🧪 This pipeline is for research purposes only.
Text-to-video
ModelScope Text-to-Video Technical Report is by Jiuniu Wang, Hangjie Yuan, Dayou Chen, Yingya Zhang, Xiang Wang, Shiwei Zhang.
The abstract from the paper is:
This paper introduces ModelScopeT2V, a text-to-video synthesis model that evolves from a text-to-image synthesis model (i.e., Stable Diffusion). ModelScopeT2V incorporates spatio-temporal blocks to ensure consistent frame generation and smooth movement transitions. The model could adapt to varying frame numbers during training and inference, rendering it suitable for both image-text and video-text datasets. ModelScopeT2V brings together three components (i.e., VQGAN, a text encoder, and a denoising UNet), totally comprising 1.7 billion parameters, in which 0.5 billion parameters are dedicated to temporal capabilities. The model demonstrates superior performance over state-of-the-art methods across three evaluation metrics. The code and an online demo are available at https://modelscope.cn/models/damo/text-to-video-synthesis/summary.
You can find additional information about Text-to-Video on the project page, original codebase, and try it out in a demo. Official checkpoints can be found at damo-vilab and cerspense.
Usage example
text-to-video-ms-1.7b
Let’s start by generating a short video with the default length of 16 frames (2s at 8 fps):
import torch
from diffusers import DiffusionPipeline
from diffusers.utils import export_to_video
pipe = DiffusionPipeline.from_pretrained("damo-vilab/text-to-video-ms-1.7b", torch_dtype=torch.float16, variant="fp16")
pipe = pipe.to("cuda")
prompt = "Spiderman is surfing"
video_frames = pipe(prompt).frames[0]
video_path = export_to_video(video_frames)
video_pathDiffusers supports different optimization techniques to improve the latency and memory footprint of a pipeline. Since videos are often more memory-heavy than images, we can enable CPU offloading and VAE slicing to keep the memory footprint at bay.
Let’s generate a video of 8 seconds (64 frames) on the same GPU using CPU offloading and VAE slicing:
import torch
from diffusers import DiffusionPipeline
from diffusers.utils import export_to_video
pipe = DiffusionPipeline.from_pretrained("damo-vilab/text-to-video-ms-1.7b", torch_dtype=torch.float16, variant="fp16")
pipe.enable_model_cpu_offload()
# memory optimization
pipe.enable_vae_slicing()
prompt = "Darth Vader surfing a wave"
video_frames = pipe(prompt, num_frames=64).frames[0]
video_path = export_to_video(video_frames)
video_pathIt just takes 7 GBs of GPU memory to generate the 64 video frames using PyTorch 2.0, “fp16” precision and the techniques mentioned above.
We can also use a different scheduler easily, using the same method we’d use for Stable Diffusion:
import torch
from diffusers import DiffusionPipeline, DPMSolverMultistepScheduler
from diffusers.utils import export_to_video
pipe = DiffusionPipeline.from_pretrained("damo-vilab/text-to-video-ms-1.7b", torch_dtype=torch.float16, variant="fp16")
pipe.scheduler = DPMSolverMultistepScheduler.from_config(pipe.scheduler.config)
pipe.enable_model_cpu_offload()
prompt = "Spiderman is surfing"
video_frames = pipe(prompt, num_inference_steps=25).frames[0]
video_path = export_to_video(video_frames)
video_pathHere are some sample outputs:
 |  |
cerspense/zeroscope_v2_576w & cerspense/zeroscope_v2_XL
Zeroscope are watermark-free model and have been trained on specific sizes such as 576x320 and 1024x576.
