AI/diffusers
1
0
Fork 0
mirror of https://github.com/huggingface/diffusers.git synced 2026-01-27 17:22:53 +03:00
Code Activity

Diffedit Zero-Shot Inpainting Pipeline (#2837)

Browse Source 
* Update Pix2PixZero Auto-correlation Loss

* Add Stable Diffusion DiffEdit pipeline

* Add draft documentation and import code

* Bugfixes and refactoring

* Add option to not decode latents in the inversion process

* Harmonize preprocessing

* Revert "Update Pix2PixZero Auto-correlation Loss"

This reverts commit b218062fed.

* Update annotations

* rename `compute_mask` to `generate_mask`

* Update documentation

* Update docs

* Update Docs

* Fix copy

* Change shape of output latents to batch first

* Update docs

* Add first draft for tests

* Bugfix and update tests

* Add `cross_attention_kwargs` support for all pipeline methods

* Fix Copies

* Add support for PIL image latents

Add support for mask broadcasting

Update docs and tests

Align `mask` argument to `mask_image`

Remove height and width arguments

* Enable MPS Tests

* Move example docstrings

* Fix test

* Fix test

* fix pipeline inheritance

* Harmonize `prepare_image_latents` with StableDiffusionPix2PixZeroPipeline

* Register modules set to `None` in config for `test_save_load_optional_components`

* Move fixed logic to specific test class

* Clean changes to other pipelines

* Update new tests to coordinate with #2953

* Update slow tests for better results

* Safety to avoid potential problems with torch.inference_mode

* Add reference in SD Pipeline Overview

* Fix tests again

* Enforce determinism in noise for generate_mask

* Fix copies

* Widen test tolerance for fp16 based on `test_stable_diffusion_upscale_pipeline_fp16`

* Add LoraLoaderMixin and update `prepare_image_latents`

* clean up repeat and reg

* bugfix

* Remove invalid args from docs

Suppress spurious warning by repeating image before latent to mask gen
This commit is contained in:
clarencechen
2023-04-28 03:58:26 -07:00
committed by GitHub
parent d464214464
commit be0bfcec4d
9 changed files with 2227 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
docs/source/en/_toctree.yml
View File

@@ -203,6 +203,8 @@
title: Text-to-Image Generation with ControlNet Conditioning
- local: api/pipelines/stable_diffusion/model_editing
title: Text-to-Image Model Editing
- local: api/pipelines/stable_diffusion/diffedit
title: DiffEdit
title: Stable Diffusion
- local: api/pipelines/stable_diffusion_2
title: Stable Diffusion 2

360
docs/source/en/api/pipelines/stable_diffusion/diffedit.mdx Normal file
View File

