Add demo.

.gitignore

@@ -0,0 +1,4 @@
+output.png
+output-rembg.png
+__pycache__
+tmp

Dockerfile

@@ -0,0 +1,38 @@
+FROM nvidia/cuda:12.1.1-cudnn8-devel-ubuntu22.04
+
+ARG DEBIAN_FRONTEND=noninteractive
+
+ENV PYTHONUNBUFFERED=1
+
+RUN apt-get update && apt-get install --no-install-recommends -y \
+  build-essential \
+  python3.9 \
+  python3-pip \
+  git \
+  ffmpeg \
+  && apt-get clean && rm -rf /var/lib/apt/lists/*
+
+WORKDIR /code
+
+COPY ./requirements.txt /code/requirements.txt
+
+# Set up a new user named "user" with user ID 1000
+RUN useradd -m -u 1000 user
+# Switch to the "user" user
+USER user
+# Set home to the user's home directory
+ENV HOME=/home/user \
+	PATH=/home/user/.local/bin:$PATH \
+    PYTHONPATH=$HOME/app \
+	PYTHONUNBUFFERED=1 \
+	SYSTEM=spaces
+
+RUN pip3 install --no-cache-dir --upgrade -r /code/requirements.txt
+
+# Set the working directory to the user's home directory
+WORKDIR $HOME/app
+
+# Copy the current directory contents into the container at $HOME/app setting the owner to the user
+COPY --chown=user . $HOME/app
+
+CMD ["streamlit", "run", "app.py"]

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 2023 eliphat
+
+   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.

