216 hand-drawn style prompts for GPT image2: pick by number, reuse forever
xiaohu · x · 2026-09-07
GitHub user yang0 built handraw-style, a library of 216 numbered hand-drawn styles for GPT image2, each linking a reference name, a generation label, and a sample image. A companion Skill expands 'number + subject' into Chinese/English prompts so a chosen style can be reliably reused. Styles span seven groups: international humor comics, narrative picture books, flat graphic characters, Japanese illustrators, Chinese illustrators, medium/regional styles, and a contemporary Chinese supplement. Key practice: judge styles by line, color, shape, and texture rather than subject matter, and fall back to numbered reference images when the model doesn't understand style names. Entries 201–216 lack full visual descriptors yet.
Related event: handraw-style Library Offers 216 Hand-Drawn Styles for GPT Image(2 posts)→
More from Multimodal
- Astra turns photogrammetry scan of old home into editable 3D scene, hinting inverse graphics has crossed the chasm — bilawalsidhu · 2026-09-07
- Trellis2/pixel3d ComfyUI template workflow generates detailed 3D models in minutes — Disastrous-Agency675 · 2026-09-07
- "It's the artist that doesn't need the studios": Creator demos Minimax H3 workflow in ComfyUI — jnack · 2026-09-07
- Indie dev cancels $10/mo Midjourney after 3 years as GPT image generation catches up — yihui_indie · 2026-09-07
- Three diffusion papers head to ECCV: Flash-BoN, PFM and dynamic image-gen benchmarks — RisingSayak · 2026-09-07
- Motion-Omni generates real-time full-body co-speech motion aligned with spoken dialogue — PKU1898 · 2026-09-07