Maya 2016, new features and integration

Maya 2016 brings a list of changes to rendering with mental ray. This is the greatest amount of change in several versions and may be disorienting for experienced users. This begins the long awaited move to modern rendering in Maya workflows. You will also notice mental ray is truly a separate plugin for Maya and installs separately. This may allow updates out-of-sync with Maya releases. You can find the plugin download here: mental ray for maya

The emphasis is on simplified rendering using abstracted quality controls and global controls that are easy to find–changing how users interact with mental ray and reducing the decisions or options you might need to consider.

To take the most advantage of new techniques it’s recommended you use the MILA or MDL materials and area lights (including the newly integrated object lights) when possible. Newer techniques begin to rely on the core itself to optimize the scene (Light IS, MDL, GI Next, MIS, etc.) and the use of the GPU, when possible, to speed up rendering. Note that MDL will be covered later as materials are made available.


Keep an eye on the official mental ray blog for more information as well as MDL examples. You can find their introduction to 2016 here: mental ray for Maya 2016


Test Scene

Test Scene

Improvements to the mental ray 3.13 core include:

  • Improved Light Importance Sampling — scenes with many lights of any type can benefit. Especially when there is high contrast lighting and use of HDRI mapping to area lights. Lights must be physically plausible, i.e. quadratic falloff
  • Improved convergence of Progressive Rendering. This mode now better matches progressive off.
  • Improved texture caching performance of tiled and mipmapped file formats — .tif, .iff, and .exr
  • Useful Multiple Importance Sampling when using MDL materials and the option for MIS turned on (string option for now)
  • The ability to render scenes in Iray for Maya or mental ray using the same MDL materials (partial support for some effects in 3.13)
  • Improved hair shading and rendering
  • Deep Data for exr file rendering
  • GPU accelerated AO and GI faster than before using updated OptiX Prime libraries
  • The separation of the raylib component from the integration, meaning the possibility of updates to mental ray without changing the plugin (Dynamic raylib for integrations)
  • Memory management improved for large scenes
  • GI “Next”, which is an improvement on the ideas first shown in GI on the GPU. GI Next allows for all mental ray features like motion blur, lens shaders, more light paths, and more to be rendered using a new brute force based algorithm. This prototype is currently CPU only and only requires a single Quality control. GPU acceleration is eventual. More information will be available later. Officially here is the explanation of GI Next:
    • New Global Illumination “Next” (Prototype)

      This version of mental ray offers a new global illumination engine to compute indirect lighting efficiently, and without the restrictions of the previous experimental solution GI GPU. In particular, it supports motion blur and works well with custom shading effects, like lens distortion and depth-of-field, volume effects, and subsurface scattering. It covers all the indirect lighting features of the traditional Final Gathering approach, like “color bleed” diffuse-diffuse bounces even passing through transparent surfaces or seen in reflections. Caustics are currently not handled. In contrast to the traditional GI implementations, the new engine is utilizing a brute-force algorithm without any caching, to guarantee uncompromising and consistent quality in static images and in animations. This also makes it extremely easy to use.

      The new global illumination engine is considered a prototype in this version because it is under continuous development. It currently runs only on CPU, but is planned to take advantage of the GPU going forward. It should work reliably with typical scenes and setups, and with most third-party shaders. It can be enabled and controlled with string options and on the command line of standalone mental ray.

The render settings are the most impacted by integration changes. Others will progress over time, but this section has the most effect on users rendering images.

The settings are now separated into fewer and better defined tabs:

    • Common – this remains nearly the same except the addition of a control for the much improved color management
    • Quality – the most common controls for image quality are here. This has a few main categories:
      • Overall Quality – this controls Unified Sampling
      • Lighting Quality – this applies to Light Importance Sampling used for the lights in the scene
      • Environment Lighting Quality – this applies to the IBL lighting
      • Indirect Diffuse Quality –  Basically your indirect lighting: GI Next, GI GPU, and Final Gathering is controlled here
      • Material Quality – MILA only overall quality for all effects (glossy, scatter, etc) MDL does not use this control as it is completely in the core’s control
      • Trace depth for your ray depth including diffuse bounces
      • Geometry displacement quality for motion blurred objects

