Mochi 1 Preview (original) (raw)

Only a research preview of the model weights is available at the moment.

Mochi 1 is a video generation model by Genmo with a strong focus on prompt adherence and motion quality. The model features a 10B parameter Asmmetric Diffusion Transformer (AsymmDiT) architecture, and uses non-square QKV and output projection layers to reduce inference memory requirements. A single T5-XXL model is used to encode prompts.

Mochi 1 preview is an open state-of-the-art video generation model with high-fidelity motion and strong prompt adherence in preliminary evaluation. This model dramatically closes the gap between closed and open video generation systems. The model is released under a permissive Apache 2.0 license.

Make sure to check out the Schedulers guide to learn how to explore the tradeoff between scheduler speed and quality, and see the reuse components across pipelines section to learn how to efficiently load the same components into multiple pipelines.

Quantization

Quantization helps reduce the memory requirements of very large models by storing model weights in a lower precision data type. However, quantization may have varying impact on video quality depending on the video model.

Refer to the Quantization overview to learn more about supported quantization backends and selecting a quantization backend that supports your use case. The example below demonstrates how to load a quantized MochiPipeline for inference with bitsandbytes.

import torch from diffusers import BitsAndBytesConfig as DiffusersBitsAndBytesConfig, MochiTransformer3DModel, MochiPipeline from diffusers.utils import export_to_video from transformers import BitsAndBytesConfig as BitsAndBytesConfig, T5EncoderModel

quant_config = BitsAndBytesConfig(load_in_8bit=True) text_encoder_8bit = T5EncoderModel.from_pretrained( "genmo/mochi-1-preview", subfolder="text_encoder", quantization_config=quant_config, torch_dtype=torch.float16, )

quant_config = DiffusersBitsAndBytesConfig(load_in_8bit=True) transformer_8bit = MochiTransformer3DModel.from_pretrained( "genmo/mochi-1-preview", subfolder="transformer", quantization_config=quant_config, torch_dtype=torch.float16, )

pipeline = MochiPipeline.from_pretrained( "genmo/mochi-1-preview", text_encoder=text_encoder_8bit, transformer=transformer_8bit, torch_dtype=torch.float16, device_map="balanced", )

video = pipeline( "Close-up of a cats eye, with the galaxy reflected in the cats eye. Ultra high resolution 4k.", num_inference_steps=28, guidance_scale=3.5 ).frames[0] export_to_video(video, "cat.mp4")

Generating videos with Mochi-1 Preview

The following example will download the full precision mochi-1-preview weights and produce the highest quality results but will require at least 42GB VRAM to run.

import torch from diffusers import MochiPipeline from diffusers.utils import export_to_video

pipe = MochiPipeline.from_pretrained("genmo/mochi-1-preview")

pipe.enable_model_cpu_offload() pipe.enable_vae_tiling()

prompt = "Close-up of a chameleon's eye, with its scaly skin changing color. Ultra high resolution 4k."

with torch.autocast("cuda", torch.bfloat16, cache_enabled=False): frames = pipe(prompt, num_frames=85).frames[0]

export_to_video(frames, "mochi.mp4", fps=30)

Using a lower precision variant to save memory

The following example will use the bfloat16 variant of the model and requires 22GB VRAM to run. There is a slight drop in the quality of the generated video as a result.

import torch from diffusers import MochiPipeline from diffusers.utils import export_to_video

pipe = MochiPipeline.from_pretrained("genmo/mochi-1-preview", variant="bf16", torch_dtype=torch.bfloat16)

pipe.enable_model_cpu_offload() pipe.enable_vae_tiling()

prompt = "Close-up of a chameleon's eye, with its scaly skin changing color. Ultra high resolution 4k." frames = pipe(prompt, num_frames=85).frames[0]

export_to_video(frames, "mochi.mp4", fps=30)

Reproducing the results from the Genmo Mochi repo

The Genmo Mochi implementation uses different precision values for each stage in the inference process. The text encoder and VAE use torch.float32, while the DiT uses torch.bfloat16 with the attention kernel set to EFFICIENT_ATTENTION. Diffusers pipelines currently do not support setting different dtypes for different stages of the pipeline. In order to run inference in the same way as the original implementation, please refer to the following example.

The original Mochi implementation zeros out empty prompts. However, enabling this option and placing the entire pipeline under autocast can lead to numerical overflows with the T5 text encoder.

When enabling force_zeros_for_empty_prompt, it is recommended to run the text encoding step outside the autocast context in full precision.