app.py

@@ -0,0 +1,258 @@
+import os
+import sys
+import numpy
+import torch
+import rembg
+import urllib.request
+from PIL import Image
+import streamlit as st
+
+
+img_example_counter = 0
+iret_base = 'resources/examples'
+iret = [
+    dict(rimageinput=os.path.join(iret_base, x), dispi=os.path.join(iret_base, x))
+    for x in sorted(os.listdir(iret_base))
+]
+
+
+class SAMAPI:
+    predictor = None
+
+    @staticmethod
+    @st.cache_resource
+    def get_instance(sam_checkpoint=None):
+        if SAMAPI.predictor is None:
+            if sam_checkpoint is None:
+                sam_checkpoint = "tmp/sam_vit_h_4b8939.pth"
+            if not os.path.exists(sam_checkpoint):
+                urllib.request.urlretrieve(
+                    "https://dl.fbaipublicfiles.com/segment_anything/sam_vit_h_4b8939.pth",
+                    sam_checkpoint
+                )
+            device = "cuda:0" if torch.cuda.is_available() else "cpu"
+            model_type = "default"
+
+            from segment_anything import sam_model_registry, SamPredictor
+
+            sam = sam_model_registry[model_type](checkpoint=sam_checkpoint)
+            sam.to(device=device)
+
+            predictor = SamPredictor(sam)
+            SAMAPI.predictor = predictor
+        return SAMAPI.predictor
+
+    @staticmethod
+    def segment_api(rgb, mask=None, bbox=None, sam_checkpoint=None):
+        """
+
+        Parameters
+        ----------
+        rgb : np.ndarray h,w,3 uint8
+        mask: np.ndarray h,w bool
+
+        Returns
+        -------
+
+        """
+        np = numpy
+        predictor = SAMAPI.get_instance(sam_checkpoint)
+        predictor.set_image(rgb)
+        if mask is None and bbox is None:
+            box_input = None
+        else:
+            # mask to bbox
+            if bbox is None:
+                y1, y2, x1, x2 = np.nonzero(mask)[0].min(), np.nonzero(mask)[0].max(), np.nonzero(mask)[1].min(), \
+                                 np.nonzero(mask)[1].max()
+            else:
+                x1, y1, x2, y2 = bbox
+            box_input = np.array([[x1, y1, x2, y2]])
+        masks, scores, logits = predictor.predict(
+            box=box_input,
+            multimask_output=True,
+            return_logits=False,
+        )
+        mask = masks[-1]
+        return mask
+
+
+def image_examples(samples, ncols, return_key=None, example_text="Examples"):
+    global img_example_counter
+    trigger = False
+    with st.expander(example_text, True):
+        for i in range(len(samples) // ncols):
+            cols = st.columns(ncols)
+            for j in range(ncols):
+                idx = i * ncols + j
+                if idx >= len(samples):
+                    continue
+                entry = samples[idx]
+                with cols[j]:
+                    st.image(entry['dispi'])
+                    img_example_counter += 1
+                    with st.columns(5)[2]:
+                        this_trigger = st.button('\+', key='imgexuse%d' % img_example_counter)
+                    trigger = trigger or this_trigger
+                    if this_trigger:
+                        trigger = entry[return_key]
+    return trigger
+
+
+def segment_img(img: Image):
+    output = rembg.remove(img)
+    mask = numpy.array(output)[:, :, 3] > 0
+    sam_mask = SAMAPI.segment_api(numpy.array(img)[:, :, :3], mask)
+    segmented_img = Image.new("RGBA", img.size, (0, 0, 0, 0))
+    segmented_img.paste(img, mask=Image.fromarray(sam_mask))
+    return segmented_img
+
+
+def segment_6imgs(zero123pp_imgs):
+    imgs = [zero123pp_imgs.crop([0, 0, 320, 320]),
+            zero123pp_imgs.crop([320, 0, 640, 320]),
+            zero123pp_imgs.crop([0, 320, 320, 640]),
+            zero123pp_imgs.crop([320, 320, 640, 640]),
+            zero123pp_imgs.crop([0, 640, 320, 960]),
+            zero123pp_imgs.crop([320, 640, 640, 960])]
+    segmented_imgs = []
+    for i, img in enumerate(imgs):
+        output = rembg.remove(img)
+        mask = numpy.array(output)[:, :, 3]
+        mask = SAMAPI.segment_api(numpy.array(img)[:, :, :3], mask)
+        data = numpy.array(img)[:,:,:3]
+        data[mask == 0] = [255, 255, 255]
+        segmented_imgs.append(data)
+    result = numpy.concatenate([
+        numpy.concatenate([segmented_imgs[0], segmented_imgs[1]], axis=1),
+        numpy.concatenate([segmented_imgs[2], segmented_imgs[3]], axis=1),
+        numpy.concatenate([segmented_imgs[4], segmented_imgs[5]], axis=1)
+    ])
+    return Image.fromarray(result)
+
+
+def expand2square(pil_img, background_color):
+    width, height = pil_img.size
+    if width == height:
+        return pil_img
+    elif width > height:
+        result = Image.new(pil_img.mode, (width, width), background_color)
