v0.2.3-alpha - Per-Node State Registry for Multi-DazzleCommand Workflows
Dazzle Command v0.2.3-alpha
Workflow orchestration node for ComfyUI — coordinate seed control and execution gates from a single play/pause toggle. Part of the DazzleNodes collection.
What's New in v0.2.3-alpha
Per-Node State Registry (#5)
The core fix for multi-DazzleCommand workflows. Each DazzleCommand now maintains independent state in a per-node registry (DazzleCommandState class), replacing the global sys._dazzle_command_state singleton that caused cross-talk between DazzleCommand instances.
- DC-1 set to Play and DC-2 set to Pause now operate independently
- Each PBE reads its own connected DazzleCommand's state via the signal noodle
- Each SmartResCalc reads its connected DC's seed intent via
_dazzle_dc_idmarker - API endpoint keys state by
nodeIdfrom JS
Multi-Node Seed Fix (#4)
"Reuse last seed" now works correctly in multi-node workflows. Combined with SmartResCalc v0.11.3's per-node lookup and PBE v0.4.2-alpha's deterministic preview filenames:
- Seed lock verified: same seed across play repeats
- Cache-transparent: repeat runs execute in 0.01s (fully cached)
Example Workflows
Added example workflow JSON files demonstrating DazzleCommand with different KSampler configurations.
Installation
Install via ComfyUI Manager or clone:
cd ComfyUI/custom_nodes
git clone https://github.com/DazzleNodes/ComfyUI-DazzleCommand.gitVersion History (0.2.x)
| Version | Key Change |
|---|---|
| v0.2.3-alpha | Current — per-node state registry, multi-DC support |
| v0.2.2-alpha | Signal carries active_state (interim fix) |
| v0.2.1-alpha | Dynamic JS imports for DazzleNodes aggregate |
| v0.2.0-alpha | 5-mode seed control, cache-transparent orchestration |
Companion Versions
For full multi-pipeline orchestration:
- SmartResCalc v0.11.3 — per-node state lookup
- PBE v0.4.2-alpha — deterministic preview filenames, cache-compatible
Platform Support
| Platform | Status |
|---|---|
| Windows 11 + CUDA | Tested |
| Linux | Expected to work |
| macOS | Expected to work |
Requirements
- Python 3.10+
- ComfyUI (latest recommended)
- No additional pip dependencies