Seamless video join (loop) with Wan 2.1 VACE — video join v1.3

Use this workflow to join two video clips into one video or make a loop from a single video clip; it takes video clips as input and produces a combined video.

Its 65 nodes load videos, stitch images, encode with Wan VACE, sample, decode, and combine the result into video.

For a join, provide two video clips; for a loop, provide one video clip.

Suggestion · not verified

The listed release is video join v1.3 from Brainburns894.

The workflow lists these node packs: ComfyLiterals, comfyui_essentials, comfyui-kjnodes, comfyui-logicutils, comfyui-videohelpersuite, ComfyUI-WanVideoWrapper, rgthree-comfy, and scale-image-to-total-pixels-advanced.

It also lists these model files: umt5-xxl-enc-bf16.safetensors, wan_2.1_vae.safetensors, Wan2_1_self_forcing_sid_v2_1_3B_lora_rank_32_fp16.safetensors, Wan2_1_VACE_1_3B_preview_bf16.safetensors, and Wan2_1-T2V-1_3B_bf16.safetensors.

To match the listed release, set the base model to Wan Video 1.3B t2v.

Suggestion · not verified

Before use, verify that your setup has each listed model file under its exact filename.

Suggestion · not verified

No source URL is listed for Wan2_1_self_forcing_sid_v2_1_3B_lora_rank_32_fp16.safetensors or Wan2_1-T2V-1_3B_bf16.safetensors; verify how you will obtain them. License terms are not specified, so check them before use.

Suggestion · not verified