WanAnimatePlus Easy SamplerSettings
Same easy controls, but the sampling happens elsewhere
- model
- image_embeds
- text_embeds
- samples
- context_options
- uni3c_embeds
- sampler_inputs
Easy Sampler runs sampling directly and hands you latents. Easy SamplerSettings is its mirror image: the exact same seven controls - model, image_embeds, steps, cfg, shift, seed, force_offload, scheduler, plus the optional text_embeds, samples, context_options, and uni3c_embeds - but instead of sampling, it packages everything into one SAMPLER_ARGS output for WanAnimatePlus SamplerFromSettings to execute.
Why would you split it? Layout. When one node has both the settings panel and the live preview, the graph gets crowded and the preview gets buried. This two-node split lets you keep the easy settings in one corner and put SamplerFromSettings somewhere it has room to show its progress, while the description of SamplerFromSettings states the motivation plainly: it exists "to look cleaner, useful for the live preview as the main sampler node has become a messy monster." So this is a tidiness node more than a capability node - same engine, same defaults, just a different place in the graph.
The controls are identical, so the advice carries over. steps 30, cfg 6, shift 5, scheduler unipc are the sensible Wan defaults. Drop cfg to ~1.0 when running distilled speed LoRAs; raise steps if you're chasing quality and can afford the time. force_offload on means the model moves back to the offload device after the run, which matters for iterating on small cards. samples from WanAnimatePlus Encode gets you v2v; context_options from WanAnimatePlus ContextOptions gets you clips past the 81-frame window. Nothing is silently dropped - sampler_inputs carries the whole bag.
The one thing to remember: Easy SamplerSettings does not sample. If you wire it up and no video appears, check whether sampler_inputs actually reaches a SamplerFromSettings node - a dangling SAMPLER_ARGS wire is the silent killer in these split-graph setups. The README groups it with Easy Sampler as the recommended path for SCAIL-2 and WanAnimate workflows, precisely because both keep the full sampler functionality while exposing only common controls.
Install via ComfyUI Manager (search "WanAnimatePlus") or git clone https://github.com/wuwukaka/ComfyUI-WanAnimatePlus into custom_nodes, restart. And the usual fork-wide reminder: keep the whole chain on WanAnimatePlus nodes so the SAMPLER_ARGS layout matches what the sampler expects.
Inputs (12)
| Name | Type | Default | Description |
|---|---|---|---|
| model | WANVIDEOMODEL | — | |
| image_embeds | WANVIDIMAGE_EMBEDS | — | |
| steps | INT | 30 | — |
| cfg | FLOAT | 6.000–30 | — |
| shift | FLOAT | 5.000–1000 | — |
| seed | INT | 00–18446744073709550000 | — |
| force_offload | BOOLEAN | true | Moves the model to the offload device after sampling |
| scheduler | COMBO | unipc | 21 options: unipc, unipc/beta, dpm++, dpm++/beta, dpm++_sde, dpm++_sde/beta, +15 |
| text_embedsopt | WANVIDEOTEXTEMBEDS | — | |
| samplesopt | LATENT | init Latents to use for video2video process | |
| context_optionsopt | WANVIDCONTEXT | — | |
| uni3c_embedsopt | UNI3C_EMBEDS | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| sampler_inputs | SAMPLER_ARGS | — |