+        result.paste(pil_img, (0, (width - height) // 2))
+        return result
+    else:
+        result = Image.new(pil_img.mode, (height, height), background_color)
+        result.paste(pil_img, ((height - width) // 2, 0))
+        return result
+
+
+@st.cache_data
+def check_dependencies():
+    reqs = []
+    try:
+        import diffusers
+    except ImportError:
+        import traceback
+        traceback.print_exc()
+        print("Error: `diffusers` not found.", file=sys.stderr)
+        reqs.append("diffusers==0.20.2")
+    else:
+        if not diffusers.__version__.startswith("0.20"):
+            print(
+                f"Warning: You are using an unsupported version of diffusers ({diffusers.__version__}), which may lead to performance issues.",
+                file=sys.stderr
+            )
+            print("Recommended version is `diffusers==0.20.2`.", file=sys.stderr)
+    try:
+        import transformers
+    except ImportError:
+        import traceback
+        traceback.print_exc()
+        print("Error: `transformers` not found.", file=sys.stderr)
+        reqs.append("transformers==4.29.2")
+    if torch.__version__ < '2.0':
+        try:
+            import xformers
+        except ImportError:
+            print("Warning: You are using PyTorch 1.x without a working `xformers` installation.", file=sys.stderr)
+            print("You may see a significant memory overhead when running the model.", file=sys.stderr)
+    if len(reqs):
+        print(f"Info: Fix all dependency errors with `pip install {' '.join(reqs)}`.")
+
+
+@st.cache_resource
+def load_zero123plus_pipeline():
+    from diffusers import DiffusionPipeline, EulerAncestralDiscreteScheduler
+    pipeline = DiffusionPipeline.from_pretrained(
+    "sudo-ai/zero123plus-v1.1", custom_pipeline="sudo-ai/zero123plus-pipeline",
+    torch_dtype=torch.float16
+    )
+    # Feel free to tune the scheduler
+    pipeline.scheduler = EulerAncestralDiscreteScheduler.from_config(
+        pipeline.scheduler.config, timestep_spacing='trailing'
+    )
+    if torch.cuda.is_available():
+        pipeline.to('cuda:0')
+    return pipeline
+
+
+check_dependencies()
+pipeline = load_zero123plus_pipeline()
+SAMAPI.get_instance()
+torch.set_grad_enabled(False)
+
+st.title("Zero123++ Demo")
+# st.caption("For faster inference without waiting in queue, you may clone the space and run it yourself.")
+prog = st.progress(0.0, "Idle")
+with st.form("imgform"):
+    pic = st.file_uploader("Upload an Image", key='imageinput', type=['png', 'jpg', 'webp'])
+    left, right = st.columns(2)
+    with left:
+        rem_input_bg = st.checkbox("Remove Input Background")
+    with right:
+        rem_output_bg = st.checkbox("Remove Output Background")
+    num_inference_steps = st.slider("Number of Inference Steps", 15, 100, 75)
+    st.caption("Diffusion Steps. For general real or synthetic objects, around 28 is enough. For objects with delicate details such as faces (either realistic or illustration), you may need 75 or more steps.")
+    cfg_scale = st.slider("Classifier Free Guidance Scale", 1.0, 10.0, 4.0)
+    seed = st.text_input("Seed", "42")
+    submit = False
+    if st.form_submit_button("Submit"):
+        submit = True
+    results_container = st.container()
+sample_got = image_examples(iret, 4, 'rimageinput')
+if sample_got:
+    pic = sample_got
+with results_container:
+    if sample_got or submit:
+        seed = int(seed)
+        torch.manual_seed(seed)
+        img = Image.open(pic)
+        left, right = st.columns(2)
+        with left:
+            st.image(img)
+            st.caption("Input Image")
+        prog.progress(0.1, "Preparing Inputs")
+        if rem_input_bg:
+            with right:
+                img = segment_img(img)
+                st.image(img)
+                st.caption("Input (Background Removed)")
+        img = expand2square(img, (127, 127, 127, 0))
+        pipeline.set_progress_bar_config(disable=True)
+        result = pipeline(
+            img,
+            num_inference_steps=num_inference_steps,
+            guidance_scale=cfg_scale,
+            generator=torch.Generator(pipeline.device).manual_seed(seed),
+            callback=lambda i, t, latents: prog.progress(0.1 + 0.8 * i / num_inference_steps, "Diffusion Step %d" % i)
+        ).images[0]
+        prog.progress(0.9, "Post Processing")
+        left, right = st.columns(2)
+        with left:
+            st.image(result)
+            st.caption("Result")
+        if rem_output_bg:
+            result = segment_6imgs(result)
+            with right:
+                st.image(result)
+                st.caption("Result (Background Removed)")
+        prog.progress(1.0, "Idle")

