Geekatplay Dev
281401e3da
Adds a stronger SDXL QR hidden-image workflow and separates reusable blueprints from the runnable workflows, alongside node and installer updates. - workflows/sdxl_qr_hidden_image_ultra_strong.json: new higher-strength variant of the SDXL QR workflow. - workflow_blueprints/: blueprint copies of the flux ghost, sd15 QR, sdxl QR, and unified router graphs, kept separate from the executable workflows. - illusion_nodes.py, install.bat, docs/WORKFLOWS.md, README.md: node and install/documentation updates to match.
1.3 KiB
1.3 KiB
Workflow Notes
The workflows/ directory now contains importable ComfyUI starter prompts instead of the old human-readable blueprint JSON. These files should open as non-empty starter canvases in ComfyUI.
The original reference notes were preserved in workflow_blueprints/.
What is in the starter prompts:
- A runnable NanoBanana preprocessing path with preview outputs, so loading and queueing the file does something useful immediately.
- SD 1.5 and SDXL starter branches with placeholder checkpoint and ControlNet loader nodes already laid out.
- A FLUX preprocessing starter focused on pattern prep and latent-mask generation, since full FLUX loader stacks vary across installs.
Recommended usage:
- Open a starter from
workflows/. - Replace the placeholder checkpoint and ControlNet selections with models that exist in your install.
- Swap
NanoBanana Text MaskforLoadImageif you want to drive the workflow from a QR image or logo file. - For FLUX, route the
NanoBanana Latent Maskoutput intoSet Latent Noise Maskinside your preferred FLUX sampler chain.
The original blueprint split still applies:
- SD 1.5 stays aggressive and is best for scannable QR or obvious hidden text.
- SDXL uses optical-pattern composition at higher resolution.
- FLUX uses low-strength ControlNet plus latent masking to avoid fighting the model's coherence.