MiniMax H3-Image explodes an espresso machine photo into 15 labeled parts
Charlietest123 · reddit · 2026-11-11
A Reddit user tested open-weights MiniMax H3-Image on a demanding editing task: turning a single photo of a vintage Italian espresso machine into an exploded technical view. The model produced 15 separate parts floating along a diagonal with leader lines and English labels, including copper and brass internals not fully visible in the input. H3-Image supports native 4K output (up to 4096×4096) without upscaling, with editing in the same model. The author notes it's a visual concept, not an engineering-accurate diagram — and shares the full prompt.
More from Multimodal
- Stanford's DyCraft Uses Agentic Decomposition to Cut Physical Errors in Long-Horizon 4D Scene Generation — StanfordUniversity · 2026-11-11
- Midjourney SREF pick: --2523103199 for loose pencil-sketch portraits — tisch_eins · 2026-11-11
- H3 Image Tips: 16MP via ComfyUI Workflow Plus Detailed Prompts Yields Best Results — gatortux · 2026-11-11
- Gemini-generated endless loop of quantum cryostats self-assembling — hapticdata · 2026-11-11
- Dad brings his daughter's crayon drawings to life with AI for $1.75, and it's unsettling — Illustrious_Ad_4292 · 2026-11-11
- MOSS-Image-SFT vs MiniMax-H3-Image (Pruned): 1:1 image benchmark across 10 art mediums — SkirtSwimming8950 · 2026-11-11