Decoding the latents in full precision is very memory intensive. You will need at least 70GB VRAM to generate the 163 frames in this example. To reduce memory, either reduce the number of frames or run the decoding step in `torch.bfloat16`.

import torch from torch.nn.attention import SDPBackend, sdpa_kernel

from diffusers import MochiPipeline from diffusers.utils import export_to_video from diffusers.video_processor import VideoProcessor

pipe = MochiPipeline.from_pretrained("genmo/mochi-1-preview", force_zeros_for_empty_prompt=True) pipe.enable_vae_tiling() pipe.enable_model_cpu_offload()

prompt = "An aerial shot of a parade of elephants walking across the African savannah. The camera showcases the herd and the surrounding landscape."

with torch.no_grad(): prompt_embeds, prompt_attention_mask, negative_prompt_embeds, negative_prompt_attention_mask = ( pipe.encode_prompt(prompt=prompt) )

with torch.autocast("cuda", torch.bfloat16): with sdpa_kernel(SDPBackend.EFFICIENT_ATTENTION): frames = pipe( prompt_embeds=prompt_embeds, prompt_attention_mask=prompt_attention_mask, negative_prompt_embeds=negative_prompt_embeds, negative_prompt_attention_mask=negative_prompt_attention_mask, guidance_scale=4.5, num_inference_steps=64, height=480, width=848, num_frames=163, generator=torch.Generator("cuda").manual_seed(0), output_type="latent", return_dict=False, )[0]

video_processor = VideoProcessor(vae_scale_factor=8) has_latents_mean = hasattr(pipe.vae.config, "latents_mean") and pipe.vae.config.latents_mean is not None has_latents_std = hasattr(pipe.vae.config, "latents_std") and pipe.vae.config.latents_std is not None if has_latents_mean and has_latents_std: latents_mean = ( torch.tensor(pipe.vae.config.latents_mean).view(1, 12, 1, 1, 1).to(frames.device, frames.dtype) ) latents_std = ( torch.tensor(pipe.vae.config.latents_std).view(1, 12, 1, 1, 1).to(frames.device, frames.dtype) ) frames = frames * latents_std / pipe.vae.config.scaling_factor + latents_mean else: frames = frames / pipe.vae.config.scaling_factor

with torch.no_grad(): video = pipe.vae.decode(frames.to(pipe.vae.dtype), return_dict=False)[0]

video = video_processor.postprocess_video(video)[0] export_to_video(video, "mochi.mp4", fps=30)

Running inference with multiple GPUs

It is possible to split the large Mochi transformer across multiple GPUs using the device_map and max_memory options in from_pretrained. In the following example we split the model across two GPUs, each with 24GB of VRAM.

import torch from diffusers import MochiPipeline, MochiTransformer3DModel from diffusers.utils import export_to_video

model_id = "genmo/mochi-1-preview" transformer = MochiTransformer3DModel.from_pretrained( model_id, subfolder="transformer", device_map="auto", max_memory={0: "24GB", 1: "24GB"} )

pipe = MochiPipeline.from_pretrained(model_id, transformer=transformer) pipe.enable_model_cpu_offload() pipe.enable_vae_tiling()

with torch.autocast(device_type="cuda", dtype=torch.bfloat16, cache_enabled=False): frames = pipe( prompt="Close-up of a chameleon's eye, with its scaly skin changing color. Ultra high resolution 4k.", negative_prompt="", height=480, width=848, num_frames=85, num_inference_steps=50, guidance_scale=4.5, num_videos_per_prompt=1, generator=torch.Generator(device="cuda").manual_seed(0), max_sequence_length=256, output_type="pil", ).frames[0]

export_to_video(frames, "output.mp4", fps=30)

Using single file loading with the Mochi Transformer

You can use from_single_file to load the Mochi transformer in its original format.

Diffusers currently doesn't support using the FP8 scaled versions of the Mochi single file checkpoints.

import torch from diffusers import MochiPipeline, MochiTransformer3DModel from diffusers.utils import export_to_video

model_id = "genmo/mochi-1-preview"

ckpt_path = "https://huggingface.co/Comfy-Org/mochi_preview_repackaged/blob/main/split_files/diffusion_models/mochi_preview_bf16.safetensors"

transformer = MochiTransformer3DModel.from_pretrained(ckpt_path, torch_dtype=torch.bfloat16)

pipe = MochiPipeline.from_pretrained(model_id, transformer=transformer) pipe.enable_model_cpu_offload() pipe.enable_vae_tiling()

with torch.autocast(device_type="cuda", dtype=torch.bfloat16, cache_enabled=False): frames = pipe( prompt="Close-up of a chameleon's eye, with its scaly skin changing color. Ultra high resolution 4k.", negative_prompt="", height=480, width=848, num_frames=85, num_inference_steps=50, guidance_scale=4.5, num_videos_per_prompt=1, generator=torch.Generator(device="cuda").manual_seed(0), max_sequence_length=256, output_type="pil", ).frames[0]

export_to_video(frames, "output.mp4", fps=30)