@@ -0,0 +1,360 @@
<!--Copyright 2023 The HuggingFace Team. All rights reserved.
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.
-->
# Zero-shot Diffusion-based Semantic Image Editing with Mask Guidance
## Overview
[DiffEdit: Diffusion-based semantic image editing with mask guidance](https://arxiv.org/abs/2210.11427) by Guillaume Couairon, Jakob Verbeek, Holger Schwenk, and Matthieu Cord.
The abstract of the paper is the following:
*Image generation has recently seen tremendous advances, with diffusion models allowing to synthesize convincing images for a large variety of text prompts. In this article, we propose DiffEdit, a method to take advantage of text-conditioned diffusion models for the task of semantic image editing, where the goal is to edit an image based on a text query. Semantic image editing is an extension of image generation, with the additional constraint that the generated image should be as similar as possible to a given input image. Current editing methods based on diffusion models usually require to provide a mask, making the task much easier by treating it as a conditional inpainting task. In contrast, our main contribution is able to automatically generate a mask highlighting regions of the input image that need to be edited, by contrasting predictions of a diffusion model conditioned on different text prompts. Moreover, we rely on latent inference to preserve content in those regions of interest and show excellent synergies with mask-based diffusion. DiffEdit achieves state-of-the-art editing performance on ImageNet. In addition, we evaluate semantic image editing in more challenging settings, using images from the COCO dataset as well as text-based generated images.*
Resources:
* [Paper](https://arxiv.org/abs/2210.11427).
* [Blog Post with Demo](https://blog.problemsolversguild.com/technical/research/2022/11/02/DiffEdit-Implementation.html).
* [Implementation on Github](https://github.com/Xiang-cd/DiffEdit-stable-diffusion/).
## Tips
* The pipeline can generate masks that can be fed into other inpainting pipelines. Check out the code examples below to know more.
* In order to generate an image using this pipeline, both an image mask (manually specified or generated using `generate_mask`)
and a set of partially inverted latents (generated using `invert`) _must_ be provided as arguments when calling the pipeline to generate the final edited image.
Refer to the code examples below for more details.
* The function `generate_mask` exposes two prompt arguments, `source_prompt` and `target_prompt`,
that let you control the locations of the semantic edits in the final image to be generated. Let's say,
you wanted to translate from "cat" to "dog". In this case, the edit direction will be "cat -> dog". To reflect
this in the generated mask, you simply have to set the embeddings related to the phrases including "cat" to
`source_prompt_embeds` and "dog" to `target_prompt_embeds`. Refer to the code example below for more details.
* When generating partially inverted latents using `invert`, assign a caption or text embedding describing the
overall image to the `prompt` argument to help guide the inverse latent sampling process. In most cases, the
source concept is sufficently descriptive to yield good results, but feel free to explore alternatives.
Please refer to [this code example](#generating-image-captions-for-inversion) for more details.
* When calling the pipeline to generate the final edited image, assign the source concept to `negative_prompt`
and the target concept to `prompt`. Taking the above example, you simply have to set the embeddings related to
the phrases including "cat" to `negative_prompt_embeds` and "dog" to `prompt_embeds`. Refer to the code example
below for more details.
* If you wanted to reverse the direction in the example above, i.e., "dog -> cat", then it's recommended to:
* Swap the `source_prompt` and `target_prompt` in the arguments to `generate_mask`.
* Change the input prompt for `invert` to include "dog".
* Swap the `prompt` and `negative_prompt` in the arguments to call the pipeline to generate the final edited image.
* Note that the source and target prompts, or their corresponding embeddings, can also be automatically generated. Please, refer to [this discussion](#generating-source-and-target-embeddings) for more details.
## Available Pipelines:
| Pipeline | Tasks
|---|---|
| [StableDiffusionDiffEditPipeline](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_diffedit.py) | *Text-Based Image Editing*
<!-- TODO: add Colab -->
