魔兽世界如何处理盔甲交换的图形？
我一直在试图解决这个问题.. 永远，当我已经弄清楚了很多事情时.. 有一些事情我似乎无法理解。
首先，它们是否以某种方式合并纹理？例如，您可以有一个前臂，显示皮肤纹理、护腕纹理（没有添加几何体），并在其顶部放置手套（有时带有额外的几何体，据我所知，没有重叠的多边形。这是如何完成的
是否有带有和不带有额外手套多边形的单独模型部件，以及每种可能的组合的单独纹理文件？不存在。有简单的护腕纹理，并且它们以某种方式合并。
他们是否会切换材质 ID 或其他内容？
I've been trying to figure this out for .. forever and I while i've gotten my head around a good deal of it .. there are a few things I just can't seem to grasp.
First, do they merge textures somehow? For example, you can have a forearm that shows skin texture, bracer texture (no added geometry) and a glove on top of it (sometimes with extra geometry, and from what I can tell, there are no overlapping polys. How is that done?
Are there separate model parts for the forearm with and without the extra glove poly, and separate texture files for every possible combination? That seems unrealistic. From what I can tell those files don't exist. There are simple bracer textures and they are somehow merged.
Are they switching out material IDs or something?
如果你对这篇内容有疑问，欢迎到本站社区发帖提问 参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱，如果其他用户或者作者回复了您的评论，将不能在第一时间通知您！
发布评论
评论（1）
模型系统相当复杂，确实没有一个答案。他们使用多种技术，并且必须平衡多边形数量和纹理大小，以将内容大小和目标机器性能保持在合理的范围内。
一些可以开始的地方：
http://en.wikipedia.org/wiki/MPQ
< a href="http://www.zezula.net/en/mpq/stormlib.html" rel="nofollow">http://www.zezula.net/en/mpq/stormlib.html
<一个href="http://wowmapview.sourceforge.net/wowmodelview" rel="nofollow">http://wowmapview.sourceforge.net/wowmodelview
http://www.wowmodelviewer.org/
四处挖掘，也许使用一些工具，尝试找到人们尝试记录格式的地方。这是除了仔细、批判性地观看比赛之外的最好方法。还要阅读有关现代建模技术的知识，让您的头脑爆炸一点，然后一切都会开始陷入困境。
如果您查看 MPQ 内容，您可以看到模型和使用的纹理。这些将使您对正在发生的事情有一个很好的了解。即使在玩游戏时，如果您停下来思考您所看到的内容，例如交换设备时基本模型如何保持不变。每个装备似乎都有自己匹配的纹理（实际上是多个），您可以开始了解它们的基本纹理 - 多边形网格系统如何工作。您还必须记住，不同的视频设置可能有非常不同的管道（设置和渲染事件链），并且相对于您原来的问题可能会有所不同。
在正常模式下，可能很少有哇侧软件纹理混合/合成，因为，例如，即使每件的类别和种族数量的组合矩阵也会有大量的“编译”纹理来跟踪和存储。在大多数模式下，大多数事物都会与硬件混合/映射。还要记住，还有大量其他事情正在发生，例如照明、环境映射、着色器和朋友……
3D 建模已成为一个非常复杂的主题。然而，通常您可以考虑管道和源材料。在 wow 中的许多情况下，基础网格正在完成幻象的大部分工作，并且在运行时单独定义额外的 polly 并按程序“附加”到基础模型（就像前臂周围的手套，与旧的静态关键帧网格合成相反）。如果你仔细观察，你会发现开发人员和美工人员一有机会就重复使用基本模型，并创造性地在看起来像裸体的物体上使用纹理。 “物品”中隐藏着额外的信息及其相关内容，描述了幻象需要发生的情况（某些情况下它被硬编码在游戏和脚本中）。有时基本顶点只是被拉伸，有时只有纹理，有时是单独的网格。
编辑：
我真的没有回答关于“无缝”网格和缺乏装甲“堆积”的部分。对于一般情况，一个非常通用的术语和技术是“剔除”或剔除不必要的几何形状。诚实而直接的答案是，在现代世界中，这是开发人员和艺术家以及我所说的“脚本”作者（包括世界脚本和物质脚本和代码）的共同努力。好的动态模型图形有一个计划和一个架构来处理所有这些问题和OP提到的问题（并且游戏的大部分外观是决定的，特别是魔兽世界，考虑到这些常见问题）。因此，它实际上取决于艺术家的技能和理解以及引擎的设计，但从来没有真正作为绝对代码驱动的一体化解决方案完成。作为简单化系统的常见策略的一个例子，如果艺术家定义了手套较宽部分的“端盖”，以便在假定的手臂或护腕中具有某种“出血”区域，那么他们就有能力使手套能够与手臂上可能存在的任何物体配合使用，只要没有任何东西可能刺穿手套的凸缘手臂部分即可。从那里开始有很多方法，但在这一点上，即使是广义的可见性剔除步骤，前世界变换，在完全不了解世界环境或化身状态的情况下，也可以剔除看不见的波利。然后，您可以缓存并重用该模型的风格一段时间。当您了解自己的架构和系统时，您可以比这个简单的示例做得更好。但是，重要的是，美术人员知道这会发生，因为手套的末端是密封的，并且知道他们正在维持游戏的可扩展性预算。因此，这更接近于关于处理在运行时决定的多组重叠几何的 OP 问题。
艺术家/模型师也可能只是决定不做任何事情，并且不设置任何上限，因为他们知道看到手套的视角是罕见的，并且没有那么多波利的，和/或更昂贵。如果你看看更精致的装备的肩部和衣领设计，你真的可以开始看到艺术和工程的结合，就像《魔兽世界》所经历的那样。许多项圈都是扁平的，只是简单的扁平网格，每一面都有纹理，可能是相同的纹理。
The model system is quite complex, there is really no one answer. They use a mix of techniques and have to balance poly count and texture size to keep the content size and target machine performance somewhere in the realm of reasonable.