Quality - defaults

  • Scene – scene-wide controls are here. Settings that affect an entire scene or the scene output that are not related to quality are here. Some controls may be greyed out or in progress for revisions. This section includes:
    • camera settings
    • framebuffers – important – notice that framebuffers are now named framebuffers with correct naming. This may take getting used-to but it is much more flexible and powerful than before. More on this later.
    • environment shading
    • motion blur
    • scene lights
    • scene materials –
      • scene-wide scatter scale (MILA)
      • Clamp output (MILA) useful for handling fireflies and hotspots in a render
    • scene textures
    • user data

Screenshot 2015-03-26 19.47.11

  • Configuration – controls that affect scene translation and interactivity are here

Screenshot 2015-03-26 19.48.44

  • Diagnostics – controls for turning on and off features as well as image diagnostics and overrides are here

Screenshot 2015-03-26 19.49.14

You will also notice a checkbox for “Advanced” which exposes more controls. These will be covered later. Typically this provides a little more granular control or exposes a useful legacy feature for older scenes.

The creation menu has also added convenient ways to create common mental ray lights, including object lights. Selected objects will be turned into an object light in your scene. Area lights are also easily textured and sampled for rendering using a new light shader. Note that controls for the Physical Area Light are connected to the area light main controls. This is done so that users will begin to use the most obvious controls instead of those hidden as a light or material shader. New users will find this to be easiest while legacy users will need to adjust their workflow to simplify.

Create Lights menu

Create Lights menu

About David

I am a VFX artist that specializes in Lighting and Rendering. I spend a fair amount of my time supplying clients with artistic solutions as well as technology solutions. With a background in fine art and technical animation training, I strive to bridge the divide between the artist and technologist.

Posted on April 17, 2015, in Uncategorized and tagged , . Bookmark the permalink. 48 Comments.

  1. Very cool. I’ve been following the mental ray development lately.
    I still think it is very user ‘un-friendly’. What it desperately needs is maybe a new framebuffer, like vray. But since I’ve lost touch with for a couple of years now, doing work for a vray based studio, I have a lot of catching up to do.

    Is your scene file exampled above available to download anywhere?

    • Mental ray quality controls are typically much easier than other controls exposed by some renderers and allow for improvements without disrupting artist workflow. Future plans include improving mental ray viewing with GPU enhancements in the viewport as mentioned in the NVIDIA forums.

    • Unfriendly, we don’t want an iPad app either 🙂

  2. Thanks for the info 🙂

  3. Improved Light Importance Sampling — scenes with many lights of any type can benefit. Especially when there is high contrast lighting and use of HDRI mapping to area lights. Lights must be physically plausible, i.e. quadratic falloff
    ===========================================
    Any or all types of lights, including object lights, as long as the fall off is set to quadratic fall off ?

    No more what you mentioned in this blog post; https://elementalray.wordpress.com/2013/04/21/object-lights-mental-ray-3-11/ for creating object lights, you simply tell Mental Ray to make an object a light ?

    No mention if I can bake AO GPU 😦 OptiX Prime seems like a fast library for RT in Maya.
    All the new settings changes are a bit to digest, I won’t be migrating to 2016 yet, gives me some time to digest 🙂

    • Yes, all physical light types (this naturally excludes directional, makes no sense to sample it specially) Spots, point, area, object/area, and photometric are handled. Be sure that IES profiles with an obvious spotlight profile are used in a spotlight to be sampled correctly.
      At the end of this post you can see that object lights are now integrated into the menu system. With an object selected, choose “object light” from the creation menu and it will be converted.
      GPU results cannot be baked at this time.

  4. not so clever of nvidia to choose deeptile over deepscanline for the deep exr output. Nuke only supports deepscanline EXRs as of version 9.0v5 so the deep option in pretty much useless at the moment… at least for nuke users…

    • the deeptile format can be converted to scanline using imf_copy (option -s), a commandline tool that can be found in the bin directory of your mentalray installation.

      • yes of course, but that just creates the need for an additionnal annoying conversion step instead of a comp-ready exr.
        While I understand that tiled rendering is more memory efficient, please remember that just until recently (Maya 2013) mental ray rendered regular EXRs in tiled format aswell (which was annoying) and then disabled it by default, because it “is preferred to store final images that are compatible with most external tools.” (quoting the mental ray 3.10 release notes)

        So why not make it right from beginning this time, or even better, why not let the user choose deeptile or deepscanline based on their needs 🙂

      • They typically choose the format that’s most efficient. If this conflicts with what Nuke is looking to support then maybe someone needs to complain to The Foundry that they should support the better format…. In the meantime a choice would be a good compromise.

