Merge branch 'christopher-beckham/fix_flux_controlnet_modes' of github.com:christopher-beckham/diffusers into christopher-beckham/fix_flux_controlnet_modes

Author: christopher-beckham

docs/source/en/api/pipelines/animatediff.md

@@ -914,6 +914,89 @@ export_to_gif(frames, "animatelcm-motion-lora.gif")
     </tr>
 </table>
 
+## Using FreeNoise
+
+[FreeNoise: Tuning-Free Longer Video Diffusion via Noise Rescheduling](https://arxiv.org/abs/2310.15169) by Haonan Qiu, Menghan Xia, Yong Zhang, Yingqing He, Xintao Wang, Ying Shan, Ziwei Liu.
+
+FreeNoise is a sampling mechanism that can generate longer videos with short-video generation models by employing noise-rescheduling, temporal attention over sliding windows, and weighted averaging of latent frames. It also can be used with multiple prompts to allow for interpolated video generations. More details are available in the paper.
+
+The currently supported AnimateDiff pipelines that can be used with FreeNoise are:
+- [`AnimateDiffPipeline`]
+- [`AnimateDiffControlNetPipeline`]
+- [`AnimateDiffVideoToVideoPipeline`]
+- [`AnimateDiffVideoToVideoControlNetPipeline`]
+
+In order to use FreeNoise, a single line needs to be added to the inference code after loading your pipelines.
+
+```diff
++ pipe.enable_free_noise()
+```
+
+After this, either a single prompt could be used, or multiple prompts can be passed as a dictionary of integer-string pairs. The integer keys of the dictionary correspond to the frame index at which the influence of that prompt would be maximum. Each frame index should map to a single string prompt. The prompts for intermediate frame indices, that are not passed in the dictionary, are created by interpolating between the frame prompts that are passed. By default, simple linear interpolation is used. However, you can customize this behaviour with a callback to the `prompt_interpolation_callback` parameter when enabling FreeNoise.
+
+Full example:
+
+```python
+import torch
+from diffusers import AutoencoderKL, AnimateDiffPipeline, LCMScheduler, MotionAdapter
+from diffusers.utils import export_to_video, load_image
+
+# Load pipeline
+dtype = torch.float16
+motion_adapter = MotionAdapter.from_pretrained("wangfuyun/AnimateLCM", torch_dtype=dtype)
+vae = AutoencoderKL.from_pretrained("stabilityai/sd-vae-ft-mse", torch_dtype=dtype)
+
+pipe = AnimateDiffPipeline.from_pretrained("emilianJR/epiCRealism", motion_adapter=motion_adapter, vae=vae, torch_dtype=dtype)
+pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config, beta_schedule="linear")
+
+pipe.load_lora_weights(
+    "wangfuyun/AnimateLCM", weight_name="AnimateLCM_sd15_t2v_lora.safetensors", adapter_name="lcm_lora"
+)
+pipe.set_adapters(["lcm_lora"], [0.8])
+
+# Enable FreeNoise for long prompt generation
+pipe.enable_free_noise(context_length=16, context_stride=4)
+pipe.to("cuda")
+
+# Can be a single prompt, or a dictionary with frame timesteps
+prompt = {
+    0: "A caterpillar on a leaf, high quality, photorealistic",
+    40: "A caterpillar transforming into a cocoon, on a leaf, near flowers, photorealistic",
+    80: "A cocoon on a leaf, flowers in the backgrond, photorealistic",
+    120: "A cocoon maturing and a butterfly being born, flowers and leaves visible in the background, photorealistic",
+    160: "A beautiful butterfly, vibrant colors, sitting on a leaf, flowers in the background, photorealistic",
+    200: "A beautiful butterfly, flying away in a forest, photorealistic",
+    240: "A cyberpunk butterfly, neon lights, glowing",
+}
+negative_prompt = "bad quality, worst quality, jpeg artifacts"
+
+# Run inference
+output = pipe(
+    prompt=prompt,
+    negative_prompt=negative_prompt,
+    num_frames=256,
+    guidance_scale=2.5,
+    num_inference_steps=10,
+    generator=torch.Generator("cpu").manual_seed(0),
+)
+
+# Save video
+frames = output.frames[0]
+export_to_video(frames, "output.mp4", fps=16)
+```
+
+### FreeNoise memory savings
+
+Since FreeNoise processes multiple frames together, there are parts in the modeling where the memory required exceeds that available on normal consumer GPUs. The main memory bottlenecks that we identified are spatial and temporal attention blocks, upsampling and downsampling blocks, resnet blocks and feed-forward layers. Since most of these blocks operate effectively only on the channel/embedding dimension, one can perform chunked inference across the batch dimensions. The batch dimension in AnimateDiff are either spatial (`[B x F, H x W, C]`) or temporal (`B x H x W, F, C`) in nature (note that it may seem counter-intuitive, but the batch dimension here are correct, because spatial blocks process across the `B x F` dimension while the temporal blocks process across the `B x H x W` dimension). We introduce a `SplitInferenceModule` that makes it easier to chunk across any dimension and perform inference. This saves a lot of memory but comes at the cost of requiring more time for inference.
+
+```diff
+# Load pipeline and adapters
+# ...
++ pipe.enable_free_noise_split_inference()
++ pipe.unet.enable_forward_chunking(16)
+```
+
+The call to `pipe.enable_free_noise_split_inference` method accepts two parameters: `spatial_split_size` (defaults to `256`) and `temporal_split_size` (defaults to `16`). These can be configured based on how much VRAM you have available. A lower split size results in lower memory usage but slower inference, whereas a larger split size results in faster inference at the cost of more memory.
 
