My stuff

  • My Workflows

  • Liked Workflows

  • Following Workflows

Go to OpenArt main site
Upload workflow

LongerFantasyTalking

5.0

0 reviews
3
913
150
2
Description

Result Video:


Many versions of workflows show you how to create a 5-second video and showcase the results, but few explain how to create a longer virtual human video—over one minute in length.

This workflow is built upon the basic structure of WanVideo combined with FantasyTalking, enhanced with IndexTTS and Loop. It also includes a custom-designed audio segmentation node, as well as Uni3C for stable visuals.

If the MultiTalking branch is officially adopted by ComfyUI in the future, you can also use MultiTalking as part of this workflow.

The video shown here was automatically generated using this workflow, preserving the original output exactly. Since I’m a Chinese speaker, the subtitles were generated in Chinese, which caused the auto-generated English subtitles to overflow outside the frame. If you need English subtitles, make sure each line isn’t too long, or manually edit the subtitles.

With just 12GB of VRAM, this workflow completed the entire video—including TTS—in approximately 3,000 seconds. The video was divided into 30 segments of 81 frames each, resulting in 30 sampling cycles.

Some visual jitter is still present, but it has been significantly reduced.

Discussion

(No comments yet)

Loading...

Author

67
121.8K
2.0K
495.8K

No reviews yet

  • - latest (2 months ago)

Primitive Nodes (32)

AddSubtitlesToVideo (1)

AudioListGenerator (1)

CreateCFGScheduleFloatList (1)

DisplayAny (1)

DownloadAndLoadWav2VecModel (1)

FantasyTalkingModelLoader (1)

FantasyTalkingWav2VecEmbeds (1)

Fast Groups Bypasser (rgthree) (1)

GeminiAPI (1)

ImageResizeKJv2 (1)

IndexTTSRun (1)

LoadWanVideoT5TextEncoder (1)

SaveAudio (1)

SetNode (1)

TextBox1 (1)

VHS_SelectFilename (1)

WanVideoBlockSwap (1)

WanVideoClipVisionEncode (1)

WanVideoDecode (1)

WanVideoEncode (1)

WanVideoImageToVideoEncode (1)

WanVideoLoraSelect (1)

WanVideoModelLoader (1)

WanVideoSampler (1)

WanVideoTextEncode (1)

WanVideoUni3C_ControlnetLoader (1)

WanVideoUni3C_embeds (1)

WanVideoVAELoader (1)

easy batchAnything (1)

easy forLoopEnd (1)

easy forLoopStart (1)

easy indexAnything (1)

Custom Nodes (18)

ComfyUI

  • - RepeatImageBatch (1)

  • - CLIPVisionLoader (1)

  • - LoadImage (1)

  • - ImageUpscaleWithModel (1)

  • - UpscaleModelLoader (1)

  • - ImageFromBatch (1)

ComfyUI Easy Use

  • - easy float (1)

  • - GetImageSize+ (1)

  • - LayerUtility: PurgeVRAM (2)

  • - LoadAudio (1)

  • - VHS_VideoCombine (2)

  • - VHS_LoadVideoPath (1)

  • - VHS_VideoInfoSource (1)

  • - INTConstant (1)

  • - StringFunction|pysssss (1)

  • - ShowText|pysssss (1)

Checkpoints (0)

LoRAs (1)

Wan\Wan21_T2V_14B_lightx2v_cfg_step_distill_lora_rank32.safetensors