diffusers-support/pipeline.py

@@ -0,0 +1,398 @@
+from typing import Any, Dict, Optional
+from diffusers.models import AutoencoderKL, UNet2DConditionModel
+from diffusers.schedulers import KarrasDiffusionSchedulers
+
+import numpy
+import torch
+import torch.nn as nn
+import torch.utils.checkpoint
+import torch.distributed
+import transformers
+from collections import OrderedDict
+from PIL import Image
+from torchvision import transforms
+from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer
+
+import diffusers
+from diffusers import (
+    AutoencoderKL,
+    DDPMScheduler,
+    DiffusionPipeline,
+    EulerAncestralDiscreteScheduler,
+    UNet2DConditionModel,
+    ImagePipelineOutput
+)
+from diffusers.image_processor import VaeImageProcessor
+from diffusers.models.attention_processor import Attention, AttnProcessor, XFormersAttnProcessor, AttnProcessor2_0
+from diffusers.utils.import_utils import is_xformers_available
+
+
+def to_rgb_image(maybe_rgba: Image.Image):
+    if maybe_rgba.mode == 'RGB':
+        return maybe_rgba
+    elif maybe_rgba.mode == 'RGBA':
+        rgba = maybe_rgba
+        img = numpy.random.randint(127, 128, size=[rgba.size[1], rgba.size[0], 3], dtype=numpy.uint8)
+        img = Image.fromarray(img, 'RGB')
+        img.paste(rgba, mask=rgba.getchannel('A'))
+        return img
+    else:
+        raise ValueError("Unsupported image type.", maybe_rgba.mode)
+
+
+class ReferenceOnlyAttnProc(torch.nn.Module):
+    def __init__(
+        self,
+        chained_proc,
+        enabled=False,
+        name=None
+    ) -> None:
+        super().__init__()
+        self.enabled = enabled
+        self.chained_proc = chained_proc
+        self.name = name
+
+    def __call__(
+        self, attn: Attention, hidden_states, encoder_hidden_states=None, attention_mask=None,
+        mode="w", ref_dict: dict = None, is_cfg_guidance = False
+    ) -> Any:
+        if encoder_hidden_states is None:
+            encoder_hidden_states = hidden_states
+        if self.enabled and is_cfg_guidance:
+            res0 = self.chained_proc(attn, hidden_states[:1], encoder_hidden_states[:1], attention_mask)
+            hidden_states = hidden_states[1:]
+            encoder_hidden_states = encoder_hidden_states[1:]
+        if self.enabled:
+            if mode == 'w':
+                ref_dict[self.name] = encoder_hidden_states
+            elif mode == 'r':
+                encoder_hidden_states = torch.cat([encoder_hidden_states, ref_dict.pop(self.name)], dim=1)
+            elif mode == 'm':
+                encoder_hidden_states = torch.cat([encoder_hidden_states, ref_dict[self.name]], dim=1)
+            else:
+                assert False, mode
+        res = self.chained_proc(attn, hidden_states, encoder_hidden_states, attention_mask)
+        if self.enabled and is_cfg_guidance:
+            res = torch.cat([res0, res])
+        return res
+
+
+class RefOnlyNoisedUNet(torch.nn.Module):
+    def __init__(self, unet: UNet2DConditionModel, train_sched: DDPMScheduler, val_sched: EulerAncestralDiscreteScheduler) -> None:
+        super().__init__()
+        self.unet = unet
+        self.train_sched = train_sched
+        self.val_sched = val_sched
+
+        unet_lora_attn_procs = dict()
+        for name, _ in unet.attn_processors.items():
+            if torch.__version__ >= '2.0':
+                default_attn_proc = AttnProcessor2_0()
+            elif is_xformers_available():
+                default_attn_proc = XFormersAttnProcessor()
+            else:
+                default_attn_proc = AttnProcessor()
+            unet_lora_attn_procs[name] = ReferenceOnlyAttnProc(
+                default_attn_proc, enabled=name.endswith("attn1.processor"), name=name
+            )
+        unet.set_attn_processor(unet_lora_attn_procs)
+
+    def __getattr__(self, name: str):
+        try:
+            return super().__getattr__(name)
+        except AttributeError:
+            return getattr(self.unet, name)
+
+    def forward_cond(self, noisy_cond_lat, timestep, encoder_hidden_states, class_labels, ref_dict, is_cfg_guidance, **kwargs):
+        if is_cfg_guidance:
+            encoder_hidden_states = encoder_hidden_states[1:]
+            class_labels = class_labels[1:]
+        self.unet(
+            noisy_cond_lat, timestep,
+            encoder_hidden_states=encoder_hidden_states,
+            class_labels=class_labels,
+            cross_attention_kwargs=dict(mode="w", ref_dict=ref_dict),
+            **kwargs
+        )
+
+    def forward(
+        self, sample, timestep, encoder_hidden_states, class_labels=None,
+        *args, cross_attention_kwargs,
+        down_block_res_samples=None, mid_block_res_sample=None,
+        **kwargs
+    ):
+        cond_lat = cross_attention_kwargs['cond_lat']
+        is_cfg_guidance = cross_attention_kwargs.get('is_cfg_guidance', False)
+        noise = torch.randn_like(cond_lat)
+        if self.training:
+            noisy_cond_lat = self.train_sched.add_noise(cond_lat, noise, timestep)
+            noisy_cond_lat = self.train_sched.scale_model_input(noisy_cond_lat, timestep)
+        else:
+            noisy_cond_lat = self.val_sched.add_noise(cond_lat, noise, timestep.reshape(-1))
+            noisy_cond_lat = self.val_sched.scale_model_input(noisy_cond_lat, timestep.reshape(-1))
+        ref_dict = {}
+        self.forward_cond(
+            noisy_cond_lat, timestep,
+            encoder_hidden_states, class_labels,
+            ref_dict, is_cfg_guidance, **kwargs
+        )
+        weight_dtype = self.unet.dtype
+        return self.unet(
+            sample, timestep,
+            encoder_hidden_states, *args,
+            class_labels=class_labels,
+            cross_attention_kwargs=dict(mode="r", ref_dict=ref_dict, is_cfg_guidance=is_cfg_guidance),
+            down_block_additional_residuals=[
+                sample.to(dtype=weight_dtype) for sample in down_block_res_samples
+            ] if down_block_res_samples is not None else None,
+            mid_block_additional_residual=(
+                mid_block_res_sample.to(dtype=weight_dtype)
+                if mid_block_res_sample is not None else None
+            ),
+            **kwargs
+        )
+
+
+def scale_latents(latents):
+    latents = (latents - 0.22) * 0.75
+    return latents
+
+
+def unscale_latents(latents):
+    latents = latents / 0.75 + 0.22
+    return latents
+
+
+def scale_image(image):
+    image = image * 0.5 / 0.8
+    return image
+
+
+def unscale_image(image):
+    image = image / 0.5 * 0.8
+    return image
+
+
+class DepthControlUNet(torch.nn.Module):
+    def __init__(self, unet: RefOnlyNoisedUNet, controlnet: Optional[diffusers.ControlNetModel] = None, conditioning_scale=1.0) -> None:
+        super().__init__()
+        self.unet = unet
+        if controlnet is None:
+            self.controlnet = diffusers.ControlNetModel.from_unet(unet.unet)
+        else:
+            self.controlnet = controlnet
+        DefaultAttnProc = AttnProcessor2_0
+        if is_xformers_available():
+            DefaultAttnProc = XFormersAttnProcessor
+        self.controlnet.set_attn_processor(DefaultAttnProc())
+        self.conditioning_scale = conditioning_scale
+
+    def __getattr__(self, name: str):
+        try:
+            return super().__getattr__(name)
+        except AttributeError:
+            return getattr(self.unet, name)
+
+    def forward(self, sample, timestep, encoder_hidden_states, class_labels=None, *args, cross_attention_kwargs: dict, **kwargs):
+        cross_attention_kwargs = dict(cross_attention_kwargs)
+        control_depth = cross_attention_kwargs.pop('control_depth')
+        down_block_res_samples, mid_block_res_sample = self.controlnet(
+            sample,
+            timestep,
+            encoder_hidden_states=encoder_hidden_states,
+            controlnet_cond=control_depth,
