VRay or Mental Ray

作者 xdxu 发布于 . 分类 Blog, Rendering

I personally find Vray much better and regret how much time I wasted with Mental Ray. But it’s for my needs, if you have thousands of cores maybe it’s fine using Mental ray, if the pipeline is built around it. Also, MR is implemented differently in different packages. In Maya it’s implemented the worst, then Max.

Vray is the guy who does the job, and does well. Mental ray is a guy from an oriental country, which uses clever names for everything, uses complex solutions, and you really have to devote yourself to it like a pupil to a Master. You can achieve similar output with both engines mostly, but it will be much harder with MR for majority of people. I mean the learning curve. It doesn’t mean MR is worse, it’s just not as polished to everyday use, and it had to catch up for Vray, copying its materials scheme (and in some instances even surpassing it).

Mental ray has interesting solutions for speeding up renderings, but it also has some problems, and when you see them in the middle of the project, you see you can’t do much about it, and it’s almost never like this with Vray.

For example MR has (had?) terrible artifacts in specular – small dots, which just don’t go away whatever crazy sampling you use. Very slow motion blur… (I hope it was improved, but from what I’ve read it’s so deep in the core you have to rewrite so much it doesn’t make sense).

And there are many small bugs like this with MR. It drives you nuts, when you don’t have a programmer nearby, who could sort it out. And all you can do is hope and wait until a fix is released, and it takes years.

MR community is hard to deal with, people there are mostly programmers and the usual answer is “you can read the manual… it’s very good”. If you are a programmer, then it’s really good. But if you are not, then good luck with it, when you have a problem.

So at first it looks like MR is a great renderer, but indeed for work it’s quite slow to work with, not so comfortable, and takes a lot of time to learn, whereas the results you get are identical to Vray (and I think for 95% of people it will be like this, unless you really want to use some advantages MR has, like production shaders, custom shaders etc).

Mental Ray 与 VRay 之争

作者 xdxu 发布于 . 分类 Rendering, Research

    如果单就 GI 算法而言,Vray 提供的 GI 解决方案比 Mental Ray 要灵活和先进,目前而言,在直接照明阶段的一次反弹 GI 部分,Mental Ray 有 final gathering, Vray则有 irradiance map,这两种都是同一类型极其接近的算法,虽然在某些细节上处理方式不一样,提供给用户的可调手段各有差别,但在这里我们不做具体比较。

    但在次级 GI 引擎方面,Mental Ray一直沿用传统的 photon map 算法,Vray 则有自主开发的 light cache,并且也有 photon map,还有 brute force。

    photon map 是正向光线追踪的,从效率而言不及 light cache,实用性方面也不及 light cache,photon map 对光源的要求比较多,天光,自发光体都是不发射 photons 的,但这不是最要命的,photon map 算法的正向光线追踪特性使得一个场景的 GI 效果够不够好,更多取决于用户来评估其需要多少 photons,易用性不如 light cache,而且,也因为是正向光线追踪,所以阴暗角落和细小几何边缘是这个算法很难处理到位的,light cache 在这方面比 photon map 要好很多,这就是为什么 Vray 更适合室内、建筑动画的最主要原因。