Claude
ec2d8fe326
Seven production pipeline utility nodes for ComfyUI: - PD Project Context: top-level metadata/token source - PD Shot Token: lightweight template string resolver - PD Output Namer: structured output filename builder - PD Network Output: image writer with UNC path support - PD Switch: multi-channel wildcard input router (up to 8×8) - PD Wedge: iteration node with optional auto-queue injection - PD Rasterizer: dry-run wedge cartesian product previewer Core infrastructure: - core/compat.py: legacy + Nodes 2.0 (V3) API compatibility shim - core/tokens.py: standalone token resolution engine - core/utils.py: path normalization and directory helpers - web/pipedream_rasterizer.js: minimal JS for rasterizer preview https://claude.ai/code/session_015UV4rjSvFatMiNtKhd1sT3
1.3 KiB
1.3 KiB
Pipedream Workflow Library
Example workflows demonstrating how to use Pipedream nodes. Each subdirectory contains a workflow.json (when available) and a README explaining the setup.
Loading Workflows
- Drag the
workflow.jsonfile into ComfyUI, or use File > Load - All workflows require the Pipedream node pack to be installed
Progression
The workflows are numbered as an intentional learning path:
| # | Workflow | What It Teaches |
|---|---|---|
| 01 | Basic Output Namer | Wire PD Output Namer with default tokens, feed into Save Image |
| 02 | Network Output | Replace Save Image with PD Network Output, UNC path setup |
| 03 | Project Context + Tokens | Use PD Project Context as the token root; context flows downstream |
| 04 | Switch Node | Single-channel and multi-channel switch; manual vs value-driven mode |
| 05 | Single Wedge | One PD Wedge iterating over a prompt list; queue injection basics |
| 06 | Multi-Wedge + Rasterizer | Two wedges (prompt + seed), Rasterizer preview, cartesian product |
| 07 | Full Production Shot | All nodes together in a complete production pipeline |
Workflows 01–04 use only core ComfyUI nodes as the generative backbone. Workflows 05–07 use a simple SD1.5/SDXL backbone as placeholder.