                sources=["upload"], type="numpy", label="Image", height=384
+            )
+            with gr.Row():
+                with gr.Column(scale=2, variant="compact"):
+                    prompt = gr.Textbox(label="Output Prompt", interactive=True)
+                with gr.Column(scale=1, variant="compact", min_width=160):
+                    n_prompt = gr.Textbox(
+                        label="Negative Prompt",
+                        value="lowres, worst quality, bad anatomy, bad hands, text, extra digit, fewer digits, cropped, low quality, jpeg artifacts, signature, watermark, username",
+                    )
+            with gr.Row():
+                input_undo_steps = gr.Dropdown(
+                    label="Operation Steps",
+                    value=[850, 875, 900, 925, 950, 975],
+                    choices=list(range(0, 1000, 25)),
+                    multiselect=True,
+                )
+                num_sets = gr.Slider(
+                    label="Num Sets", minimum=1, maximum=10, value=4, step=1
+                )
+            with gr.Row():
+                seed = gr.Slider(
+                    label="Seed", minimum=0, maximum=50000, step=1, value=37462
+                )
+                image_width = gr.Slider(
+                    label="Target size", minimum=512, maximum=1024, value=768, step=32
+                )
+                steps = gr.Slider(
+                    label="Steps", minimum=1, maximum=32, value=16, step=1
+                )
+                cfg = gr.Slider(
+                    label="CFG Scale", minimum=1.0, maximum=16, value=5, step=0.05
+                )
+            key_gen_button = gr.Button(value="Generate Sketch", interactive=False)
+
+        with gr.Column():
+            gr.Markdown("#### Sketch Outputs")
+            result_gallery = gr.Gallery(
+                height=384, object_fit="contain", label="Sketch Outputs", columns=4
+            )
+            gr.Markdown("#### Line Art Outputs")
+            lineart_result = gr.Gallery(
+                height=384,
+                object_fit="contain",
+                label="LineArt outputs",
+            )
+
+    input_fg.change(
+        lambda x: [
+            interrogator_process(x) if x is not None else "",
+            gr.update(interactive=True),
+        ],
+        inputs=[input_fg],
+        outputs=[prompt, key_gen_button],
+    )
+
+    key_gen_button.click(
+        fn=process,
+        inputs=[
+            input_fg,
+            prompt,
+            input_undo_steps,
+            image_width,
+            seed,
+            steps,
+            n_prompt,
+            cfg,
+            num_sets,
+        ],
+        outputs=[result_gallery, lineart_result],
+    ).then(
+        lambda: gr.update(interactive=True),
+        outputs=[key_gen_button],
+    )
+
+block.queue().launch()