## Usage example
### Based on an input image with a caption
When the pipeline is conditioned on an input image, we first obtain partially inverted latents from the input image using a
`DDIMInverseScheduler` with the help of a caption. Then we generate an editing mask to identify relevant regions in the image using the source and target prompts. Finally,
the inverted noise and generated mask is used to start the generation process.
First, let's load our pipeline:
```py
import torch
from diffusers import DDIMScheduler, DDIMInverseScheduler, StableDiffusionPix2PixZeroPipeline
sd_model_ckpt = "stabilityai/stable-diffusion-2-1"
pipeline = StableDiffusionDiffEditPipeline.from_pretrained(
sd_model_ckpt,
torch_dtype=torch.float16,
safety_checker=None,
)
pipeline.scheduler = DDIMScheduler.from_config(pipeline.scheduler.config)
pipeline.inverse_scheduler = DDIMInverseScheduler.from_config(pipeline.scheduler.config)
pipeline.enable_model_cpu_offload()
pipeline.enable_vae_slicing()
generator = torch.manual_seed(0)
```
Then, we load an input image to edit using our method:
```py
from diffusers.utils import load_image
img_url = "https://github.com/Xiang-cd/DiffEdit-stable-diffusion/raw/main/assets/origin.png"
raw_image = load_image(img_url).convert("RGB").resize((768, 768))
```
Then, we employ the source and target prompts to generate the editing mask:
```py
# See the "Generating source and target embeddings" section below to
# automate the generation of these captions with a pre-trained model like Flan-T5 as explained below.
source_prompt = "a bowl of fruits"
target_prompt = "a basket of fruits"
mask_image = pipeline.generate_mask(
image=raw_image,
source_prompt=source_prompt,
target_prompt=target_prompt,
generator=generator,
)
```
Then, we employ the caption and the input image to get the inverted latents:
```py
inv_latents = pipeline.invert(prompt=source_prompt, image=raw_image, generator=generator).latents
```
Now, generate the image with the inverted latents and semantically generated mask:
```py
image = pipeline(
prompt=target_prompt,
mask_image=mask_image,
image_latents=inv_latents,
generator=generator,
negative_prompt=source_prompt,
).images[0]
image.save("edited_image.png")
```
## Generating image captions for inversion
The authors originally used the source concept prompt as the caption for generating the partially inverted latents. However, we can also leverage open source and public image captioning models for the same purpose.
Below, we provide an end-to-end example with the [BLIP](https://huggingface.co/docs/transformers/model_doc/blip) model
for generating captions.
First, let's load our automatic image captioning model:
```py
import torch
from transformers import BlipForConditionalGeneration, BlipProcessor
captioner_id = "Salesforce/blip-image-captioning-base"
processor = BlipProcessor.from_pretrained(captioner_id)
model = BlipForConditionalGeneration.from_pretrained(captioner_id, torch_dtype=torch.float16, low_cpu_mem_usage=True)
```
Then, we define a utility to generate captions from an input image using the model:
```py
@torch.no_grad()
def generate_caption(images, caption_generator, caption_processor):
text = "a photograph of"
inputs = caption_processor(images, text, return_tensors="pt").to(device="cuda", dtype=caption_generator.dtype)
caption_generator.to("cuda")
outputs = caption_generator.generate(**inputs, max_new_tokens=128)
# offload caption generator
caption_generator.to("cpu")
caption = caption_processor.batch_decode(outputs, skip_special_tokens=True)[0]
return caption
```
Then, we load an input image for conditioning and obtain a suitable caption for it:
```py
from diffusers.utils import load_image
img_url = "https://github.com/Xiang-cd/DiffEdit-stable-diffusion/raw/main/assets/origin.png"
raw_image = load_image(img_url).convert("RGB").resize((768, 768))
caption = generate_caption(raw_image, model, processor)
```
Then, we employ the generated caption and the input image to get the inverted latents:
```py
from diffusers import DDIMInverseScheduler, DDIMScheduler
pipeline = StableDiffusionDiffEditPipeline.from_pretrained(
"stabilityai/stable-diffusion-2-1", torch_dtype=torch.float16
)
pipeline = pipeline.to("cuda")
pipeline.enable_model_cpu_offload()
pipeline.enable_vae_slicing()
pipeline.scheduler = DDIMScheduler.from_config(pipeline.scheduler.config)
pipeline.inverse_scheduler = DDIMInverseScheduler.from_config(pipeline.scheduler.config)
generator = torch.manual_seed(0)
inv_latents = pipeline.invert(prompt=caption, image=raw_image, generator=generator).latents
```
Now, generate the image with the inverted latents and semantically generated mask from our source and target prompts:
```py
source_prompt = "a bowl of fruits"