One should first generate a video using the lower resolution checkpoint cerspense/zeroscope_v2_576w with TextToVideoSDPipeline,
which can then be upscaled using VideoToVideoSDPipeline and cerspense/zeroscope_v2_XL.
import torch
from diffusers import DiffusionPipeline, DPMSolverMultistepScheduler
from diffusers.utils import export_to_video
from PIL import Image
pipe = DiffusionPipeline.from_pretrained("cerspense/zeroscope_v2_576w", torch_dtype=torch.float16)
pipe.enable_model_cpu_offload()
# memory optimization
pipe.unet.enable_forward_chunking(chunk_size=1, dim=1)
pipe.enable_vae_slicing()
prompt = "Darth Vader surfing a wave"
video_frames = pipe(prompt, num_frames=24).frames[0]
video_path = export_to_video(video_frames)
video_pathNow the video can be upscaled:
pipe = DiffusionPipeline.from_pretrained("cerspense/zeroscope_v2_XL", torch_dtype=torch.float16)
pipe.scheduler = DPMSolverMultistepScheduler.from_config(pipe.scheduler.config)
pipe.enable_model_cpu_offload()
# memory optimization
pipe.unet.enable_forward_chunking(chunk_size=1, dim=1)
pipe.enable_vae_slicing()
video = [Image.fromarray(frame).resize((1024, 576)) for frame in video_frames]
video_frames = pipe(prompt, video=video, strength=0.6).frames[0]
video_path = export_to_video(video_frames)
video_pathHere are some sample outputs:
 |
Tips
Video generation is memory-intensive and one way to reduce your memory usage is to set enable_forward_chunking on the pipeline’s UNet so you don’t run the entire feedforward layer at once. Breaking it up into chunks in a loop is more efficient.
Check out the Text or image-to-video guide for more details about how certain parameters can affect video generation and how to optimize inference by reducing memory usage.
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.
TextToVideoSDPipeline
class diffusers.TextToVideoSDPipeline
< source >( vae: AutoencoderKL text_encoder: CLIPTextModel tokenizer: CLIPTokenizer unet: UNet3DConditionModel scheduler: KarrasDiffusionSchedulers )
Parameters
- vae (AutoencoderKL) — Variational Auto-Encoder (VAE) Model to encode and decode images to and from latent representations.
- text_encoder (
CLIPTextModel) — Frozen text-encoder (clip-vit-large-patch14). - tokenizer (
CLIPTokenizer) — A CLIPTokenizer to tokenize text. - unet (UNet3DConditionModel) — A UNet3DConditionModel to denoise the encoded video latents.
- scheduler (SchedulerMixin) —
A scheduler to be used in combination with
unetto denoise the encoded image latents. Can be one of DDIMScheduler, LMSDiscreteScheduler, or PNDMScheduler.
Pipeline for text-to-video generation.
This model inherits from DiffusionPipeline. Check the superclass documentation for the generic methods implemented for all pipelines (downloading, saving, running on a particular device, etc.).
The pipeline also inherits the following loading methods:
- load_textual_inversion() for loading textual inversion embeddings
- load_lora_weights() for loading LoRA weights
- save_lora_weights() for saving LoRA weights
__call__
< source >( prompt: Union = None height: Optional = None width: Optional = None num_frames: int = 16 num_inference_steps: int = 50 guidance_scale: float = 9.0 negative_prompt: Union = None eta: float = 0.0 generator: Union = None latents: Optional = None prompt_embeds: Optional = None negative_prompt_embeds: Optional = None output_type: Optional = 'np' return_dict: bool = True callback: Optional = None callback_steps: int = 1 cross_attention_kwargs: Optional = None clip_skip: Optional = None ) → TextToVideoSDPipelineOutput or tuple
Parameters
- prompt (
strorList[str], optional) — The prompt or prompts to guide image generation. If not defined, you need to passprompt_embeds. - height (
int, optional, defaults toself.unet.config.sample_size * self.vae_scale_factor) — The height in pixels of the generated video. - width (
int, optional, defaults toself.unet.config.sample_size * self.vae_scale_factor) — The width in pixels of the generated video. - num_frames (
int, optional, defaults to 16) — The number of video frames that are generated. Defaults to 16 frames which at 8 frames per seconds amounts to 2 seconds of video. - num_inference_steps (
int, optional, defaults to 50) — The number of denoising steps. More denoising steps usually lead to a higher quality videos at the expense of slower inference. - guidance_scale (
float, optional, defaults to 7.5) — A higher guidance scale value encourages the model to generate images closely linked to the textpromptat the expense of lower image quality. Guidance scale is enabled whenguidance_scale > 1. - negative_prompt (
