用参数化提示词模板让 MiniMax H3 Image 生成完整漫画页
有人把 MiniMax H3 Image 玩出了漫画工作流：题材、分镜、网点、拟声词全做成可换参数，换一下就出整页新漫画。模板免费公开在回复区，想批量产漫画页或研究排版控制的话直接抄走改。
Kōda（@aimikoda）基于 MiniMax 的 H3 Image 模型构建了一套参数化提示词模板，用于生成完整漫画页。
模板将漫画页拆解为题材、分镜布局、网点纸（tones）和拟声词（SFX）等可替换参数，用户改动任一参数即可得到风格完全不同的页面。作者演示的效果被 Hailuo AI 官方账号转发，提示词模板已在推文回复区公开。
这类工作流的核心在于把单张图生成升级为带版式结构的多格页面生成，分镜、留白和文字气泡都需要在提示词中显式约束。对做漫画创作、内容营销素材或想研究图像模型排版控制的人来说，模板可以直接套用改造。
straight back to reading manga as a kid and imagining the next page 🦸 Kōda @aimikoda MiniMax H3 Image × Manga Built a parametric prompt template for generating full manga pages. Swap the genre, panel layout, tones and SFX and you get a completely different page. #H3Image @MiniMax_AI Prompt template in the replies 👇 🔗 View Quoted Tweet📊 2 ⚡
This report is categorized under AITOP's "How-to" desk, focusing on China's AI industry and domestic foundation model ecosystem. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and Chinese AI companies expanding and localizing across Southeast Asia and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include MiniMax, H3 Image, 图像生成, 提示词工程.
Source attribution: curated from Hailuo AI (@Hailuo_AI).