A few places to get started:
http://en.wikipedia.org/wiki/MPQ
http://www.zezula.net/en/mpq/stormlib.html
http://wowmapview.sourceforge.net/wowmodelview
http://www.wowmodelviewer.org/
Dig around, maybe use some of the tools, try to find the places people have tried to document the formats. That's the best way, other than watching the game carefully and critically. Also go read about modern modeling techniques, let you head explode a little, then it will all start to sink in.
If you look at the MPQ content, you can see the models, and the textures used. These will give you a really good idea about whats going on. And even when playing the game, if you stop and think about what you see, like how the base models remain unchanged when you swap the equipment. And how each equip piece seems to have its own matched texture (multiples really) and you can begin to get an idea how they basic texture - poly mesh system works. You will also have to keep in mind that different video settings may have very different pipes (the chain of setup and rendering events) and may work differently relative to your original question.
There is probably very little wow-side software texture blending/compositing in normal modes as, for one, the matrix of combinations of even the number of classes and races per piece would have a huge number of "compiled" textures to track and store. Most things would be blended/mapped with the hardware in most modes. Also keep in mind there are a ton of other things going on, like lighting, environment mapping, shaders and friends...
3D modeling has become a very complex subject. Generally however you can think in terms of the pipe, and the source material. In many cases in wow the base mesh is doing much of the work of the illusion, and its not a big deal anymore to have extra polly's defined separately and procedurally "affixed" to the base model at run time (like the wide part of a glove around a forearm, and as opposed to the old static keyframe mesh compositing). If you watch closely you can see the devs and the artists reusing the base models every chance they get or can, and creatively use textures on what otherwise would look like a naked body. There is extra info buried in the "item" and its associated content describing what needs to happen for the illusion (some cases its buried hard coded in the game and scripts). Sometimes a base vertex(es) are jsut extruded, sometimes there is only a texture, sometiems a seperate mesh.
Edit:
I really didnt answer the part about "seemless" meshes and lack of armor "buildup". A very general term and technique for general cases is "culling" or the scooping away of unnessary geometry. The honest and streight forward answer is that this is, in the modern world, a joint effort by the devs and the artists, as well as what I'll call the "script" writers, both of world scripts and meterial scripts and code. Good dynamic-model graphics have a plan and an architecture for dealing wiht all these issues and the ones the OP mentioned (and a great deal of the look of the game is decided, esp with WoW, with these common issues in mind). So it really depends on the skill and understanding of the artists and the design of the engine, but never is really done as an absolute code driven all-in-one solution. As an example of a common tactic with a simplistic system, If an artist defines an "end cap", of the wide part of the glove to have a sort of "bleed" area into the assumed arm or bracer, then they have the ability to make the glove work with anything that might be on the arm, so long as there is nothing that might poke through the flanging arm part of the glove. Many ways to go from there, but at this point even a generalized visibility culling step, pre-world transform, with no knowedge of the world enviroment or avatar state at all, could cull the unseen polly's. You could then cache and reuse that flavor of the model for a while. When you know your own arch and systems, you can do much better than this simplistic example. But, the important thing is that the artist knew that would happen since the end of the glove is sealed, and knew they were maintaining the games scalability budget. So this is a little closer to the OPs question about dealing with mutiple sets of overlapping gemoetry decided at run time.
The artist/modeler may also just decide to not do anything about it, and put no end cap because they understand the the viewing angle of seeing up the glove is rare-ish and not that many polly's, and/or more expensive. If you look at the shoulder and collar designs of the more elaborate gear, you can really begin to see the mix of art and engineering sometng like WoW goes through. Many of the collars are flat, just simple flat meshes with a texture on each side, possibly the same texture.