+            conditioning_scale=self.conditioning_scale,
+            return_dict=False,
+        )
+        return self.unet(
+            sample,
+            timestep,
+            encoder_hidden_states=encoder_hidden_states,
+            down_block_res_samples=down_block_res_samples,
+            mid_block_res_sample=mid_block_res_sample,
+            cross_attention_kwargs=cross_attention_kwargs
+        )
+
+
+class ModuleListDict(torch.nn.Module):
+    def __init__(self, procs: dict) -> None:
+        super().__init__()
+        self.keys = sorted(procs.keys())
+        self.values = torch.nn.ModuleList(procs[k] for k in self.keys)
+
+    def __getitem__(self, key):
+        return self.values[self.keys.index(key)]
+
+
+class SuperNet(torch.nn.Module):
+    def __init__(self, state_dict: Dict[str, torch.Tensor]):
+        super().__init__()
+        state_dict = OrderedDict((k, state_dict[k]) for k in sorted(state_dict.keys()))
+        self.layers = torch.nn.ModuleList(state_dict.values())
+        self.mapping = dict(enumerate(state_dict.keys()))
+        self.rev_mapping = {v: k for k, v in enumerate(state_dict.keys())}
+
+        # .processor for unet, .self_attn for text encoder
+        self.split_keys = [".processor", ".self_attn"]
+
+        # we add a hook to state_dict() and load_state_dict() so that the
+        # naming fits with `unet.attn_processors`
+        def map_to(module, state_dict, *args, **kwargs):
+            new_state_dict = {}
+            for key, value in state_dict.items():
+                num = int(key.split(".")[1])  # 0 is always "layers"
+                new_key = key.replace(f"layers.{num}", module.mapping[num])
+                new_state_dict[new_key] = value
+
+            return new_state_dict
+
+        def remap_key(key, state_dict):
+            for k in self.split_keys:
+                if k in key:
+                    return key.split(k)[0] + k
+            return key.split('.')[0]
+
+        def map_from(module, state_dict, *args, **kwargs):
+            all_keys = list(state_dict.keys())
+            for key in all_keys:
+                replace_key = remap_key(key, state_dict)
+                new_key = key.replace(replace_key, f"layers.{module.rev_mapping[replace_key]}")
+                state_dict[new_key] = state_dict[key]
+                del state_dict[key]
+
+        self._register_state_dict_hook(map_to)
+        self._register_load_state_dict_pre_hook(map_from, with_module=True)
+
+
+class Zero123PlusPipeline(diffusers.StableDiffusionPipeline):
+    tokenizer: transformers.CLIPTokenizer
+    text_encoder: transformers.CLIPTextModel
+    vision_encoder: transformers.CLIPVisionModelWithProjection
+
+    feature_extractor_clip: transformers.CLIPImageProcessor
+    unet: UNet2DConditionModel
+    scheduler: diffusers.schedulers.KarrasDiffusionSchedulers
+
+    vae: AutoencoderKL
+    ramping: nn.Linear
+
+    feature_extractor_vae: transformers.CLIPImageProcessor
+
+    depth_transforms_multi = transforms.Compose([
+        transforms.ToTensor(),
+        transforms.Normalize([0.5], [0.5])
+    ])
+
+    def __init__(
+        self,
+        vae: AutoencoderKL,
+        text_encoder: CLIPTextModel,
+        tokenizer: CLIPTokenizer,
+        unet: UNet2DConditionModel,
+        scheduler: KarrasDiffusionSchedulers,
+        vision_encoder: transformers.CLIPVisionModelWithProjection,
+        feature_extractor_clip: CLIPImageProcessor, 
+        feature_extractor_vae: CLIPImageProcessor,
+        ramping_coefficients: Optional[list] = None,
+        safety_checker=None,
+    ):
+        DiffusionPipeline.__init__(self)
+
+        self.register_modules(
+            vae=vae, text_encoder=text_encoder, tokenizer=tokenizer,
+            unet=unet, scheduler=scheduler, safety_checker=None,
+            vision_encoder=vision_encoder,
+            feature_extractor_clip=feature_extractor_clip,
+            feature_extractor_vae=feature_extractor_vae