target_prompt = "a basket of fruits"
mask_image = pipeline.generate_mask(
image=raw_image,
source_prompt=source_prompt,
target_prompt=target_prompt,
generator=generator,
)
image = pipeline(
prompt=target_prompt,
mask_image=mask_image,
image_latents=inv_latents,
generator=generator,
negative_prompt=source_prompt,
).images[0]
image.save("edited_image.png")
```
## Generating source and target embeddings
The authors originally required the user to manually provide the source and target prompts for discovering
edit directions. However, we can also leverage open source and public models for the same purpose.
Below, we provide an end-to-end example with the [Flan-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5) model
for generating source an target embeddings.
**1. Load the generation model**:
```py
import torch
from transformers import AutoTokenizer, T5ForConditionalGeneration
tokenizer = AutoTokenizer.from_pretrained("google/flan-t5-xl")
model = T5ForConditionalGeneration.from_pretrained("google/flan-t5-xl", device_map="auto", torch_dtype=torch.float16)
```
**2. Construct a starting prompt**:
```py
source_concept = "bowl"
target_concept = "basket"
source_text = f"Provide a caption for images containing a {source_concept}. "
"The captions should be in English and should be no longer than 150 characters."
target_text = f"Provide a caption for images containing a {target_concept}. "
"The captions should be in English and should be no longer than 150 characters."
```
Here, we're interested in the "bowl -> basket" direction.
**3. Generate prompts**:
We can use a utility like so for this purpose.
```py
@torch.no_grad
def generate_prompts(input_prompt):
input_ids = tokenizer(input_prompt, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(
input_ids, temperature=0.8, num_return_sequences=16, do_sample=True, max_new_tokens=128, top_k=10
)
return tokenizer.batch_decode(outputs, skip_special_tokens=True)
```
And then we just call it to generate our prompts:
```py
source_prompts = generate_prompts(source_text)
target_prompts = generate_prompts(target_text)
```
We encourage you to play around with the different parameters supported by the
`generate()` method ([documentation](https://huggingface.co/docs/transformers/main/en/main_classes/text_generation#transformers.generation_tf_utils.TFGenerationMixin.generate)) for the generation quality you are looking for.
**4. Load the embedding model**:
Here, we need to use the same text encoder model used by the subsequent Stable Diffusion model.
```py
from diffusers import StableDiffusionDiffEditPipeline
pipeline = StableDiffusionDiffEditPipeline.from_pretrained(
"stabilityai/stable-diffusion-2-1", torch_dtype=torch.float16
)
pipeline = pipeline.to("cuda")
pipeline.enable_model_cpu_offload()
pipeline.enable_vae_slicing()
generator = torch.manual_seed(0)
```
**5. Compute embeddings**:
```py
import torch
@torch.no_grad()
def embed_prompts(sentences, tokenizer, text_encoder, device="cuda"):
embeddings = []
for sent in sentences:
text_inputs = tokenizer(
sent,
padding="max_length",
max_length=tokenizer.model_max_length,
truncation=True,
return_tensors="pt",
)
text_input_ids = text_inputs.input_ids
prompt_embeds = text_encoder(text_input_ids.to(device), attention_mask=None)[0]
embeddings.append(prompt_embeds)
return torch.concatenate(embeddings, dim=0).mean(dim=0).unsqueeze(0)
source_embeddings = embed_prompts(source_prompts, pipeline.tokenizer, pipeline.text_encoder)
target_embeddings = embed_prompts(target_captions, pipeline.tokenizer, pipeline.text_encoder)
```
And you're done! Now, you can use these embeddings directly while calling the pipeline:
```py
from diffusers import DDIMInverseScheduler, DDIMScheduler
from diffusers.utils import load_image
pipeline.scheduler = DDIMScheduler.from_config(pipeline.scheduler.config)
pipeline.inverse_scheduler = DDIMInverseScheduler.from_config(pipeline.scheduler.config)
img_url = "https://github.com/Xiang-cd/DiffEdit-stable-diffusion/raw/main/assets/origin.png"
raw_image = load_image(img_url).convert("RGB").resize((768, 768))
mask_image = pipeline.generate_mask(
image=raw_image,
source_prompt_embeds=source_embeds,
target_prompt_embeds=target_embeds,
generator=generator,
)
inv_latents = pipeline.invert(
prompt_embeds=source_embeds,
image=raw_image,
generator=generator,
).latents
images = pipeline(
mask_image=mask_image,
image_latents=inv_latents,
prompt_embeds=target_embeddings,
negative_prompt_embeds=source_embeddings,
generator=generator,
).images
images[0].save("edited_image.png")
```
## StableDiffusionDiffEditPipeline
[[autodoc]] StableDiffusionDiffEditPipeline
- all
- generate_mask
- invert
- __call__

