WanAnimatePlus Easy Sampler

The full WanAnimatePlus Sampler is a wall of inputs - APG omegas, RoPE functions, chain guidance, loop args, half a dozen optional embeds. If you just want to render a clip, that's a lot of surface area to trip over. Easy Sampler is the same sampling engine with the panel stripped down to the seven controls that actually matter, and the README explicitly recommends it for both SCAIL-2 and plain WanAnimate workflows. Start here; graduate to the full sampler only when you need its extra knobs.

The seven knobs. model (your WanAnimatePlus ModelLoader output) and image_embeds (from AnimateEmbeds, SCAIL_2 Embeds, or Bernini) are the plumbing. Then: steps (default 30), cfg (default 6), shift (default 5), seed, force_offload (on - moves the model to the offload device after sampling so your VRAM frees up), and scheduler (default unipc, with 21 choices including the euler/beta fast favorite, dpm++ variants, and distilled lcm/longcat_distill_euler).

Those defaults are worth a second thought for Wan. Community consensus is that flow-matching Wan wants euler/beta or unipc with shift around 5 - the defaults here match the wrapper's established recipe. Speed LoRAs (LightX2V and friends) force CFG down toward 1.0, so if you're running a distilled model, drop cfg accordingly; at full quality, keep it at or above 3.5.

What you can still add. text_embeds (from the pack's TextEncodeCached) if you're using a text prompt. samples for video-to-video - plug WanAnimatePlus Encode's output here to start from existing footage instead of noise. context_options for long clips (from WanAnimatePlus ContextOptions), and uni3c_embeds if you're in a Uni3C control workflow. Nothing you wire in here is wasted; it all lands on the same full sampler path underneath, which is the point of the node.

Outputs. samples (the latents, for WanAnimatePlus Decode) and denoised_samples (the pre-VAE denoised latents, mostly useful for previewing or inspecting the model's raw output).

Install via ComfyUI Manager (search "WanAnimatePlus") or git clone https://github.com/wuwukaka/ComfyUI-WanAnimatePlus into custom_nodes, restart. One fork-wide rule to keep in your back pocket: use the WanAnimatePlus chain end to end. This sampler expects image embeds produced by the fork's embeds nodes - mixed-wrapper graphs are where "why is my output degraded" threads come from.