Extensions/ComfyUI_StreamingT2V
ComfyUI Extension

ComfyUI_StreamingT2V

ComfyUI_StreamingT2V

By chaojie·Created 2 years ago·Updated 2 years ago· 36
chaojie/ComfyUI_StreamingT2V
Nodes24
On cloudLocal install
Categoryutils, StreamingT2V
Stars36
Updated2 years ago

Nodes (24)

LoadText_StreamingT2V

A boring text-file reader that the prompt-travel workflow quietly depends on

utils
PromptTravelIndex

The little node that answers 'which prompt is running at frame N?'

StreamingT2V
SaveText_StreamingT2V

The save half of StreamingT2V's state loop

utils
StreamingT2VLoaderAnimateDiff

The all-in-one loader that crams an entire long-video pipeline into one node

StreamingT2V
StreamingT2VLoaderAnimateDiffModel

The AnimateDiff seed generator, loaded on its own

StreamingT2V
StreamingT2VLoaderEnhanceModel

Turn 256px long videos into 720p

StreamingT2V
StreamingT2VLoaderModelscopeModel

The Modelscope T2V short-video generator, on its own

StreamingT2V
StreamingT2VLoaderModelscopeT2V

Modelscope seed, StreamingT2V extension, enhance built in

StreamingT2V
StreamingT2VLoaderStreamModel

Load just the autoregressive heart of StreamingT2V

StreamingT2V
StreamingT2VLoaderSVD

SVD seed, StreamingT2V extension, enhance

StreamingT2V
StreamingT2VLoaderSVDModel

The image-to-video seed generator

StreamingT2V
StreamingT2VLoaderVidXTendModel

VidXTend, the 5GB StreamingT2V that runs on 10GB VRAM

StreamingT2V
StreamingT2VRunEnhanceStep

Upscale your long 256px video to 720p with one pass

StreamingT2V
StreamingT2VRunI2V

SVD seed, StreamingT2V extension, enhance

StreamingT2V
StreamingT2VRunLongStep

Extend your seed into a long 256px video

StreamingT2V
StreamingT2VRunLongStepVidXTendPipeline

VidXTend long extension, IMAGE in, IMAGE out

StreamingT2V
StreamingT2VRunLongStepVidXTendPipelineCustomRef

VidXTend extension with custom reference frames for appearance control

StreamingT2V
StreamingT2VRunLongStepVidXTendPipelineCustomRefOutExtendOnly

The VidXTend variant that conditions on the whole video so far

StreamingT2V
StreamingT2VRunLongStepVidXTendPipelinePromptTravel

VidXTend long extension where the prompt changes mid-video

StreamingT2V
StreamingT2VRunShortStepAnimateDiff

16 frames of AnimateDiff, the seed every long video starts from

StreamingT2V
StreamingT2VRunShortStepModelscopeT2V

The 16-frame seed that starts every StreamingT2V long video

StreamingT2V
StreamingT2VRunShortStepSVD

One still image becomes the opening of a long video

StreamingT2V
StreamingT2VRunT2V

The all-in-one node that turns a prompt into a minutes-long mp4

StreamingT2V
VHS_FILENAMES_STRING_StreamingT2V

A tiny bridge that turns a VHS filename into a plain string

StreamingT2V
Readme

ComfyUI-StreamingT2V

Painebenjamin created the VidXTend model, which reduces the size from the original 25.6GB to 5GB. When using the VidXTend process, there is no need to manually download the model, as it will be downloaded automatically.

If you use the original model, please download the StreamingT2V model and place it in folder ComfyUI/model/checkpoints

VidXTend (<10G VRAM) | Original (<15G VRAM)

workflow

Recommended complete Workflow

<img src="workflow/wf_final.png" raw=true>

Recommended STAGE 2 Workflow

<img src="workflow/wf_VidXTend.png" raw=true>

Test prompt travel cyclist Workflow

<img src="workflow/wf_prompttravel.png" raw=true>

Open the Auto Queue option, Stop it when you want to.

<img src="workflow/cyclist.png" raw=true> <details> <summary>More Workflow</summary>

Step 1 (It's not necessary.) Generate short video (16x256x256):

You can freely generate a 16x256x256 video without using the process below. You can even use videos downloaded from the internet.

AnimateDiff Option: https://github.com/chaojie/ComfyUI_StreamingT2V/blob/main/workflow/wf_short_ad.json

<img src="workflow/wf_short_ad.png" raw=true>

Modelscope Option: https://github.com/chaojie/ComfyUI_StreamingT2V/blob/main/workflow/wf_short_ms.json

<img src="workflow/wf_short_ms.png" raw=true>

SVD Option: https://github.com/chaojie/ComfyUI_StreamingT2V/blob/main/workflow/wf_short_svd.json

<img src="workflow/wf_short_svd.png" raw=true>

Step 2 Generate long video (nx256x256):

VidXTend Option: https://github.com/chaojie/ComfyUI_StreamingT2V/blob/main/workflow/wf_VidXTend.json

<img src="workflow/wf_VidXTend.png" raw=true>

Original Option: https://github.com/chaojie/ComfyUI_StreamingT2V/blob/main/workflow/wf_long.json

<img src="workflow/wf_long.png" raw=true>

Step 3 Generate long enhanced video (nx512x512):

https://github.com/chaojie/ComfyUI_StreamingT2V/blob/main/workflow/wf_enhance.json

<img src="workflow/wf_enhance.png" raw=true>

You can generate all by one workflow (MORE VRAM)

ModelscopeT2V (T2V)

https://github.com/chaojie/ComfyUI_StreamingT2V/blob/main/workflow/wf.json

<img src="workflow/wf.png" raw=true>

AnimateDiff (T2V)

https://github.com/chaojie/ComfyUI_StreamingT2V/blob/main/workflow/wf_ad.json

<img src="workflow/wf_ad.png" raw=true>

SVD (I2V)

https://github.com/chaojie/ComfyUI_StreamingT2V/blob/main/workflow/wf_svd.json

<img src="workflow/wf_svd.png" raw=true>

StreamingT2V

</details>