diffusers_helper/attention.py

@@ -0,0 +1,86 @@
+from typing import Optional
+
+import torch
+
+try:
+    from xformers.ops import memory_efficient_attention
+    XFORMERS_AVAIL = True
+except ImportError:
+    XFORMERS_AVAIL = False
+
+
+class AttnProcessor2_0_xformers:
+    def __call__(
+        self,
+        attn,
+        hidden_states: torch.Tensor,
+        encoder_hidden_states: Optional[torch.Tensor] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        temb: Optional[torch.Tensor] = None,
+        *args,
+        **kwargs,
+    ) -> torch.Tensor:
+        residual = hidden_states
+        if attn.spatial_norm is not None:
+            hidden_states = attn.spatial_norm(hidden_states, temb)
+
+        input_ndim = hidden_states.ndim
+
+        if input_ndim == 4:
+            batch_size, channel, height, width = hidden_states.shape
+            hidden_states = hidden_states.view(batch_size, channel, height * width).transpose(1, 2)
+
+        batch_size, sequence_length, _ = (
+            hidden_states.shape if encoder_hidden_states is None else encoder_hidden_states.shape
+        )
+
+        if attention_mask is not None:
+            attention_mask = attn.prepare_attention_mask(attention_mask, sequence_length, batch_size)
+            # scaled_dot_product_attention expects attention_mask shape to be
+            # (batch, heads, source_length, target_length)
+            attention_mask = attention_mask.view(batch_size, attn.heads, -1, attention_mask.shape[-1])
+
+        if attn.group_norm is not None:
+            hidden_states = attn.group_norm(hidden_states.transpose(1, 2)).transpose(1, 2)
+
+        query = attn.to_q(hidden_states)
+
+        if encoder_hidden_states is None:
+            encoder_hidden_states = hidden_states
+        elif attn.norm_cross:
+            encoder_hidden_states = attn.norm_encoder_hidden_states(encoder_hidden_states)
+
+        key = attn.to_k(encoder_hidden_states)
+        value = attn.to_v(encoder_hidden_states)
+
+        inner_dim = key.shape[-1]
+        head_dim = inner_dim // attn.heads
+
+        query = query.view(batch_size, -1, attn.heads, head_dim)
+
+        key = key.view(batch_size, -1, attn.heads, head_dim)
+        value = value.view(batch_size, -1, attn.heads, head_dim)
+
+        # the output of sdp = (batch, num_heads, seq_len, head_dim)
+        # TODO: add support for attn.scale when we move to Torch 2.1
+        hidden_states = memory_efficient_attention(
+            query, key, value, attention_mask, p=0.0
+        )
+
+        hidden_states = hidden_states.reshape(batch_size, -1, attn.heads * head_dim)
+        hidden_states = hidden_states.to(query.dtype)
+
+        # linear proj
+        hidden_states = attn.to_out[0](hidden_states)
+        # dropout
+        hidden_states = attn.to_out[1](hidden_states)
+
+        if input_ndim == 4:
+            hidden_states = hidden_states.transpose(-1, -2).reshape(batch_size, channel, height, width)
+
+        if attn.residual_connection:
+            hidden_states = hidden_states + residual
+
+        hidden_states = hidden_states / attn.rescale_output_factor
+
+        return hidden_states

diffusers_helper/cat_cond.py

@@ -0,0 +1,24 @@
+import torch
+
+
+def unet_add_concat_conds(unet, new_channels=4):
+    with torch.no_grad():
+        new_conv_in = torch.nn.Conv2d(4 + new_channels, unet.conv_in.out_channels, unet.conv_in.kernel_size, unet.conv_in.stride, unet.conv_in.padding)
+        new_conv_in.weight.zero_()
+        new_conv_in.weight[:, :4, :, :].copy_(unet.conv_in.weight)
+        new_conv_in.bias = unet.conv_in.bias
+        unet.conv_in = new_conv_in
+
+    unet_original_forward = unet.forward
+
+    def hooked_unet_forward(sample, timestep, encoder_hidden_states, **kwargs):
+        cross_attention_kwargs = {k: v for k, v in kwargs['cross_attention_kwargs'].items()}
+        c_concat = cross_attention_kwargs.pop('concat_conds')
+        kwargs['cross_attention_kwargs'] = cross_attention_kwargs
+
+        c_concat = torch.cat([c_concat] * (sample.shape[0] // c_concat.shape[0]), dim=0).to(sample)
+        new_sample = torch.cat([sample, c_concat], dim=1)
+        return unet_original_forward(new_sample, timestep, encoder_hidden_states, **kwargs)
+
+    unet.forward = hooked_unet_forward
+    return

diffusers_helper/code_cond.py

@@ -0,0 +1,34 @@
+import torch
+
+from diffusers.models.embeddings import TimestepEmbedding, Timesteps
+
+
+def unet_add_coded_conds(unet, added_number_count=1):
+    unet.add_time_proj = Timesteps(256, True, 0)
+    unet.add_embedding = TimestepEmbedding(256 * added_number_count, 1280)
+
+    def get_aug_embed(emb, encoder_hidden_states, added_cond_kwargs):
+        coded_conds = added_cond_kwargs.get("coded_conds")
+        batch_size = coded_conds.shape[0]
+        time_embeds = unet.add_time_proj(coded_conds.flatten())
+        time_embeds = time_embeds.reshape((batch_size, -1))
+        time_embeds = time_embeds.to(emb)
+        aug_emb = unet.add_embedding(time_embeds)
+        return aug_emb
+
+    unet.get_aug_embed = get_aug_embed
+
+    unet_original_forward = unet.forward
+
+    def hooked_unet_forward(sample, timestep, encoder_hidden_states, **kwargs):
+        cross_attention_kwargs = {k: v for k, v in kwargs['cross_attention_kwargs'].items()}
+        coded_conds = cross_attention_kwargs.pop('coded_conds')
+        kwargs['cross_attention_kwargs'] = cross_attention_kwargs
+
+        coded_conds = torch.cat([coded_conds] * (sample.shape[0] // coded_conds.shape[0]), dim=0).to(sample.device)
+        kwargs['added_cond_kwargs'] = dict(coded_conds=coded_conds)
+        return unet_original_forward(sample, timestep, encoder_hidden_states, **kwargs)
+
+    unet.forward = hooked_unet_forward
+
+    return

