图形学与人机交互2025arxiv.org
Terrain Diffusion：Perlin 噪声的基于扩散的继任者
Terrain Diffusion: A Diffusion-Based Successor to Perlin Noise
进阶 · 约 60 分钟读完前置知识：扩散模型基础、Perlin 噪声与程序化地形生成、MultiDiffusion
导读
这篇论文提出 Terrain Diffusion，用扩散模型生成地形，目标是作为 Perlin 噪声在无限、实时地形生成场景中的后继方案。方法采用分层流程：粗粒度行星布局→潜空间扩散生成 46km 瓦片→一致性解码器输出高精度高程图，并用 InfiniteDiffusion 扩展 MultiDiffusion 以支持无界、无缝生成。作者也承认它目前比 Perlin 噪声慢、尚未达到生产可用，控制手段有限，实时 Minecraft 集成中瓶颈在 Minecraft 自身生成逻辑。
HN 关注点：是否真能替代 Perlin 噪声，讨论焦点集中在速度、可控性、新颖性以及论文表述是否清晰。
HN 评论精选
这几条评论里：质疑 1 补充 1 亲历 2
亲历作者澄清：速度不及 Perlin 噪声、尚未生产可用、控制有限，但实时性尚可，Minecraft 集成瓶颈在游戏自身逻辑。
我是论文作者，很高兴看到它已经被发到这里。补充澄清几点：这个架构并不像 Perlin 噪声那么快；我认为在大多数应用里，若不显著增加算力，很难看到对 Perlin 噪声的重大改进。不过它也没有慢到不能实时使用：在 Minecraft 集成中（单张 RTX 3090 Ti），生成速度的瓶颈远在 Minecraft 自身的生成逻辑上。我同意它对多数任务还不是“生产可用”：主要问题是（1）地形按真实尺度生成，对多数应用来说太大；（2）用户只能控制初始高程图，而且非常粗糙。好在我认为这两个问题都能较快解决；（1）更偏地形生成特有问题，我有一些想法；（2）主要是因为没时间工程化这么多功能（而且系统已经相当密集），很多现有的 diffusion conditioning 工作可以移植过来。另外帖子标题漏掉了论文标题的关键部分：“in Infinite, Real-Time Terrain Generation”；我不指望它在其他应用中取代 Perlin 噪声，有界生成场景下手工流程仍然更好。
xandergos查看英文原文
Hi everyone, I wrote this paper. Cool to see it has been posted here already. I want to clarify some points on things other people have mentioned: - This architecture is not as fast as Perlin noise. IMO it is unlikely we will see any significant improvement on Perlin noise without a significant increase in compute, at least for most applications. Nonetheless, this system is not too slow for real-time use. In the Minecraft integration, for instance, the bottleneck in generation speed is by far Minecraft's own generation logic (on one RTX 3090 Ti). - I agree that this is not "production-ready" for most tasks. The main issue is that (1) terrain is generated at realistic scales, which are too big for most applications, and (2) the only control the user has is the initial elevation map, which is very coarse. Thankfully, I expect both of these issues to be fixed pretty quickly. (1) is more specific to terrain generation, but I have a number of ideas on how to fix it. (2) is mostly an issue simply because I did not have the time to engineer a system with this many features (and as-is, the system is quite dense). I believe a lot of existing work on diffusion conditioning could be adapted here. - The post title misses one key part of the paper title: "in Infinite, Real-Time Terrain Generation." I don't expect this to replace perlin noise in other applications. And for bounded generation, manual workflows are still superior. - The top level input is perlin noise because it is genuinely the best tool for generating terrain at continental scale. If I had more time on my hands, I would like to use some sort of plate tectonics simulator to generate that layout, but for something simple, reasonably realistic, and infinite, perlin noise is pretty much unbeatable. Even learned methods perform on-par with perlin noise at this scale because the data is so simple.
补充提出能否扩展到非地球类地形（无大气/无水/强火山活动）的问题。
我想知道，要让这类模型生成非地球地形需要什么条件。比如，用它生成没有大气、没有水循环的行星，或者像木卫一 Io 那样火山活动剧烈的行星。
Imnimo查看英文原文
I wonder what it would take to adapt a model like this to generate non-Earthlike terrain. For example, if you were using it to make planets without atmospheres and without water cycles, or planets like Io with rampant volcanism.
亲历亲历者认为速度瓶颈仍在，预生成场景面临成熟工具链竞争；期待更强的条件控制研究。
我硕士期间做过非常类似的工作。我一直没解决速度问题，从论文看他们似乎也没解决。最终，我的工作（我预计这项工作也是）只适用于预生成地形；而在这个场景里，你要面对的是非常成熟、有大量工具来操控地形生成的生态。如果作者后续研究更强的地形输出条件控制能力，会很有意思。
euleriancon查看英文原文
I worked on something very similar for my master's degree. The problem I could never solve was the speed, and from reading the paper it doesn't seem like they managed to solve that either. In the end, for my work, and I expect for this work, it is only usable for pre generated terrains and in that case you are up against very mature ecosystems with a lot of tooling to manipulate and control terrain generation. It'll be interesting to see of the authors follow up this paper with research into even stronger ability to condition and control terrain outputs.