+        )
+        self.register_to_config(ramping_coefficients=ramping_coefficients)
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
+
+    def prepare(self):
+        train_sched = DDPMScheduler.from_config(self.scheduler.config)
+        if isinstance(self.unet, UNet2DConditionModel):
+            self.unet = RefOnlyNoisedUNet(self.unet, train_sched, self.scheduler).eval()
+
+    def add_controlnet(self, controlnet: Optional[diffusers.ControlNetModel] = None, conditioning_scale=1.0):
+        self.prepare()
+        self.unet = DepthControlUNet(self.unet, controlnet, conditioning_scale)
+        return SuperNet(OrderedDict([('controlnet', self.unet.controlnet)]))
+
+    def encode_condition_image(self, image: torch.Tensor):
+        image = self.vae.encode(image).latent_dist.sample()
+        return image
+
+    @torch.no_grad()
+    def __call__(
+        self,
+        image: Image.Image = None,
+        prompt = "",
+        *args,
+        num_images_per_prompt: Optional[int] = 1,
+        guidance_scale=4.0,
+        depth_image: Image.Image = None,
+        output_type: Optional[str] = "pil",
+        width=640,
+        height=960,
+        num_inference_steps=28,
+        return_dict=True,
+        **kwargs
+    ):
+        self.prepare()
+        if image is None:
+            raise ValueError("Inputting embeddings not supported for this pipeline. Please pass an image.")
+        assert not isinstance(image, torch.Tensor)
+        image = to_rgb_image(image)
+        image_1 = self.feature_extractor_vae(images=image, return_tensors="pt").pixel_values
+        image_2 = self.feature_extractor_clip(images=image, return_tensors="pt").pixel_values
+        if depth_image is not None and hasattr(self.unet, "controlnet"):
+            depth_image = to_rgb_image(depth_image)
+            depth_image = self.depth_transforms_multi(depth_image).to(
+                device=self.unet.controlnet.device, dtype=self.unet.controlnet.dtype
+            )
+        image = image_1.to(device=self.vae.device, dtype=self.vae.dtype)
+        image_2 = image_2.to(device=self.vae.device, dtype=self.vae.dtype)
+        cond_lat = self.encode_condition_image(image)
+        if guidance_scale > 1:
+            negative_lat = self.encode_condition_image(torch.zeros_like(image))
+            cond_lat = torch.cat([negative_lat, cond_lat])
+        encoded = self.vision_encoder(image_2, output_hidden_states=False)
+        global_embeds = encoded.image_embeds
+        global_embeds = global_embeds.unsqueeze(-2)
+        
+        encoder_hidden_states = self._encode_prompt(
+            prompt,
+            self.device,
+            num_images_per_prompt,
+            False
+        )
+        ramp = global_embeds.new_tensor(self.config.ramping_coefficients).unsqueeze(-1)
+        encoder_hidden_states = encoder_hidden_states + global_embeds * ramp
+        cak = dict(cond_lat=cond_lat)
+        if hasattr(self.unet, "controlnet"):
+            cak['control_depth'] = depth_image
+        latents: torch.Tensor = super().__call__(
+            None,
+            *args,
+            cross_attention_kwargs=cak,
+            guidance_scale=guidance_scale,
+            num_images_per_prompt=num_images_per_prompt,
+            prompt_embeds=encoder_hidden_states,
+            num_inference_steps=num_inference_steps,
+            output_type='latent',
+            width=width,
+            height=height,
+            **kwargs
+        ).images
+        latents = unscale_latents(latents)
+        if not output_type == "latent":
+            image = unscale_image(self.vae.decode(latents / self.vae.config.scaling_factor, return_dict=False)[0])
+        else:
+            image = latents
+
+        image = self.image_processor.postprocess(image, output_type=output_type)
+        if not return_dict:
+            return (image,)
+
+        return ImagePipelineOutput(images=image)