1
docs/source/en/api/pipelines/stable_diffusion/overview.mdx
View File

@@ -36,6 +36,7 @@ For more details about how Stable Diffusion works and how it differs from the ba
| [StableDiffusionAttendAndExcitePipeline](./attend_and_excite) | **Experimental** – *Text-to-Image Generation * | | [Attend-and-Excite: Attention-Based Semantic Guidance for Text-to-Image Diffusion Models](https://huggingface.co/spaces/AttendAndExcite/Attend-and-Excite)
| [StableDiffusionPix2PixZeroPipeline](./pix2pix_zero) | **Experimental** – *Text-Based Image Editing * | | [Zero-shot Image-to-Image Translation](https://arxiv.org/abs/2302.03027)
| [StableDiffusionModelEditingPipeline](./model_editing) | **Experimental** – *Text-to-Image Model Editing * | | [Editing Implicit Assumptions in Text-to-Image Diffusion Models](https://arxiv.org/abs/2303.08084)
| [StableDiffusionDiffEditPipeline](./diffedit) | **Experimental** – *Text-Based Image Editing * | | [DiffEdit: Diffusion-based semantic image editing with mask guidance](https://arxiv.org/abs/2210.11427)

1
src/diffusers/__init__.py
View File

@@ -134,6 +134,7 @@ else:
StableDiffusionAttendAndExcitePipeline,
StableDiffusionControlNetPipeline,
StableDiffusionDepth2ImgPipeline,
StableDiffusionDiffEditPipeline,
StableDiffusionImageVariationPipeline,
StableDiffusionImg2ImgPipeline,
StableDiffusionInpaintPipeline,

1
src/diffusers/pipelines/__init__.py
View File

@@ -60,6 +60,7 @@ else:
StableDiffusionAttendAndExcitePipeline,
StableDiffusionControlNetPipeline,
StableDiffusionDepth2ImgPipeline,
StableDiffusionDiffEditPipeline,
StableDiffusionImageVariationPipeline,
StableDiffusionImg2ImgPipeline,
StableDiffusionInpaintPipeline,

2
src/diffusers/pipelines/stable_diffusion/__init__.py
View File

@@ -75,10 +75,12 @@ try:
except OptionalDependencyNotAvailable:
from ...utils.dummy_torch_and_transformers_objects import (
StableDiffusionDepth2ImgPipeline,
StableDiffusionDiffEditPipeline,
StableDiffusionPix2PixZeroPipeline,
)
else:
from .pipeline_stable_diffusion_depth2img import StableDiffusionDepth2ImgPipeline
from .pipeline_stable_diffusion_diffedit import StableDiffusionDiffEditPipeline
from .pipeline_stable_diffusion_pix2pix_zero import StableDiffusionPix2PixZeroPipeline

1530
src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_diffedit.py Normal file
View File

File diff suppressed because it is too large Load Diff

15
src/diffusers/utils/dummy_torch_and_transformers_objects.py
View File

@@ -242,6 +242,21 @@ class StableDiffusionDepth2ImgPipeline(metaclass=DummyObject):
requires_backends(cls, ["torch", "transformers"])
class StableDiffusionDiffEditPipeline(metaclass=DummyObject):
_backends = ["torch", "transformers"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["torch", "transformers"])
@classmethod
def from_config(cls, *args, **kwargs):
requires_backends(cls, ["torch", "transformers"])
@classmethod
def from_pretrained(cls, *args, **kwargs):
requires_backends(cls, ["torch", "transformers"])
class StableDiffusionImageVariationPipeline(metaclass=DummyObject):
_backends = ["torch", "transformers"]

315
tests/pipelines/stable_diffusion_2/test_stable_diffusion_diffedit.py Normal file
View File