质疑质疑新颖性：本质像 Perlin 噪声缩放后做 image-to-image 生成细节瓦片，论文表述模糊，且顶层输入有界时为何不能直接用 MultiDiffusion。
这篇论文让人很难看清他们到底做了什么。例如：他们说 MultiDiffusion 局限于有界域，所有窗口都必须落在固定有限画布内，限制了它在无界世界或连续流式环境中的适用性；然后提出 InfiniteDiffusion 解除这个限制。但另一处又说：层级从粗粒度行星模型开始，从粗略的程序化或用户提供布局生成世界基本结构；下一阶段核心潜空间扩散模型把结构变成 46km 瓦片；最后一致性解码器扩展为高保真高程图。所以真正新的东西只是稍微更好的无缝扩散图像生成。但我们又是在程序化布局的层级上生成。TLDR：拿 Perlin 噪声，缩放，然后 image-to-image 作为种子生成细节瓦片？人们早就在这么做了。不新颖。新颖之处只是让细节瓦片稍微更好看。论文把这一点讲得很绕，挺烦人。考虑到顶层输入本来就有界（比如用户输入）而非无限，我不清楚为什么不能直接用 MultiDiffusion。
noodletheworld查看英文原文
Mm. This paper makes it hard to understand what they've done. For example: > MultiDiffusion remains confined to bounded domains: all windows must lie within a fixed finite canvas, limiting its applicability to unbounded worlds or continuously streamed environments. > We introduce InfiniteDiffusion, an extension of MultiDiffusion that lifts this constraint. By reformulating the sampling process to operate over an effectively infinite domain, InfiniteDiffusion supports seamless, consistent generation at scale. …but: > The hierarchy begins with a coarse planetary model, which generates the basic structure of the world from a rough, procedural or user-provided layout. The next stage is the core latent diffusion model, which transforms that structure into realistic 46km tiles in latent space. Finally, a consistency decoder expands these latents into a high-fidelity elevation map. So, the novel thing here is slightly better seemless diffusion image gen. …but, we generate using a heirsrchy based on a procedural layout. So basocally, tldr: take perlin noise, resize it, and then image-2-image use it as a seed to generate detailed tiles? People have already been doing this. Its not novel. The novel part here is making the detailed tiles slightly nicer. Eh. :shrug: The paper obfuscates this, quite annoyingly. Its unclear to me why you cant just use multi diffusion for this, given your top level input is already bounded (eg. User input) and not infinite.
历次讨论
- 2025Hacker News 讨论帖154 分 · 42 评
同领域还值得读
像素不是小方块 (1995)
A Pixel Is Not A Little Square (1995)
这篇 1995 年的经典备忘录讨论数字图像中像素应如何理解：像素不是有面积的小方块，而是对连续图像的采样点，重建连续图像需要合适的重建滤波器。
图形学与人机交互2014 · 372 分 · 241 评进阶 · 约 20 分钟编辑精选自动化的讽刺（Ironies of Automation，1983）
Ironies of Automation (1983)
这篇 1983 年的经典论文讨论自动化（尤其过程控制）中的悖论：自动化越先进，人类操作员在异常处理和人工接管中的作用可能越关键；自动化常把常规任务拿走，留下监控、诊断和罕见故障处…
图形学与人机交互2019 · 147 分 · 38 评入门 · 约 30 分钟编辑精选GDPR 同意框中的暗黑模式
Dark patterns in GDPR consent boxes
研究 GDPR 同意框中的暗黑模式，即网站如何通过界面设计引导用户同意 Cookie 与追踪。
图形学与人机交互2020 · 328 分 · 122 评 · 引用 6进阶 · 约 30 分钟用稀疏条带在 CPU 上实现高性能 2D 图形渲染（硕士论文）
High-performance 2D graphics rendering on the CPU using sparse strips
这篇硕士论文研究如何在 CPU 上实现高性能的 2D 图形渲染，核心思路是用“稀疏条带（sparse strips）”结构来表示和光栅化路径，以降低内存占用并提升渲染速度。
图形学与人机交互2025 · 281 分 · 35 评专业 · 约 90 分钟GDPR 后的暗模式：同意弹窗及其影响
Dark patterns after the GDPR: consent pop-ups and their influence
研究 GDPR 生效后网站 consent pop-ups 中的 dark patterns，以及这些弹窗如何影响用户的同意选择。
图形学与人机交互2021 · 251 分 · 228 评进阶 · 约 60 分钟