strorList[str], optional) — The prompt or prompts to guide what to not include in image generation. If not defined, you need to passnegative_prompt_embedsinstead. Ignored when not using guidance (guidance_scale < 1). - num_images_per_prompt (
int, optional, defaults to 1) — The number of images to generate per prompt. - eta (
float, optional, defaults to 0.0) — Corresponds to parameter eta (η) from the DDIM paper. Only applies to the DDIMScheduler, and is ignored in other schedulers. - generator (
torch.GeneratororList[torch.Generator], optional) — Atorch.Generatorto make generation deterministic. - latents (
torch.Tensor, optional) — Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for video generation. Can be used to tweak the same generation with different prompts. If not provided, a latents tensor is generated by sampling using the supplied randomgenerator. Latents should be of shape(batch_size, num_channel, num_frames, height, width). - prompt_embeds (
torch.Tensor, optional) — Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not provided, text embeddings are generated from thepromptinput argument. - negative_prompt_embeds (
torch.Tensor, optional) — Pre-generated negative text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not provided,negative_prompt_embedsare generated from thenegative_promptinput argument. - output_type (
str, optional, defaults to"np") — The output format of the generated video. Choose betweentorch.Tensorornp.array. - return_dict (
bool, optional, defaults toTrue) — Whether or not to return a TextToVideoSDPipelineOutput instead of a plain tuple. - callback (
Callable, optional) — A function that calls everycallback_stepssteps during inference. The function is called with the following arguments:callback(step: int, timestep: int, latents: torch.Tensor). - callback_steps (
int, optional, defaults to 1) — The frequency at which thecallbackfunction is called. If not specified, the callback is called at every step. - cross_attention_kwargs (
dict, optional) — A kwargs dictionary that if specified is passed along to theAttentionProcessoras defined inself.processor. - clip_skip (
int, optional) — Number of layers to be skipped from CLIP while computing the prompt embeddings. A value of 1 means that the output of the pre-final layer will be used for computing the prompt embeddings.
Returns
TextToVideoSDPipelineOutput or tuple
If return_dict is True, TextToVideoSDPipelineOutput is
returned, otherwise a tuple is returned where the first element is a list with the generated frames.
The call function to the pipeline for generation.
Examples:
>>> import torch
>>> from diffusers import TextToVideoSDPipeline
>>> from diffusers.utils import export_to_video
>>> pipe = TextToVideoSDPipeline.from_pretrained(
... "damo-vilab/text-to-video-ms-1.7b", torch_dtype=torch.float16, variant="fp16"
... )
>>> pipe.enable_model_cpu_offload()
>>> prompt = "Spiderman is surfing"
>>> video_frames = pipe(prompt).frames[0]
>>> video_path = export_to_video(video_frames)
>>> video_pathencode_prompt
< source >( prompt device num_images_per_prompt do_classifier_free_guidance negative_prompt = None prompt_embeds: Optional = None negative_prompt_embeds: Optional = None lora_scale: Optional = None clip_skip: Optional = None )
Parameters
- prompt (
strorList[str], optional) — prompt to be encoded device — (torch.device): torch device - num_images_per_prompt (
int) — number of images that should be generated per prompt - do_classifier_free_guidance (
bool) — whether to use classifier free guidance or not - negative_prompt (
strorList[str], optional) — The prompt or prompts not to guide the image generation. If not defined, one has to passnegative_prompt_embedsinstead. Ignored when not using guidance (i.e., ignored ifguidance_scaleis less than1). - 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 frompromptinput 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 fromnegative_promptinput argument. - lora_scale (
float, optional) — A LoRA scale that will be applied to all LoRA layers of the text encoder if LoRA layers are loaded. - clip_skip (
int, optional) — Number of layers to be skipped from CLIP while computing the prompt embeddings. A value of 1 means that the output of the pre-final layer will be used for computing the prompt embeddings.
Encodes the prompt into text encoder hidden states.
