ComfyUI-MiniMaxH3-Easy

ComfyUI-MiniMaxH3-Easy is a community-built custom node suite for ComfyUI that wraps the MiniMax H3 nodes into a smaller set of nodes. One main node covers text-to-video, image-to-video, first/last-frame generation, reference-to-video and digital-human (audio-driven) generation, and it can chain segments to build longer videos. A later addition supports MiniMax H3 Image for text-to-image and reference-image editing.

Its main additions are a unified Media Loader for images, video and audio (upload, drag, paste, reorder, replace, toggle), @ references inserted directly in prompts, optional prompt optimization through an API, and per-segment refinement (pixel resize, 3D latent upscale, low-VRAM tiling). Samplers, LoRAs, attention patches, decoders and save nodes remain ordinary ComfyUI connections. A model adapter node lets other native, community or GGUF loaders be used.

It is installed by cloning into ComfyUI/custom_nodes or through ComfyUI Manager (the README recommends the Nightly version). It requires a recent ComfyUI that includes the official MiniMax H3 nodes (0.45.0 or newer for the image workflow) and separately downloaded H3 model files. Example workflows are included in the repository; some need extra models, LoRAs or custom nodes. It runs locally, so hardware requirements depend on the chosen models.