MOSS-Image ecosystem roundup: sharpness slider, manga alpha LoRA, box inpainting
optimisticalish · reddit · 2026-10-27
A daily roundup of new MOSS-Image community releases (Oct 27):
- 18MB Sharpness Slider LoRA — moves images between sharp and blurry via LoRA strength alone, no trigger needed
- New theme LoRAs for 1950s sci-fi and 1950s western movie aesthetics
- Manga alpha-transparency LoRA trained via a special Flux.2 VAE (not chroma keying) for characters, ink effects, and speech balloons; works in color too, with ComfyUI workflow
- Red-box inpainting combined with extracted Depth reference as control, with MaskMaker nodes and workflows
- ComfyUI layout workflows for manga pages, anime e-konte storyboards, six-panel imageboards, and sketch grids
- Thicker-lineart Nu-MOSS LoRA variant
- Fizgig update: train only on the part of each image that matters (e.g. faces); users report strong character LoRA results
- Per-block strength editing for all 42 blocks in MOSS Base bf16
More from Multimodal
- ComfyUI Krea2 Prompt Styler Node Revived in a Fork With a Massive Style Library — Winter_unmuted · 2026-10-27
- Single HTML file generates retro poster designs in-browser with zero LLM calls — Afinetheorem · 2026-10-27
- A Redditor Used Copilot to Imagine and Archive Their Own '80s/'90s Winter Streetwear Concepts — Valuable-Sky2134 · 2026-10-27
- Transkun open-source piano transcription model nails MIDI, sustain events included — alexanderchen · 2026-10-27
- fal adds FLUX 3 Image Klein LoRA training: 10 reference images, 3 LoRAs per request — OdinLovis · 2026-10-27
- Guy turns his girlfriend's selfie into a playable AI game ad about popping pimples — CurieuxExplorer · 2026-10-27