younestft/comfy_everanimate

Comfy EverAnimate is a set of custom add-on nodes for ComfyUI, a visual tool that lets people build AI image and video generation pipelines by connecting blocks together on a canvas. This particular set of nodes is focused on generating longer AI videos by splitting the work into smaller pieces called chunks and then joining them together smoothly. The core problem it addresses is that AI video models typically can only generate a short clip at a time. To make a longer video, you have to generate several short clips and stitch them together, but the transition between chunks often looks jarring because each chunk starts fresh without knowing what came before. This project tries to reduce those boundary artifacts by carrying information forward from one chunk to the next, including decoded image frames and motion state. The workflow involves a master settings node that holds global configuration like resolution, chunk length, and the reference image. From there, an initial chunk node handles the first segment of the video, and one or more extension chunk nodes each handle the following segments. Each extension chunk receives the output of the previous one and uses it to anchor the continuation. A separate color correction node can smooth out brightness and color shifts specifically at the points where chunks meet. A calculator node lets you preview how many chunks a given video will require based on your chosen chunk length and the length of your pose guide video, so you can plan the workflow before running anything. Installing the nodes means cloning this repository into the custom_nodes folder inside an existing ComfyUI installation and restarting ComfyUI. The project is described as a work in progress and does not depend on third-party ComfyUI wrapper packages, working instead with ComfyUI's built-in native video nodes.