MochiPipeline

class diffusers.MochiPipeline

< source >

( scheduler: FlowMatchEulerDiscreteScheduler vae: AutoencoderKLMochi text_encoder: T5EncoderModel tokenizer: T5TokenizerFast transformer: MochiTransformer3DModel force_zeros_for_empty_prompt: bool = False )

Parameters

• transformer (MochiTransformer3DModel) — Conditional Transformer architecture to denoise the encoded video latents.
• scheduler (FlowMatchEulerDiscreteScheduler) — A scheduler to be used in combination with transformer to denoise the encoded image latents.
• vae (AutoencoderKLMochi) — Variational Auto-Encoder (VAE) Model to encode and decode videos to and from latent representations.
• text_encoder (T5EncoderModel) —T5, specifically the google/t5-v1_1-xxl variant.
• tokenizer (CLIPTokenizer) — Tokenizer of classCLIPTokenizer.
• tokenizer (T5TokenizerFast) — Second Tokenizer of classT5TokenizerFast.

The mochi pipeline for text-to-video generation.

Reference: https://github.com/genmoai/models

__call__

< source >

( prompt: typing.Union[str, typing.List[str]] = None negative_prompt: typing.Union[str, typing.List[str], NoneType] = None height: typing.Optional[int] = None width: typing.Optional[int] = None num_frames: int = 19 num_inference_steps: int = 64 timesteps: typing.List[int] = None guidance_scale: float = 4.5 num_videos_per_prompt: typing.Optional[int] = 1 generator: typing.Union[torch._C.Generator, typing.List[torch._C.Generator], NoneType] = None latents: typing.Optional[torch.Tensor] = None prompt_embeds: typing.Optional[torch.Tensor] = None prompt_attention_mask: typing.Optional[torch.Tensor] = None negative_prompt_embeds: typing.Optional[torch.Tensor] = None negative_prompt_attention_mask: typing.Optional[torch.Tensor] = None output_type: typing.Optional[str] = 'pil' return_dict: bool = True attention_kwargs: typing.Optional[typing.Dict[str, typing.Any]] = None callback_on_step_end: typing.Optional[typing.Callable[[int, int, typing.Dict], NoneType]] = None callback_on_step_end_tensor_inputs: typing.List[str] = ['latents'] max_sequence_length: int = 256 ) → ~pipelines.mochi.MochiPipelineOutput or tuple

Parameters

• prompt (str or List[str], optional) — The prompt or prompts to guide the image generation. If not defined, one has to pass prompt_embeds. instead.
• height (int, optional, defaults to self.default_height) — The height in pixels of the generated image. This is set to 480 by default for the best results.
• width (int, optional, defaults to self.default_width) — The width in pixels of the generated image. This is set to 848 by default for the best results.
• num_frames (int, defaults to 19) — The number of video frames to generate
• num_inference_steps (int, optional, defaults to 50) — The number of denoising steps. More denoising steps usually lead to a higher quality image at the expense of slower inference.
• timesteps (List[int], optional) — Custom timesteps to use for the denoising process with schedulers which support a timesteps argument in their set_timesteps method. If not defined, the default behavior when num_inference_steps is passed will be used. Must be in descending order.
• guidance_scale (float, defaults to 4.5) — Guidance scale as defined in Classifier-Free Diffusion Guidance.guidance_scale is defined as w of equation 2. of Imagen Paper. Guidance scale is enabled by setting guidance_scale > 1. Higher guidance scale encourages to generate images that are closely linked to the text prompt, usually at the expense of lower image quality.
• num_videos_per_prompt (int, optional, defaults to 1) — The number of videos to generate per prompt.
• generator (torch.Generator or List[torch.Generator], optional) — One or a list of torch generator(s)to make generation deterministic.
• latents (torch.Tensor, optional) — Pre-generated noisy latents, sampled from a Gaussian distribution, to be used as inputs for image generation. Can be used to tweak the same generation with different prompts. If not provided, a latents tensor will ge generated by sampling using the supplied random generator.
• prompt_embeds (torch.Tensor, optional) — Pre-generated text embeddings. Can be used to easily tweak text inputs, e.g. prompt weighting. If not provided, text embeddings will be generated from prompt input argument.
• prompt_attention_mask (torch.Tensor, optional) — Pre-generated attention mask for text embeddings.
• negative_prompt_embeds (torch.FloatTensor, optional) — Pre-generated negative text embeddings. For PixArt-Sigma this negative prompt should be "". If not provided, negative_prompt_embeds will be generated from negative_prompt input argument.
• negative_prompt_attention_mask (torch.FloatTensor, optional) — Pre-generated attention mask for negative text embeddings.
• output_type (str, optional, defaults to "pil") — The output format of the generate image. Choose betweenPIL: PIL.Image.Image or np.array.
• return_dict (bool, optional, defaults to True) — Whether or not to return a ~pipelines.mochi.MochiPipelineOutput instead of a plain tuple.
• attention_kwargs (dict, optional) — A kwargs dictionary that if specified is passed along to the AttentionProcessor as defined underself.processor indiffusers.models.attention_processor.
• callback_on_step_end (Callable, optional) — A function that calls at the end of each denoising steps during the inference. The function is called with the following arguments: callback_on_step_end(self: DiffusionPipeline, step: int, timestep: int, callback_kwargs: Dict). callback_kwargs will include a list of all tensors as specified bycallback_on_step_end_tensor_inputs.
• callback_on_step_end_tensor_inputs (List, optional) — The list of tensor inputs for the callback_on_step_end function. The tensors specified in the list will be passed as callback_kwargs argument. You will only be able to include variables listed in the._callback_tensor_inputs attribute of your pipeline class.
• max_sequence_length (int defaults to 256) — Maximum sequence length to use with the prompt.

