Add grouped custom sampling options

Fix ComfyUI PromptServer asset manager initialization

Reuse final prompts in detail passes

Place metadata upscaler after detailers

Add independent second-pass style transfer

Reuse bucket CLIP Vision model

Support averaged style references

Add InstantStyle reference guidance

Generate configurable image batches

Add attention-coupled regional prompting

Improve detail passes and upscale sizing

Add repeatable ComfyUI detailer passes

Prioritize regional subjects before environment

Improve regional subject composition

Stage regional prompt conditioning

Feather regional prompt masks

Preserve and isolate regional prompts

Add selectable upscale models

Add Anima generation controls

Use generation 2 names and seed

Add AlignYourSteps and scaled API output

Standardize configurable Comfy pipeline

Support configurable CFG

Support configurable generation options

Preserve workflow metadata in archives

Close ComfyUI event loop cleanly

Add Comfy-compatible Gradio image API