  5. Great improvements! I already had a chance to try it out, and most of it seems to work nicely in my current scenes. I found one show stopper on the road though – GI quality slider only changes quality of GI, but not indirect bounces and those are defaulting to 0. These days, single bounce GI is not a standard anymore, as at least 5 bounces are needed for believable indirect illumination.

    I could not find any settings for the new GI prototype. I’ve read about some way change them via “string options” in help files, but every time i click string options dropdown bar, my Maya immediately crashes 😦

    • You can find diffuse bounces under the raytrace depth controls. This should be the correct place. For VFX we still rely on a single bounce for 90% of things.

  6. Aha! I totally missed that. I guess it makes sense if I think about it with sort of “Arnold” mindset. I am still spoiled by Final Gather style separate setting. Thanks a bunch!

  7. Rodolfo damaggio

    These are great improvements David, I was wondering if the old still reliable mia x shader can benefit from new feature clamp output for getting rid off fireflies “mila” … Is that only if using the mila shaders?

  8. So David… the scene file above? Is that available for download somewhere? 😉

  9. Hi,
    I’ve noticed two problems with the new Physical Area Lights:
    – Don’t respect the “Use Ray Trace Shadows” / “Use Depth Map Shadows” options of the light.
    – “MILA_light” shader don’t have a “Visible” option inside, to turn the Primary Visibility of the light on/off.

    • It should always use raytrace shadows.
      For visibility flags it’s recommended to use the object lights instead. You can use the shape node flags for visibility (primary, reflection, etc)

      • I mean… now (using the mila_light shader) you can’t make a light that doesn’t cast shadows by turning off the “Use Ray Trace Shadows” option.

        Thanks for the tip on visibility!

      • It’s designed to be physical, completely, and take advantage of Light Importance Sampling. So you can’t turn off shadows.

      • you know, sometimes it’s useful to break the laws of Physics

        never mind… I’ve found a crack in the “physically based design” to do just that.
        you should:
        – turn off Ray Trace Shadows
        – turn on “Use mental ray shadow map overrides”
        – set the “Resolution”=0 in the Shadow Map Overrides section.

        please don’t “fix” that one 😉

      • Just don’t use that shader. Anything designed to be physical is going to restrict you to physical workflows. Most of the modern tools will be this way. Older tools will still be arbitrary. One previous complaint for Maya was shadows were always off. More features are moving toward physical simplicity.

      • I know… still, when illuminating a set, it’s useful to fill the room using a light that doesn’t cast shadows.
        The thing that worries me is that in the future, legacy methods are removed, we would be left with simpler, nicer, but less flexible ways to do things.

        Regarding the complaint about lights having shadows off by default, what can I say, I was the one that initiated that request 😉

        http://mayafeedback.autodesk.com/forums/160518-small-annoying-things-to-fix-in-maya-forum/suggestions/2956363-lights-activate-raytraced-shadows-by-default

        …but that doesn’t mean that it is a good idea that lights always have shadows on either.

      • A lot of this will eventually be handled through light path expressions (LPE) including contribution.

      • great!
        speaking of what, is there gonna be a LPE for shadows or that’s just not technically possible with the new way of rendering?

      • Light transport doesn’t typically involve shadows. Shadows are a non-thing, the absence of light. But you’ll be able to better separate light contributions.

      • it’s good to be able to have granular light contribution! 🙂

        I assume now mental ray is more similar to cycles or arnold renderers.
        (honestly, I don’t know about arnold, but) I do know cycles still have a shadow pass.
        would this be some kind of a hack then?

      • Arnold just uses regular passes in their commercial version. I’m not familiar with Cycles. Iray makes full use of LPE

  10. Sorry, I may be completely missing something here, but in the new list of passes, I don’t see a reflection pass or a shadow pass. Are they renamed to something else, or are those “to come”? In the meantime I have enabled the “legacy” options in prefs to access them in my current scenes.

    • Your glossy and specular passes are what your “reflection” passes are. The term “reflection” originally used was misleading since there is such a thing as “diffuse reflection”. The terms are more exact now.

      The passes also let you separate glossy (blurry) and specular (mirror-like) reflections into different passes.