@@ -0,0 +1,315 @@
# coding=utf-8
# Copyright 2023 HuggingFace Inc.
#
# 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.
import gc
import random
import tempfile
import unittest
import numpy as np
import torch
from PIL import Image
from transformers import CLIPTextConfig, CLIPTextModel, CLIPTokenizer
from diffusers import (
AutoencoderKL,
DDIMInverseScheduler,
DDIMScheduler,
StableDiffusionDiffEditPipeline,
UNet2DConditionModel,
)
from diffusers.utils import load_image, slow
from diffusers.utils.testing_utils import floats_tensor, require_torch_gpu, torch_device
from ..pipeline_params import TEXT_GUIDED_IMAGE_INPAINTING_BATCH_PARAMS, TEXT_GUIDED_IMAGE_INPAINTING_PARAMS
from ..test_pipelines_common import PipelineTesterMixin
torch.backends.cuda.matmul.allow_tf32 = False
class StableDiffusionDiffEditPipelineFastTests(PipelineTesterMixin, unittest.TestCase):
pipeline_class = StableDiffusionDiffEditPipeline
params = TEXT_GUIDED_IMAGE_INPAINTING_PARAMS - {"height", "width", "image"} | {"image_latents"}
batch_params = TEXT_GUIDED_IMAGE_INPAINTING_BATCH_PARAMS - {"image"} | {"image_latents"}
def get_dummy_components(self):
torch.manual_seed(0)
unet = UNet2DConditionModel(
block_out_channels=(32, 64),
layers_per_block=2,
sample_size=32,
in_channels=4,
out_channels=4,
down_block_types=("DownBlock2D", "CrossAttnDownBlock2D"),
up_block_types=("CrossAttnUpBlock2D", "UpBlock2D"),
cross_attention_dim=32,
# SD2-specific config below
attention_head_dim=(2, 4),
use_linear_projection=True,
)
scheduler = DDIMScheduler(
beta_start=0.00085,
beta_end=0.012,
beta_schedule="scaled_linear",
clip_sample=False,
set_alpha_to_one=False,
)
inverse_scheduler = DDIMInverseScheduler(
beta_start=0.00085,
beta_end=0.012,
beta_schedule="scaled_linear",
clip_sample=False,
set_alpha_to_zero=False,
)
torch.manual_seed(0)
vae = AutoencoderKL(
block_out_channels=[32, 64],
in_channels=3,
out_channels=3,
down_block_types=["DownEncoderBlock2D", "DownEncoderBlock2D"],
up_block_types=["UpDecoderBlock2D", "UpDecoderBlock2D"],
latent_channels=4,
sample_size=128,
)
torch.manual_seed(0)
text_encoder_config = CLIPTextConfig(
bos_token_id=0,
eos_token_id=2,
hidden_size=32,
intermediate_size=37,
layer_norm_eps=1e-05,
num_attention_heads=4,
num_hidden_layers=5,
pad_token_id=1,
vocab_size=1000,
# SD2-specific config below
hidden_act="gelu",
projection_dim=512,
)
text_encoder = CLIPTextModel(text_encoder_config)
tokenizer = CLIPTokenizer.from_pretrained("hf-internal-testing/tiny-random-clip")
components = {
"unet": unet,
"scheduler": scheduler,
"inverse_scheduler": inverse_scheduler,
"vae": vae,
"text_encoder": text_encoder,
"tokenizer": tokenizer,
"safety_checker": None,
"feature_extractor": None,
}
return components
def get_dummy_inputs(self, device, seed=0):
mask = floats_tensor((1, 16, 16), rng=random.Random(seed)).to(device)
latents = floats_tensor((1, 2, 4, 16, 16), rng=random.Random(seed)).to(device)
if str(device).startswith("mps"):
generator = torch.manual_seed(seed)
else:
generator = torch.Generator(device=device).manual_seed(seed)
inputs = {
"prompt": "a dog and a newt",
"mask_image": mask,
"image_latents": latents,
"generator": generator,
"num_inference_steps": 2,
"inpaint_strength": 1.0,
"guidance_scale": 6.0,
"output_type": "numpy",
}
return inputs
def get_dummy_mask_inputs(self, device, seed=0):
image = floats_tensor((1, 3, 32, 32), rng=random.Random(seed)).to(device)
image = image.cpu().permute(0, 2, 3, 1)[0]
image = Image.fromarray(np.uint8(image)).convert("RGB")
if str(device).startswith("mps"):
generator = torch.manual_seed(seed)
else:
generator = torch.Generator(device=device).manual_seed(seed)
inputs = {
"image": image,
"source_prompt": "a cat and a frog",
"target_prompt": "a dog and a newt",
"generator": generator,
"num_inference_steps": 2,
"num_maps_per_mask": 2,
"mask_encode_strength": 1.0,
"guidance_scale": 6.0,
"output_type": "numpy",
}
return inputs
def get_dummy_inversion_inputs(self, device, seed=0):
image = floats_tensor((1, 3, 32, 32), rng=random.Random(seed)).to(device)
image = image.cpu().permute(0, 2, 3, 1)[0]
image = Image.fromarray(np.uint8(image)).convert("RGB")