examples/depth_controlnet.py

@@ -0,0 +1,24 @@
+import torch
+import requests
+from PIL import Image
+from diffusers import DiffusionPipeline, EulerAncestralDiscreteScheduler, ControlNetModel
+
+# Load the pipeline
+pipeline = DiffusionPipeline.from_pretrained(
+    "sudo-ai/zero123plus-v1.1", custom_pipeline="sudo-ai/zero123plus-pipeline",
+    torch_dtype=torch.float16
+)
+pipeline.add_controlnet(ControlNetModel.from_pretrained(
+    "sudo-ai/controlnet-zp11-depth-v1", torch_dtype=torch.float16
+), conditioning_scale=0.75)
+# Feel free to tune the scheduler
+pipeline.scheduler = EulerAncestralDiscreteScheduler.from_config(
+    pipeline.scheduler.config, timestep_spacing='trailing'
+)
+pipeline.to('cuda:0')
+# Run the pipeline
+cond = Image.open(requests.get("https://d.skis.ltd/nrp/sample-data/0_cond.png", stream=True).raw)
+depth = Image.open(requests.get("https://d.skis.ltd/nrp/sample-data/0_depth.png", stream=True).raw)
+result = pipeline(cond, depth_image=depth, num_inference_steps=36).images[0]
+result.show()
+result.save("output.png")

examples/img_to_mv.py

@@ -0,0 +1,20 @@
+import torch
+import requests
+from PIL import Image
+from diffusers import DiffusionPipeline, EulerAncestralDiscreteScheduler
+
+# Load the pipeline
+pipeline = DiffusionPipeline.from_pretrained(
+    "sudo-ai/zero123plus-v1.1", custom_pipeline="sudo-ai/zero123plus-pipeline",
+    torch_dtype=torch.float16
+)
+# Feel free to tune the scheduler
+pipeline.scheduler = EulerAncestralDiscreteScheduler.from_config(
+    pipeline.scheduler.config, timestep_spacing='trailing'
+)
+pipeline.to('cuda:0')
+# Run the pipeline
+cond = Image.open(requests.get("https://d.skis.ltd/nrp/sample-data/lysol.png", stream=True).raw)
+result = pipeline(cond, num_inference_steps=75).images[0]
+result.show()
+result.save("output.png")

examples/text_to_img.py

@@ -0,0 +1,24 @@
+from diffusers import StableDiffusionXLPipeline
+import torch
+import rembg
+
+# text-to-image with SDXL for text-to-image-to-3d
+pipeline = StableDiffusionXLPipeline.from_single_file(
+    "https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/sd_xl_base_1.0.safetensors",
+    torch_dtype=torch.float16,
+    variant="fp16",
+    use_safetensors=True
+).to("cuda")
+pipeline.enable_model_cpu_offload()
+
+num_images_per_prompt = 1
+res = 1024
+text = input("Prompt > ")
+bkgd_color = "white"
+prompt = f"a ((full-body:2)) shot of a ((single:2)) {text}, isolated on {bkgd_color} background, 4k, highly detailed"
+images = pipeline(prompt=prompt, num_images_per_prompt=num_images_per_prompt, height=res, width=res).images
+image = images[0]
+image.show()
+image.save("output.png")
+image = rembg.remove(image)
+image.save("output-rembg.png")

requirements.txt

@@ -0,0 +1,10 @@
+torch
+torchvision
+numpy
+rembg
+opencv-contrib-python
+diffusers==0.20.2
+transformers==4.29.2
+streamlit==1.22.0
+altair<5
+git+https://github.com/facebookresearch/segment-anything.git