ComfyUI-Egregora-ARMD

EgregoraAdaptiveDiffusionApply

Applies ARMD to a model. This is the main node that patches the model so regional conditioning can be used during denoising. It: - reads the region plan - reads per-region conditioning - builds regional runtime batches - uses context regions for reading - writes back using feathered local ownership In v0.1.2, batching is also more efficient for mixed prompt sizes, and debug logging is clearer when batch order differs from spatial order.

Egregora ARMD · 5 inputs · 2 parameters · 1 output

EgregoraRegionalConditioning

Encodes one prompt per region. It creates: - regional positive conditioning - regional negative conditioning - placeholder positive conditioning - placeholder negative conditioning These placeholders are useful for compatibility with nodes that still expect standard conditioning objects. In v0.1.2, regional conditioning also preserves: - ccrossattn - pooledoutput when available - extra conditioning fields needed for safer SDXL-style regional injection

Egregora ARMD · 2 inputs · 2 parameters · 5 outputs

EgregoraRegionPlan

Builds the spatial layout used by ARMD. It: - aligns the working canvas - divides the image into core regions - expands them into context regions - prepares batching metadata - outputs region order text for prompt assignment - optionally works from an IMAGE, a LATENT, or both This node is the backbone of the workflow.

Egregora ARMD · 2 inputs · 6 parameters · 6 outputs

EgregoraRegionSelect

Lets you inspect a single region for: - manual prompt writing - captioning workflows - debugging region order

Egregora ARMD · 2 inputs · 1 parameter · 1 output

EgregoraRestoreOriginalSize

Restores the final image back to the original framing after padreflect alignment. This is useful because ARMD may internally pad the image to a safer working canvas, while you still want the final output to match the original framing. Note: this node can restore padding-based alignment. It cannot recover pixels removed by floorcrop, because cropping is destructive.

Egregora ARMD · 2 inputs · 0 parameters · 1 output

EgregoraRuntimeAdapterMerge

Merges multiple runtime payload adapters.

Egregora ARMD · 3 inputs · 0 parameters · 1 output

EgregoraSpatialTensorPack

Packs spatial tensors into a runtime payload adapter. Useful for advanced workflows that pass extra runtime tensors.

Egregora ARMD · 4 inputs · 4 parameters · 1 output

EgregoraStaticPayloadPack

Packs static runtime values into a runtime payload adapter.

Egregora ARMD · 0 inputs · 4 parameters · 1 output