Layer-and-mask local image editing: turboCLI reuses ComfyUI models, runs 9B FLUX on 4 GB GPU
3unjee · reddit · 2026-10-16
A developer released turboCLI (LGPL) with a different take on local image editing: instead of re-prompting one image, you build the picture from generated layers and masks. The demo generates a knight, cuts her out with InSPyReNet, rebuilds the background without her using FLUX.2 Klein 9B, then moves her, recolors just the castle, switches to night, and changes her hair while keeping her face — no brush, only generation and masking.
Key points:
- Reuses your ComfyUI models folder, nothing downloaded twice.
- turbo-offloader streams weights via ComfyUI's own memory management — the 9B model runs on a 4 GB laptop GPU.
- Speed matches ComfyUI: FLUX.2 Klein 9B in 63.3s vs 62.8s; Krea 2 Turbo 3.5 vs 3.6 s/it at 512².
Supports FLUX.2 Klein, Z-Image-Turbo, Qwen-Image 2.1, Qwen-Image-Edit-2511, Krea 2 Turbo and cutout models on CPU/CUDA/Apple MPS, fully offline; the turbopixel GUI is sourceware with a free month for testers.
Related event: turbopixel: Layer-and-Mask Local AI Image Editing Runs on 4GB VRAM(2 posts)→
More from Multimodal
- FLUX 3 vs Krea 2 vs Z-Image Turbo vs Qwen 2.1: Four-Way Image Model Comparison — Ok_Contribution8157 · 2026-10-16
- AI-Animated Short: Trapped in a Crypt with a Crypto Bro — PolicemansBeard · 2026-10-16
- Redditor Asks: Why No Video Models Built for 24-32GB Consumer VRAM? — BreadUndPeeTears · 2026-10-16
- Dev builds MCP feeding Claude 3,000 real launch videos to fix samey AI output — Designer_Landscape25 · 2026-10-16
- Mirelo SFX V2.0 launches on fal with dialogue-preserving video sound effects — gorkem · 2026-10-16
- Moss-Image samples show startlingly lifelike film-style photography — AgeNo5351 · 2026-10-16