init: carolina-unicorn workflow + model ideas backlog

ideas/WORKFLOW-IDEAS.md

@@ -0,0 +1,35 @@
+# ComfyUI Workflow Ideas
+
+## Models to add
+
+Current: v1-5-pruned-emaonly.safetensors (SD 1.5 baseline quality).
+
+Recommended additions (model path: docker volume comfyui_models):
+- FLUX.1-dev: best photorealism and text rendering
+- SDXL base + refiner: much better composition
+- SDXL Lightning: 4-step fast iteration
+- IP-Adapter: consistent character/face style injection
+- AnimateDiff: animated outputs
+
+## Planned workflows
+
+### Carolina portraits (character consistency)
+Goal: multiple scenes with the same child character
+Approach: IP-Adapter face LoRA + consistent prompt
+Status: blocked on IP-Adapter model
+
+### Children book illustrations (SD 1.5 viable now)
+Style prompt: children book illustration, watercolor painting, soft edges, pastel colors
+Works with current SD 1.5 model
+
+### Food photography (for Meater BBQ results)
+Prompt pattern: [dish], professional food photography, soft studio lighting, 50mm lens, plated
+Needs: SDXL or FLUX.1-dev
+
+### 3D print preview (img2img from slicer screenshot)
+Goal: photorealistic render before printing
+Approach: img2img on slicer preview at denoise 0.5-0.7
+
+### Home scene visualization
+Goal: visualize smart home lighting scenes
+Approach: inpaint room photos with different lighting