diffusers_helper/k_diffusion.py

@@ -0,0 +1,145 @@
+import torch
+import numpy as np
+
+from tqdm import tqdm
+
+
+@torch.no_grad()
+def sample_dpmpp_2m(model, x, sigmas, extra_args=None, callback=None, progress_tqdm=None):
+    """DPM-Solver++(2M)."""
+    extra_args = {} if extra_args is None else extra_args
+    s_in = x.new_ones([x.shape[0]])
+    sigma_fn = lambda t: t.neg().exp()
+    t_fn = lambda sigma: sigma.log().neg()
+    old_denoised = None
+
+    bar = tqdm if progress_tqdm is None else progress_tqdm
+
+    for i in bar(range(len(sigmas) - 1)):
+        denoised = model(x, sigmas[i] * s_in, **extra_args)
+        if callback is not None:
+            callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigmas[i], 'denoised': denoised})
+        t, t_next = t_fn(sigmas[i]), t_fn(sigmas[i + 1])
+        h = t_next - t
+        if old_denoised is None or sigmas[i + 1] == 0:
+            x = (sigma_fn(t_next) / sigma_fn(t)) * x - (-h).expm1() * denoised
+        else:
+            h_last = t - t_fn(sigmas[i - 1])
+            r = h_last / h
+            denoised_d = (1 + 1 / (2 * r)) * denoised - (1 / (2 * r)) * old_denoised
+            x = (sigma_fn(t_next) / sigma_fn(t)) * x - (-h).expm1() * denoised_d
+        old_denoised = denoised
+    return x
+
+
+class KModel:
+    def __init__(self, unet, timesteps=1000, linear_start=0.00085, linear_end=0.012, linear=False):
+        if linear:
+            betas = torch.linspace(linear_start, linear_end, timesteps, dtype=torch.float64)
+        else:
+            betas = torch.linspace(linear_start ** 0.5, linear_end ** 0.5, timesteps, dtype=torch.float64) ** 2
+
+        alphas = 1. - betas
+        alphas_cumprod = torch.tensor(np.cumprod(alphas, axis=0), dtype=torch.float32)
+
+        self.sigmas = ((1 - alphas_cumprod) / alphas_cumprod) ** 0.5
+        self.log_sigmas = self.sigmas.log()
+        self.sigma_data = 1.0
+        self.unet = unet
+        return
+
+    @property
+    def sigma_min(self):
+        return self.sigmas[0]
+
+    @property
+    def sigma_max(self):
+        return self.sigmas[-1]
+
+    def timestep(self, sigma):
+        log_sigma = sigma.log()
+        dists = log_sigma.to(self.log_sigmas.device) - self.log_sigmas[:, None]
+        return dists.abs().argmin(dim=0).view(sigma.shape).to(sigma.device)
+
+    def get_sigmas_karras(self, n, rho=7.):
+        ramp = torch.linspace(0, 1, n)
+        min_inv_rho = self.sigma_min ** (1 / rho)
+        max_inv_rho = self.sigma_max ** (1 / rho)
+        sigmas = (max_inv_rho + ramp * (min_inv_rho - max_inv_rho)) ** rho
+        return torch.cat([sigmas, sigmas.new_zeros([1])])
+
+    def __call__(self, x, sigma, **extra_args):
+        x_ddim_space = x / (sigma[:, None, None, None] ** 2 + self.sigma_data ** 2) ** 0.5
