Anima Artist Starter

AnimaArtistStarter is the pack's "I don't want to read a manual yet" node. It doesn't do the mixing itself - it builds the pieces you need and walks you through wiring them. The README bills it as the fastest first run: fill in the artist_table, read the in-UI guide, and you're generating. If you've been staring at the bigger preset workflow and bouncing off, this is the gentler door in.

It's a cousin of AnimaArtistBasic, but where Basic swallows the whole pipeline into one box, Starter hands you a clean artist chain plus a preset payload and tells you what to connect next. That's actually a nicer on-ramp for understanding the machinery, because you end up with the standard AnimaArtistPack → AnimaArtistPresetApply graph instead of a black box.

What you fill in

• recipe - balanced (default), strong_style, drift_auto, or prompt_passthrough. Same four recommended modes as Basic.
• artist_table - the star of the show. One artist per line, in artist | weight | layers | timing form. Just the artist name works; weight defaults to 1.0 and empty layers/timing auto-fill from the layout. So @wlop is fine, and so is @wlop | 1.2 | 0-8 | 0.0-0.45.
• layout - layer_scheduled (default, auto layers + sampling windows, recommended), even_layers (split DiT blocks evenly), or manual (use the table's routes as-is).
• intensity - recipe strength multiplier for the three mixing recipes; ignored by prompt_passthrough.
• Optional extras: normalize_weights, a global layer_mode shortcut, a custom_layer_filter for layer_mode=custom, and num_blocks for the preview (Anima default is 28).

What it hands you

Four outputs: artist_chain (feed into AnimaArtistPack.artist_chain), preset (ANIMA_PRESET, feed into AnimaArtistPresetApply.preset), advanced_options (ANIMA_OPTS), and guide - a STRING that spells out the exact wiring in the node's own UI panel. On older ComfyUI frontends that don't render that panel, drop guide into a Show Text node instead.

Install

ComfyUI Manager → search "Anima Artist Mixer Forge", or:

cd ComfyUI/custom_nodes
git clone https://github.com/peter119lee/Anima-Artist-Mixer-Forge

then restart. No extra dependencies; you need the usual Anima trio (UNET in models/diffusion_models, qwen_3_06b_base in models/text_encoders, qwen_image_vae in models/vae) and the Anima-compatible CLIP loader. Don't run it alongside the original Anima-Artist-Mixer pack - duplicate node names shadow each other.

My take

layer_scheduled is the layout you actually want for your first few runs; it spreads artists across early/mid/late layers and sampling windows so nobody stomps on anybody. And when you inevitably wonder "is this artist even doing anything?", Starter's guide points you toward the pack's diagnostics - that's where the real fun starts.