      There isn’t a built in shadow pass in this first release because Light Path Expressions are built on light transport. Shadow is an absence of light, not an actual thing itself. Shadow passes for CG are also not additive and should be avoided. For shadows for plate integration I would output shadows from the mip_matte shadow shader in a render layer (as we’ve done previously)

      • Gabriel Gazzan

        I’m curious about what the Indirect Glossy/Specular LPEs will show in practice.

        Also, I wonder if an Indirect Refraction LPE shouldn’t exist also, following that line of thinking?

  11. Sadly, with all those new features also comes the penalty of only having a single netrender license per seat of AD’s new desktop subscription model. I have no idea if this was AD’s or nVidia’s decision, but it surely forces us into looking for alternatives, now. I surely hope that the last word has not been spoken in this matter…

  12. Mental ray every day goes stranger!!!!
    Where is “Passes tab” ?!!!!!!!!!!!!!!

    • Under the scene tab is where you find the new named passes. You can find this explained in the blog post about 2016. Where I discuss the scene tab as where to find options that affect the entire scene output.

    • I don’t care much about UI changes (one can get used to them really quickly)
      but I do so about well designed features that are meant to be used in a real production environment (not just in a test lab), and also about sensible default options.

      In both areas, the latest version of Mental Ray has still some ways to go, but at the same time is headed in the good direction, I think.

      • The named framebuffers have been a long time request of production companies like The Mill. Over time this will change as LPE becomes common. The catch is making production tools easy outside of a pipeline.

      • I count this one between the (many) changes I liked to find in the new version. 🙂

        I was thinking more about some things (some small, some more important) on the engine or the integration, that in other commercial renderers seem to have received more attention.
        i.e.
        – new physical lights (mila_light) with primary visibility that can’t be turn off.
        – the “prototype” status of the new GI algorithm (I hope now that mr is a plugin, it could be upgraded more frequently).
        – important production shaders like the mip_matteshadow or mip_cameramap left hidden somewhere for the user to finde them, if they are lucky enough. Also the status of the support of these shaders with the new render algorithms is uncertain).
        – new AO pass development that does not catch transparency or bump (does not see shaders at all).
        – presets for the mila_material (it can’t currently have).
        – very rudimentary volumetric shaders (parti_volume) over the years (and no sign of attention to this), for things like volumetric lights/shadows and volumetric caustics.
        – better interactive preview of (new) mr materials in VP 2.0.
        – better interactive preview of mr lights in VP 2.0.

        In spite of the above, I do appreciate a LOT of the recent changes in mental ray, aiming towards more modern approaches to rendering.
        I’m just pointing out some things I think could still be improved in mr for Maya, in order to better compete with other solutions out there.

      • I might be able to give you a hint where some of these are going in a bit. You might also want to look into the Autodesk Beta program to help guide features. You can apply at beta.autodesk.com

      • -mila_light is not complete. More options for controlling area lights has been discussed lots of places. For now object lights can give you the control you want.
        -prototype status will change over time of course, but this also means you don’t have to wait several years until it’s “done” to try it. Also good to try on finished projects and report your findings.
        -mip shaders are more likely to be integrated in a way that doesn’t expose them as shaders. As with user_data, these are more valuable as scene options and shape attributes
        -new GPU AO is built purely for speed using geometry, it’s in a similar phase as GI Next. You will notice that Optix Prime is much younger than Optix itself and is in development
        -mila base materials are your presets. 🙂 Eventually MDL will support measured data in mental ray and presets will become more…complete I guess you could say
        -volume work has been done for Bifrost. Use this, not parti_volume
        -VP 2.0 with GPU acceleration will become the home of interactive rendering, this was mentioned on the ARC forums, this will become an “always on” rendering possibility I can imagine. However, much needs to be done inside Maya itself to make this work best.

      • that’d be great! 🙂

      • thanks for the insights, really appreciated.

    • you can activate te Legacy Pass tab in preference/rendering category/ShowMayaLegacyPasses.

  13. If ‘Caustics are currently not handled’ (in the New Global Illumination “Next” (Prototype)), are we to use the old/legacy GI way to achieve caustics?

  14. ‘Caustics are currently not handled’
    Do you have any info on when Caustics might be handled by the new GI settings?

  1. Pingback: Maya 2016, new features and integration | CGNCollect

  2. Pingback: A Look at Mental Ray's New Features & Integration With Maya 2016 - Lesterbanks

  3. Pingback: Maya Tutorials | CG Info

Leave a comment