 ## Using `from_single_file` with the MotionAdapter
 

examples/controlnet/README_sd3.md

@@ -0,0 +1,152 @@
+# ControlNet training example for Stable Diffusion 3 (SD3)
+
+The `train_controlnet_sd3.py` script shows how to implement the ControlNet training procedure and adapt it for [Stable Diffusion 3](https://arxiv.org/abs/2403.03206).
+
+## Running locally with PyTorch
+
+### Installing the dependencies
+
+Before running the scripts, make sure to install the library's training dependencies:
+
+**Important**
+
+To make sure you can successfully run the latest versions of the example scripts, we highly recommend **installing from source** and keeping the install up to date as we update the example scripts frequently and install some example-specific requirements. To do this, execute the following steps in a new virtual environment:
+
+```bash
+git clone https://github.com/huggingface/diffusers
+cd diffusers
+pip install -e .
+```
+
+Then cd in the `examples/controlnet` folder and run
+```bash
+pip install -r requirements_sd3.txt
+```
+
+And initialize an [🤗Accelerate](https://github.com/huggingface/accelerate/) environment with:
+
+```bash
+accelerate config
+```
+
+Or for a default accelerate configuration without answering questions about your environment
+
+```bash
+accelerate config default
+```
+
+Or if your environment doesn't support an interactive shell (e.g., a notebook)
+
+```python
+from accelerate.utils import write_basic_config
+write_basic_config()
+```
+
+When running `accelerate config`, if we specify torch compile mode to True there can be dramatic speedups.
+
+## Circle filling dataset
+
+The original dataset is hosted in the [ControlNet repo](https://huggingface.co/lllyasviel/ControlNet/blob/main/training/fill50k.zip). We re-uploaded it to be compatible with `datasets` [here](https://huggingface.co/datasets/fusing/fill50k). Note that `datasets` handles dataloading within the training script.
+Please download the dataset and unzip it in the directory `fill50k` in the `examples/controlnet` folder.
+
+## Training
+
+First download the SD3 model from [Hugging Face Hub](https://huggingface.co/stabilityai/stable-diffusion-3-medium). We will use it as a base model for the ControlNet training.
+> [!NOTE]
+> As the model is gated, before using it with diffusers you first need to go to the [Stable Diffusion 3 Medium Hugging Face page](https://huggingface.co/stabilityai/stable-diffusion-3-medium-diffusers), fill in the form and accept the gate. Once you are in, you need to log in so that your system knows you’ve accepted the gate. Use the command below to log in:
+
+```bash
+huggingface-cli login
+```
+
+This will also allow us to push the trained model parameters to the Hugging Face Hub platform.
+
+
+Our training examples use two test conditioning images. They can be downloaded by running
+
+```sh
+wget https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet_training/conditioning_image_1.png
+
+wget https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet_training/conditioning_image_2.png
+```
+
+Then run the following commands to train a ControlNet model.
+
+```bash
+export MODEL_DIR="stabilityai/stable-diffusion-3-medium-diffusers"
+export OUTPUT_DIR="sd3-controlnet-out"
+
+accelerate launch train_controlnet_sd3.py \
+    --pretrained_model_name_or_path=$MODEL_DIR \
+    --output_dir=$OUTPUT_DIR \
+    --train_data_dir="fill50k" \
+    --resolution=1024 \
+    --learning_rate=1e-5 \
+    --max_train_steps=15000 \
+    --validation_image "./conditioning_image_1.png" "./conditioning_image_2.png" \
+    --validation_prompt "red circle with blue background" "cyan circle with brown floral background" \
+    --validation_steps=100 \
+    --train_batch_size=1 \
+    --gradient_accumulation_steps=4
+```
+
+To better track our training experiments, we're using flags `validation_image`, `validation_prompt`, and `validation_steps` to allow the script to do a few validation inference runs. This allows us to qualitatively check if the training is progressing as expected.
+
+Our experiments were conducted on a single 40GB A100 GPU.
+
+### Inference
+
+Once training is done, we can perform inference like so:
+
+```python
+from diffusers import StableDiffusion3ControlNetPipeline, SD3ControlNetModel
+from diffusers.utils import load_image
+import torch
+
+base_model_path = "stabilityai/stable-diffusion-3-medium-diffusers"
+controlnet_path = "sd3-controlnet-out/checkpoint-6500/controlnet"
+
+controlnet = SD3ControlNetModel.from_pretrained(controlnet_path, torch_dtype=torch.float16)
+pipe = StableDiffusion3ControlNetPipeline.from_pretrained(
+    base_model_path, controlnet=controlnet
+)
+pipe.to("cuda", torch.float16)
+
+
+control_image = load_image("./conditioning_image_1.png").resize((1024, 1024))
+prompt = "pale golden rod circle with old lace background"
+
+# generate image
+generator = torch.manual_seed(0)
+image = pipe(
+    prompt, num_inference_steps=20, generator=generator, control_image=control_image
+).images[0]
+image.save("./output.png")
+```
+
+## Notes
+
+### GPU usage
+
+SD3 is a large model and requires a lot of GPU memory. 
+We recommend using one GPU with at least 80GB of memory.
+Make sure to use the right GPU when configuring the [accelerator](https://huggingface.co/docs/transformers/en/accelerate).
+
+
+## Example results
+
+#### After 500 steps with batch size 8
+
+| |  |
+|-------------------|:-------------------------:|
+|| pale golden rod circle with old lace background |
+ ![conditioning image](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet_training/conditioning_image_1.png) | ![pale golden rod circle with old lace background](https://huggingface.co/datasets/DavyMorgan/sd3-controlnet-results/resolve/main/step-500.png) |
+
+
+#### After 6500 steps with batch size 8:
+
+| |  |
+|-------------------|:-------------------------:|
+|| pale golden rod circle with old lace background |
+ ![conditioning image](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet_training/conditioning_image_1.png) | ![pale golden rod circle with old lace background](https://huggingface.co/datasets/DavyMorgan/sd3-controlnet-results/resolve/main/step-6500.png) |
+