+        x_ddim_space = x_ddim_space.to(dtype=self.unet.dtype)
+        t = self.timestep(sigma)
+        cfg_scale = extra_args['cfg_scale']
+        eps_positive = self.unet(x_ddim_space, t, return_dict=False, **extra_args['positive'])[0]
+        eps_negative = self.unet(x_ddim_space, t, return_dict=False, **extra_args['negative'])[0]
+        noise_pred = eps_negative + cfg_scale * (eps_positive - eps_negative)
+        return x - noise_pred * sigma[:, None, None, None]
+
+
+class KDiffusionSampler:
+    def __init__(self, unet, **kwargs):
+        self.unet = unet
+        self.k_model = KModel(unet=unet, **kwargs)
+
+    @torch.inference_mode()
+    def __call__(
+            self,
+            initial_latent = None,
+            strength = 1.0,
+            num_inference_steps = 25,
+            guidance_scale = 5.0,
+            batch_size = 1,
+            generator = None,
+            prompt_embeds = None,
+            negative_prompt_embeds = None,
+            cross_attention_kwargs = None,
+            same_noise_in_batch = False,
+            progress_tqdm = None,
+    ):
+
+        device = self.unet.device
+
+        # Sigmas
+
+        sigmas = self.k_model.get_sigmas_karras(int(num_inference_steps/strength))
+        sigmas = sigmas[-(num_inference_steps + 1):].to(device)
+
+        # Initial latents
+
+        if same_noise_in_batch:
+            noise = torch.randn(initial_latent.shape, generator=generator, device=device, dtype=self.unet.dtype).repeat(batch_size, 1, 1, 1)
+            initial_latent = initial_latent.repeat(batch_size, 1, 1, 1).to(device=device, dtype=self.unet.dtype)
+        else:
+            initial_latent = initial_latent.repeat(batch_size, 1, 1, 1).to(device=device, dtype=self.unet.dtype)
+            noise = torch.randn(initial_latent.shape, generator=generator, device=device, dtype=self.unet.dtype)
+
+        latents = initial_latent + noise * sigmas[0].to(initial_latent)
+
+        # Batch
+
+        latents = latents.to(device)
+        prompt_embeds = prompt_embeds.repeat(batch_size, 1, 1).to(device)
+        negative_prompt_embeds = negative_prompt_embeds.repeat(batch_size, 1, 1).to(device)
+
+        # Feeds
+
+        sampler_kwargs = dict(
+            cfg_scale=guidance_scale,
+            positive=dict(
+                encoder_hidden_states=prompt_embeds,
+                cross_attention_kwargs=cross_attention_kwargs
+            ),
+            negative=dict(
+                encoder_hidden_states=negative_prompt_embeds,
+                cross_attention_kwargs=cross_attention_kwargs,
+            )
+        )
+
+        # Sample
+
+        results = sample_dpmpp_2m(self.k_model, latents, sigmas, extra_args=sampler_kwargs, progress_tqdm=progress_tqdm)
+
+        return results