Returns

~pipelines.mochi.MochiPipelineOutput or tuple

If return_dict is True, ~pipelines.mochi.MochiPipelineOutput is returned, otherwise a tupleis returned where the first element is a list with the generated images.

Function invoked when calling the pipeline for generation.

Examples:

import torch from diffusers import MochiPipeline from diffusers.utils import export_to_video

pipe = MochiPipeline.from_pretrained("genmo/mochi-1-preview", torch_dtype=torch.bfloat16) pipe.enable_model_cpu_offload() pipe.enable_vae_tiling() prompt = "Close-up of a chameleon's eye, with its scaly skin changing color. Ultra high resolution 4k." frames = pipe(prompt, num_inference_steps=28, guidance_scale=3.5).frames[0] export_to_video(frames, "mochi.mp4")

Disable sliced VAE decoding. If enable_vae_slicing was previously enabled, this method will go back to computing decoding in one step.

Disable tiled VAE decoding. If enable_vae_tiling was previously enabled, this method will go back to computing decoding in one step.

Enable sliced VAE decoding. When this option is enabled, the VAE will split the input tensor in slices to compute decoding in several steps. This is useful to save some memory and allow larger batch sizes.

Enable tiled VAE decoding. When this option is enabled, the VAE will split the input tensor into tiles to compute decoding and encoding in several steps. This is useful for saving a large amount of memory and to allow processing larger images.

encode_prompt

< source >

( prompt: typing.Union[str, typing.List[str]] negative_prompt: typing.Union[str, typing.List[str], NoneType] = None do_classifier_free_guidance: bool = True num_videos_per_prompt: int = 1 prompt_embeds: typing.Optional[torch.Tensor] = None negative_prompt_embeds: typing.Optional[torch.Tensor] = None prompt_attention_mask: typing.Optional[torch.Tensor] = None negative_prompt_attention_mask: typing.Optional[torch.Tensor] = None max_sequence_length: int = 256 device: typing.Optional[torch.device] = None dtype: typing.Optional[torch.dtype] = None )

Parameters

• prompt (str or List[str], optional) — prompt to be encoded
• negative_prompt (str or List[str], optional) — The prompt or prompts not to guide the image generation. If not defined, one has to passnegative_prompt_embeds instead. Ignored when not using guidance (i.e., ignored if guidance_scale is less than 1).
• do_classifier_free_guidance (bool, optional, defaults to True) — Whether to use classifier free guidance or not.
• num_videos_per_prompt (int, optional, defaults to 1) — Number of videos that should be generated per prompt. torch device to place the resulting embeddings on
• prompt_embeds (torch.Tensor, optional) — Pre-generated text embeddings. Can be used to easily tweak text inputs, e.g. prompt weighting. If not provided, text embeddings will be generated from prompt input argument.
• negative_prompt_embeds (torch.Tensor, optional) — Pre-generated negative text embeddings. Can be used to easily tweak text inputs, e.g. prompt weighting. If not provided, negative_prompt_embeds will be generated from negative_prompt input argument.
• device — (torch.device, optional): torch device
• dtype — (torch.dtype, optional): torch dtype

Encodes the prompt into text encoder hidden states.

MochiPipelineOutput

class diffusers.pipelines.mochi.pipeline_output.MochiPipelineOutput

< source >

( frames: Tensor )

Parameters

• frames (torch.Tensor, np.ndarray, or List[List[PIL.Image.Image]]) — List of video outputs - It can be a nested list of length batch_size, with each sub-list containing denoised PIL image sequences of length num_frames. It can also be a NumPy array or Torch tensor of shape(batch_size, num_frames, channels, height, width).

Output class for Mochi pipelines.

< > Update on GitHub