VideoToVideoSDPipeline
class diffusers.VideoToVideoSDPipeline
< source >( vae: AutoencoderKL text_encoder: CLIPTextModel tokenizer: CLIPTokenizer unet: UNet3DConditionModel scheduler: KarrasDiffusionSchedulers )
Parameters
- vae (AutoencoderKL) — Variational Auto-Encoder (VAE) Model to encode and decode videos to and from latent representations.
- text_encoder (
CLIPTextModel) — Frozen text-encoder (clip-vit-large-patch14). - tokenizer (
CLIPTokenizer) — A CLIPTokenizer to tokenize text. - unet (UNet3DConditionModel) — A UNet3DConditionModel to denoise the encoded video latents.
- scheduler (SchedulerMixin) —
A scheduler to be used in combination with
unetto denoise the encoded image latents. Can be one of DDIMScheduler, LMSDiscreteScheduler, or PNDMScheduler.
Pipeline for text-guided video-to-video generation.
This model inherits from DiffusionPipeline. Check the superclass documentation for the generic methods implemented for all pipelines (downloading, saving, running on a particular device, etc.).
The pipeline also inherits the following loading methods:
- load_textual_inversion() for loading textual inversion embeddings
- load_lora_weights() for loading LoRA weights
- save_lora_weights() for saving LoRA weights
__call__
< source >( prompt: Union = None video: Union = None strength: float = 0.6 num_inference_steps: int = 50 guidance_scale: float = 15.0 negative_prompt: Union = None eta: float = 0.0 generator: Union = None latents: Optional = None prompt_embeds: Optional = None negative_prompt_embeds: Optional = None output_type: Optional = 'np' return_dict: bool = True callback: Optional = None callback_steps: int = 1 cross_attention_kwargs: Optional = None clip_skip: Optional = None ) → TextToVideoSDPipelineOutput or tuple
Parameters
- prompt (
strorList[str], optional) — The prompt or prompts to guide image generation. If not defined, you need to passprompt_embeds. - video (
List[np.ndarray]ortorch.Tensor) —videoframes or tensor representing a video batch to be used as the starting point for the process. Can also accept video latents asimage, if passing latents directly, it will not be encoded again. - strength (
float, optional, defaults to 0.8) — Indicates extent to transform the referencevideo. Must be between 0 and 1.videois used as a starting point, adding more noise to it the larger thestrength. The number of denoising steps depends on the amount of noise initially added. Whenstrengthis 1, added noise is maximum and the denoising process runs for the full number of iterations specified innum_inference_steps. A value of 1 essentially ignoresvideo. - num_inference_steps (
int, optional, defaults to 50) — The number of denoising steps. More denoising steps usually lead to a higher quality videos at the expense of slower inference. - guidance_scale (
float, optional, defaults to 7.5) — A higher guidance scale value encourages the model to generate images closely linked to the textpromptat the expense of lower image quality. Guidance scale is enabled whenguidance_scale > 1. - negative_prompt (
strorList[str], optional) — The prompt or prompts to guide what to not include in video generation. If not defined, you need to passnegative_prompt_embedsinstead. Ignored when not using guidance (guidance_scale < 1). - eta (
float, optional, defaults to 0.0) — Corresponds to parameter eta (η) from the DDIM paper. Only applies to the DDIMScheduler, and is ignored in other schedulers. - generator (
torch.GeneratororList[torch.Generator], optional) — Atorch.Generatorto make generation deterministic. - latents (
torch.Tensor, optional) — Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for video generation. Can be used to tweak the same generation with different prompts. If not provided, a latents tensor is generated by sampling using the supplied randomgenerator. Latents should be of shape(batch_size, num_channel, num_frames, height, width). - prompt_embeds (
torch.Tensor, optional) — Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not provided, text embeddings are generated from thepromptinput argument. - negative_prompt_embeds (
torch.Tensor, optional) — Pre-generated negative text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not provided,negative_prompt_embedsare generated from thenegative_promptinput argument. - output_type (
str, optional, defaults to"np") — The output format of the generated video. Choose betweentorch.Tensorornp.array. - return_dict (
bool, optional, defaults toTrue) — Whether or not to return a TextToVideoSDPipelineOutput instead of a plain tuple. - callback (
Callable, optional) — A function that calls everycallback_stepssteps during inference. The function is called with the following arguments:callback(step: int, timestep: int, latents: torch.Tensor). - callback_steps (
int, optional, defaults to 1) — The frequency at which thecallbackfunction is called. If not specified, the callback is called at every step. - cross_attention_kwargs (
dict, optional) — A kwargs dictionary that if specified is passed along to theAttentionProcessoras defined inself.processor. - clip_skip (
int, optional) — Number of layers to be skipped from CLIP while computing the prompt embeddings. A value of 1 means that the output of the pre-final layer will be used for computing the prompt embeddings.