diffusers_helper/utils.py

@@ -0,0 +1,136 @@
+import os
+import json
+import random
+import glob
+import torch
+import einops
+import torchvision
+
+import safetensors.torch as sf
+
+
+def write_to_json(data, file_path):
+    temp_file_path = file_path + ".tmp"
+    with open(temp_file_path, 'wt', encoding='utf-8') as temp_file:
+        json.dump(data, temp_file, indent=4)
+    os.replace(temp_file_path, file_path)
+    return
+
+
+def read_from_json(file_path):
+    with open(file_path, 'rt', encoding='utf-8') as file:
+        data = json.load(file)
+    return data
+
+
+def get_active_parameters(m):
+    return {k:v for k, v in m.named_parameters() if v.requires_grad}
+
+
+def cast_training_params(m, dtype=torch.float32):
+    for param in m.parameters():
+        if param.requires_grad:
+            param.data = param.to(dtype)
+    return
+
+
+def set_attr_recursive(obj, attr, value):
+    attrs = attr.split(".")
+    for name in attrs[:-1]:
+        obj = getattr(obj, name)
+    setattr(obj, attrs[-1], value)
+    return
+
+
+@torch.no_grad()
+def batch_mixture(a, b, probability_a=0.5, mask_a=None):
+    assert a.shape == b.shape, "Tensors must have the same shape"
+    batch_size = a.size(0)
+
+    if mask_a is None:
+        mask_a = torch.rand(batch_size) < probability_a
+
+    mask_a = mask_a.to(a.device)
+    mask_a = mask_a.reshape((batch_size,) + (1,) * (a.dim() - 1))
+    result = torch.where(mask_a, a, b)
+    return result
+
+
+@torch.no_grad()
+def zero_module(module):
+    for p in module.parameters():
+        p.detach().zero_()
+    return module
+
+
+def load_last_state(model, folder='accelerator_output'):
+    file_pattern = os.path.join(folder, '**', 'model.safetensors')
+    files = glob.glob(file_pattern, recursive=True)
+
+    if not files:
+        print("No model.safetensors files found in the specified folder.")
+        return
+
+    newest_file = max(files, key=os.path.getmtime)
+    state_dict = sf.load_file(newest_file)
+    missing_keys, unexpected_keys = model.load_state_dict(state_dict, strict=False)
+
+    if missing_keys:
+        print("Missing keys:", missing_keys)
+    if unexpected_keys:
+        print("Unexpected keys:", unexpected_keys)
+
+    print("Loaded model state from:", newest_file)
+    return
+
+
+def generate_random_prompt_from_tags(tags_str, min_length=3, max_length=32):
+    tags = tags_str.split(', ')
+    tags = random.sample(tags, k=min(random.randint(min_length, max_length), len(tags)))
+    prompt = ', '.join(tags)
+    return prompt
+
+
+def save_bcthw_as_mp4(x, output_filename, fps=10):
+    b, c, t, h, w = x.shape
+
+    per_row = b
+    for p in [6, 5, 4, 3, 2]:
+        if b % p == 0:
+            per_row = p
+            break
+
+    os.makedirs(os.path.dirname(os.path.abspath(os.path.realpath(output_filename))), exist_ok=True)
+    x = torch.clamp(x.float(), -1., 1.) * 127.5 + 127.5
+    x = x.detach().cpu().to(torch.uint8)
+    x = einops.rearrange(x, '(m n) c t h w -> t (m h) (n w) c', n=per_row)
+    torchvision.io.write_video(output_filename, x, fps=fps, video_codec='h264', options={'crf': '0'})
+    return x
+
+
+def save_bcthw_as_png(x, output_filename):
+    os.makedirs(os.path.dirname(os.path.abspath(os.path.realpath(output_filename))), exist_ok=True)
+    x = torch.clamp(x.float(), -1., 1.) * 127.5 + 127.5
+    x = x.detach().cpu().to(torch.uint8)
+    x = einops.rearrange(x, 'b c t h w -> c (b h) (t w)')
+    torchvision.io.write_png(x, output_filename)
+    return output_filename
+
+
+def add_tensors_with_padding(tensor1, tensor2):
+    if tensor1.shape == tensor2.shape:
+        return tensor1 + tensor2
+
+    shape1 = tensor1.shape
+    shape2 = tensor2.shape
+
+    new_shape = tuple(max(s1, s2) for s1, s2 in zip(shape1, shape2))
+
+    padded_tensor1 = torch.zeros(new_shape)
+    padded_tensor2 = torch.zeros(new_shape)
+
+    padded_tensor1[tuple(slice(0, s) for s in shape1)] = tensor1
+    padded_tensor2[tuple(slice(0, s) for s in shape2)] = tensor2
+
+    result = padded_tensor1 + padded_tensor2
+    return result

requirements.txt

@@ -0,0 +1,15 @@
+diffusers==0.28.0
+transformers==4.41.1
+gradio==4.31.5
+bitsandbytes==0.43.1
+accelerate==0.30.1
+protobuf==3.20
+opencv-python
+tensorboardX
+safetensors
+pillow
+einops
+torch
+torchvision
+dghs-imgutils
+spaces