examples/controlnet/requirements_sd3.txt

@@ -0,0 +1,8 @@
+accelerate>=0.16.0
+torchvision
+transformers>=4.25.1
+ftfy
+tensorboard
+Jinja2
+datasets
+wandb

examples/controlnet/test_controlnet.py

@@ -115,3 +115,24 @@ def test_controlnet_sdxl(self):
             run_command(self._launch_args + test_args)
 
             self.assertTrue(os.path.isfile(os.path.join(tmpdir, "diffusion_pytorch_model.safetensors")))
+
+
+class ControlNetSD3(ExamplesTestsAccelerate):
+    def test_controlnet_sd3(self):
+        with tempfile.TemporaryDirectory() as tmpdir:
+            test_args = f"""
+            examples/controlnet/train_controlnet_sd3.py
+            --pretrained_model_name_or_path=DavyMorgan/tiny-sd3-pipe
+            --dataset_name=hf-internal-testing/fill10
+            --output_dir={tmpdir}
+            --resolution=64
+            --train_batch_size=1
+            --gradient_accumulation_steps=1
+            --controlnet_model_name_or_path=DavyMorgan/tiny-controlnet-sd3
+            --max_train_steps=4
+            --checkpointing_steps=2
+            """.split()
+
+            run_command(self._launch_args + test_args)
+
+            self.assertTrue(os.path.isfile(os.path.join(tmpdir, "diffusion_pytorch_model.safetensors")))