Returns
TextToVideoSDPipelineOutput or tuple
If return_dict is True, TextToVideoSDPipelineOutput is
returned, otherwise a tuple is returned where the first element is a list with the generated frames.
The call function to the pipeline for generation.
Examples:
>>> import torch
>>> from diffusers import DiffusionPipeline, DPMSolverMultistepScheduler
>>> from diffusers.utils import export_to_video
>>> pipe = DiffusionPipeline.from_pretrained("cerspense/zeroscope_v2_576w", torch_dtype=torch.float16)
>>> pipe.scheduler = DPMSolverMultistepScheduler.from_config(pipe.scheduler.config)
>>> pipe.to("cuda")
>>> prompt = "spiderman running in the desert"
>>> video_frames = pipe(prompt, num_inference_steps=40, height=320, width=576, num_frames=24).frames[0]
>>> # safe low-res video
>>> video_path = export_to_video(video_frames, output_video_path="./video_576_spiderman.mp4")
>>> # let's offload the text-to-image model
>>> pipe.to("cpu")
>>> # and load the image-to-image model
>>> pipe = DiffusionPipeline.from_pretrained(
... "cerspense/zeroscope_v2_XL", torch_dtype=torch.float16, revision="refs/pr/15"
... )
>>> pipe.scheduler = DPMSolverMultistepScheduler.from_config(pipe.scheduler.config)
>>> pipe.enable_model_cpu_offload()
>>> # The VAE consumes A LOT of memory, let's make sure we run it in sliced mode
>>> pipe.vae.enable_slicing()
>>> # now let's upscale it
>>> video = [Image.fromarray(frame).resize((1024, 576)) for frame in video_frames]
>>> # and denoise it
>>> video_frames = pipe(prompt, video=video, strength=0.6).frames[0]
>>> video_path = export_to_video(video_frames, output_video_path="./video_1024_spiderman.mp4")
>>> video_pathencode_prompt
< source >( prompt device num_images_per_prompt do_classifier_free_guidance negative_prompt = None prompt_embeds: Optional = None negative_prompt_embeds: Optional = None lora_scale: Optional = None clip_skip: Optional = None )
Parameters
- prompt (
strorList[str], optional) — prompt to be encoded device — (torch.device): torch device - num_images_per_prompt (
int) — number of images that should be generated per prompt - do_classifier_free_guidance (
bool) — whether to use classifier free guidance or not - negative_prompt (
strorList[str], optional) — The prompt or prompts not to guide the image generation. If not defined, one has to passnegative_prompt_embedsinstead. Ignored when not using guidance (i.e., ignored ifguidance_scaleis less than1). - 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 frompromptinput 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 fromnegative_promptinput argument. - lora_scale (
float, optional) — A LoRA scale that will be applied to all LoRA layers of the text encoder if LoRA layers are loaded. - clip_skip (
int, optional) — Number of layers to be skipped from CLIP while computing the prompt embeddings. A value of 1 means that the output of the pre-final layer will be used for computing the prompt embeddings.
Encodes the prompt into text encoder hidden states.
TextToVideoSDPipelineOutput
class diffusers.pipelines.text_to_video_synthesis.TextToVideoSDPipelineOutput
< source >( frames: Union )
Output class for text-to-video pipelines.
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)