Open-source ComfyUI workflows add slot-based reference control to Nano Banana Pro
Cute-Appointment6874 · reddit · 2026-09-13
Developer haristahir1 released two open-source ComfyUI workflows built around Nano Banana Pro (bring your own Google API key; free via Vertex AI's $300 trial credit), designed to replace prompt-roulette image generation with slot-based control:
- Multi Reference Shot: composites one frame from multiple references, each assigned to a specific aspect — lighting, composition, blocking — plus character images for consistency;
- Character Swap: drops your own character into any shot, preserving the original framing, light and pose, and can swap clothes in the same pass while keeping the frame's shape.
The core idea is per-slot control: each reference image contributes only what its slot says, so you fix the pose by adjusting blocking and face nodes independently instead of rewriting one giant prompt. Both workflows are on GitHub; all sample images in the post were produced with them.
Related event: Two Open-Source ComfyUI Workflows Unleash Nano Banana Pro(2 posts)→
More from Multimodal
- 10-Person Startup Phonic Claims #1 on Five Speech Benchmarks With Phonic-1 — graceisford · 2026-10-31
- Halloween doorway: Midjourney prompt builds a giant moth crossing a star-forest cliff — tisch_eins · 2026-10-31
- Open-source ComfyUI Motion Previs Studio brings 3D camera blocking to AI video — navarisun · 2026-10-31
- Open-Source ComfyUI Motion Previs Studio Blocks 3D Shots Before Video Generation — navarisun · 2026-10-31
- Hooking Mage's MCP into Claude generated narrative image sets in one shot — AIandDesign · 2026-10-31
- At 3 am she stops scrolling and a whole town answers in songs — AI short film — ownubie · 2026-10-31