用AI把角色概念一键生成精准像素的我的世界皮肤
BLOCK: An Open-Source Bi-Stage MLLM Character-to-Skin Pipeline for Minecraft
- 分两阶段：先用大模型生成前后双视角预览图，再转为皮肤贴图
- 支持任意概念生成，输出与原版游戏像素完全对齐
- 开源完整流程，适合游戏开发与创意设计者使用
我们提出 extbf{BLOCK}，一个开源的双阶段角色到皮肤生成流水线，可将任意角色概念生成像素精确的 Minecraft 皮肤。BLOCK 将问题分解为（i）基于大型多模态模型（MLLM）的3D预览合成阶段，采用精心设计的提示与参考模板，生成一致的双面板（正面/背面）斜视风格预览图；（ii）基于微调后的 FLUX.2 模型的皮肤解码阶段，将预览图转换为皮肤图集图像。我们进一步提出 extbf{EvolveLoRA}，一种渐进式 LoRA 教学策略（文本到图像 → 图像到图像 → 预览到皮肤），通过前一阶段的适配器初始化下一阶段，提升稳定性与效率。BLOCK 已发布所有提示模板与微调权重，支持可复现的角色到皮肤生成。
原文摘要 · Abstract (English)
We present \textbf{BLOCK}, an open-source bi-stage character-to-skin pipeline that generates pixel-perfect Minecraft skins from arbitrary character concepts. BLOCK decomposes the problem into (i) a \textbf{3D preview synthesis stage} driven by a large multimodal model (MLLM) with a carefully designed prompt-and-reference template, producing a consistent dual-panel (front/back) oblique-view Minecraft-style preview; and (ii) a \textbf{skin decoding stage} based on a fine-tuned FLUX.2 model that translates the preview into a skin atlas image. We further propose \textbf{EvolveLoRA}, a progressive LoRA curriculum (text-to-image $\rightarrow$ image-to-image $\rightarrow$ preview-to-skin) that initializes each phase from the previous adapter to improve stability and efficiency. BLOCK is released with all prompt templates and fine-tuned weights to support reproducible character-to-skin generation.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。