if str(device).startswith("mps"):
generator = torch.manual_seed(seed)
else:
generator = torch.Generator(device=device).manual_seed(seed)
inputs = {
"image": image,
"prompt": "a cat and a frog",
"generator": generator,
"num_inference_steps": 2,
"inpaint_strength": 1.0,
"guidance_scale": 6.0,
"decode_latents": True,
"output_type": "numpy",
}
return inputs
def test_save_load_optional_components(self):
if not hasattr(self.pipeline_class, "_optional_components"):
return
components = self.get_dummy_components()
pipe = self.pipeline_class(**components)
pipe.to(torch_device)
pipe.set_progress_bar_config(disable=None)
# set all optional components to None and update pipeline config accordingly
for optional_component in pipe._optional_components:
setattr(pipe, optional_component, None)
pipe.register_modules(**{optional_component: None for optional_component in pipe._optional_components})
inputs = self.get_dummy_inputs(torch_device)
output = pipe(**inputs)[0]
with tempfile.TemporaryDirectory() as tmpdir:
pipe.save_pretrained(tmpdir)
pipe_loaded = self.pipeline_class.from_pretrained(tmpdir)
pipe_loaded.to(torch_device)
pipe_loaded.set_progress_bar_config(disable=None)
for optional_component in pipe._optional_components:
self.assertTrue(
getattr(pipe_loaded, optional_component) is None,
f"`{optional_component}` did not stay set to None after loading.",
)
inputs = self.get_dummy_inputs(torch_device)
output_loaded = pipe_loaded(**inputs)[0]
max_diff = np.abs(output - output_loaded).max()
self.assertLess(max_diff, 1e-4)
def test_mask(self):
device = "cpu"
components = self.get_dummy_components()
pipe = self.pipeline_class(**components)
pipe.to(device)
pipe.set_progress_bar_config(disable=None)
inputs = self.get_dummy_mask_inputs(device)
mask = pipe.generate_mask(**inputs)
mask_slice = mask[0, -3:, -3:]
self.assertEqual(mask.shape, (1, 16, 16))
expected_slice = np.array([0] * 9)
max_diff = np.abs(mask_slice.flatten() - expected_slice).max()
self.assertLessEqual(max_diff, 1e-3)
self.assertEqual(mask[0, -3, -4], 0)
def test_inversion(self):
device = "cpu"
components = self.get_dummy_components()
pipe = self.pipeline_class(**components)
pipe.to(device)
pipe.set_progress_bar_config(disable=None)
inputs = self.get_dummy_inversion_inputs(device)
image = pipe.invert(**inputs).images
image_slice = image[0, -1, -3:, -3:]
self.assertEqual(image.shape, (2, 32, 32, 3))
expected_slice = np.array(
[0.5150, 0.5134, 0.5043, 0.5376, 0.4694, 0.51050, 0.5015, 0.4407, 0.4799],
)
max_diff = np.abs(image_slice.flatten() - expected_slice).max()
self.assertLessEqual(max_diff, 1e-3)
@require_torch_gpu
@slow
class StableDiffusionDiffEditPipelineIntegrationTests(unittest.TestCase):
def tearDown(self):
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
@classmethod
def setUpClass(cls):
raw_image = load_image(
"https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/diffedit/fruit.png"
)
raw_image = raw_image.convert("RGB").resize((768, 768))
cls.raw_image = raw_image
def test_stable_diffusion_diffedit_full(self):
generator = torch.manual_seed(0)
pipe = StableDiffusionDiffEditPipeline.from_pretrained(
"stabilityai/stable-diffusion-2-1", safety_checker=None, torch_dtype=torch.float16
)
pipe.scheduler = DDIMScheduler.from_config(pipe.scheduler.config)
pipe.inverse_scheduler = DDIMInverseScheduler.from_config(pipe.scheduler.config)
pipe.enable_model_cpu_offload()
pipe.set_progress_bar_config(disable=None)
source_prompt = "a bowl of fruit"
target_prompt = "a bowl of pears"
mask_image = pipe.generate_mask(
image=self.raw_image,
source_prompt=source_prompt,
target_prompt=target_prompt,
generator=generator,
)
inv_latents = pipe.invert(
prompt=source_prompt, image=self.raw_image, inpaint_strength=0.7, generator=generator
).latents
image = pipe(
prompt=target_prompt,
mask_image=mask_image,
image_latents=inv_latents,
generator=generator,
negative_prompt=source_prompt,
inpaint_strength=0.7,
output_type="numpy",
).images[0]
expected_image = (
np.array(
load_image(
"https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main"
"/diffedit/pears.png"
).resize((768, 768))
)
/ 255
)
assert np.abs((expected_image - image).max()) < 5e-1