Still Life with RenderMan 20
Dylan Sisson and originally published in 3D Artist Magazine
Complete Maya Project Included
"Still Life with RenderMan" rendered 100% in camera.
This tutorial will focus on lighting and look development in RenderMan for Maya. The goal is to introduce RenderMan's new RIS technology and show how the simplified workflows can deliver photorealistic results quickly. The objects in the still life were chosen specifically to demonstrate a broad range of effects, concepts, and workflows.
This tutorial also demonstrates many key features in the latest release of
. By reading this tutorial you can get a taste of some of the new features in this major release.
While RenderMan is quite capable of producing custom secondary passes for compositing, in this tutorial everything is rendered “in camera" during one pass. There are no pre-passes such as shadow maps, point clouds, or whatnot … everything from subsurface scattering, to caustics, to color grading is all done in camera within a single pass. There is no use of Nuke or Photoshop, and the only post-process is the amazing new Denoiser (Step #15).
Now that RenderMan 20 is released, you can try it out for yourself. If you're not already a RenderMan user, you can download
Free Non-Commercial RenderMan and follow along with the tutorial.
The hero of this still life is the original Melitta teapot that Martin Newell purchased in 1974 in Salt Lake City which he used as reference to create the popular CGI model, the Utah Teapot.
Maya Project Files
All of the the assets are all included to render this still life with RenderMan for Maya, including the Maya scene, textures, shaders, etc. Feel free to follow along and experiment on your own. The Maya Project is attached at the bottom of the page.
Ok, here we go ...
PART A - SETTING THE STAGE
First we'll arrange the composition of the still life and prepare for shading and lighting.
Step 1: Compose the Scene
The main subject of the still life is the teapot, which was modelled using photo references from the
. Other models were then added to the scene to show specific features & effects: a rug to show hair rendering, fruit to show subsurface scattering, a candle to show emissive volumes, etc. These object quickly created quite a clutter, so to properly direct the eye it was important to create a compelling composition. For this particular image, the Golden Ratio
was used to guide the placement of the objects. An image of the Golden Ratio was attached to the camera plane in Maya and used to quickly layout the objects.
The arrangement of the objects was guided by the golden ratio.
Step 2: Interactive Rendering
The majority of this project was created during
sessions to leverage the acceleration RIS technology provides for both lighting and look development. Interactive rendering makes lighting setup fast and natural. The special RenderMan “Lighting Panel" makes working with lights efficient … edit the intensity of groups of lights all at once and bookmark your favorite light settings as desired. From the UI, you have the option of starting an IPR session in Maya's Render View or using Pixar's own image tool “it." Interactive rendering can be a real time saver.
An interactive rendering session in RenderMan for Maya.
Step 3: Verify the Geometry
It's always a good idea to check out the geometry in your scene at the beginning of a project and make sure that there's nothing too abnormal lurking about, and an upcoming feature that will help in RenderMan 20 is a new Visualizer Integrator which allows scene geometry to be viewed as RenderMan “sees" it. You can visualize wireframes, normals, etc. Because the Visualizer is extremely lightweight it can be used to maneuver through highly complex scenes during an interactive render. Start up an interactive render and select the Visualizer Integrator from the Sampling tab in the globals and look at the scene. For our project the wireframes check out, but if there were issues with the geometry or parameterization, that would be apparent here.
Wireframe, flat shading, and ST parameterization shown in the Visualizer Integrator.
Did you know? Subdivision Surfaces
Pixar did the original research and development for subdivision surfaces, and recently Pixar has promoted this standard for subdivision surfaces with the Open Subdiv project. Open Subdiv is a set of open source libraries which have been adopted in Autodesk's Maya and across the industry. This means the subdivision surface you see in Maya will be the same as you get out of RenderMan. You may choose to work with Maya's subdivision surface primitives or simply work with a polygonal mesh and render it as a subdivision surface by attaching special attributes to the geometry, which is the technique used for this still life. RenderMan translates creases on poly meshes, so complex topologies like the candlestick can be modelled. Tip: for easy set up, try smoothing any poly mesh (by pressing “3" in Maya) and RenderMan will treat it as a subdivision surface … no extra attributes required.
PART B - LIGHTING
Fire up the interactive renderer. It's time for lighting.
Step 4: Make the Key Light
The first thing to do is create a key light to direct the viewer's eye toward the teapot. For maximum control and efficiency RenderMan has its own
, and next we'll create one from the RenderMan menu (PxrStdAreaLight) as the key light. Start up an interactive session (RenderMan > IPR Render) and select the light. To create a dramatic effect, position the light to the right of the table and focus it on the teapot as shown in the image. Now dial in the lighting for the key: increase the exposure, adjust the color temperature, narrow the light profile to tighten the illumination, and finally enable “barn doors" which provides fine control for constraining the light.
The main lighting is provided by a single key light.
Step 5: Create Fill Lights
To emphasize different elements of the composition we can use fill lights to highlight specific objects. With the interactive session running from the previous step, simply create more lights, and position them as you see fit … for example to fill in the teapot, to cast highlights on the grapes, or to add rim lighting the candlestick. In order to “paint" a specific object with light, like the teapot, an extremely tight Light Profile can be used (such as a setting of “10") which will allow the light to illuminate the teapot, but not objects adjacent to it.
Several fill lights are positioned to highlight certain areas.
Step 6: Lighting Transparent Objects
Next make sure that the wine glass renders properly. In order to achieve good results most efficiently, pay attention to the “Max Specular Depth" setting in the Sampling tab of the globals. For a ray to travel all the way through the wine glass, it must pass through at least two surfaces before exiting. In the image you can see the results from a Max Specular Depth of 2, 4, and 8. If Max Specular Depth is set incorrectly you may see black areas in the glass. Note that Max Specular Depth can be assigned on a per object basis from the RenderMan Attribute menu.
The wine glasses different Specular Max Depth: from left to right 2, 4, and 8.
Step 7: Create Caustics
For the wine glass, it's desirable for light to refract through the glass, but notice that when the Path Tracer integrator is enabled, via the Sampling tab in the globals, the shadows cast by the wine glass are solid black. Path tracers simply aren't that good at rendering caustics, so RenderMan provides a VCM integrator which uses bi-directional path tracing to efficiently resolve caustic effects for both transparent and reflective objects. Switch to the VCM integrator and that's it. In the image you can see caustics are automatically generated for the wine glass and the golden rings.
The image on the left uses path tracing, while VCM was used to create the caustics on the right.
Did you know? RIS and Reyes
It is important to be aware that RenderMan essentially has two rendering modes, RIS and Reyes, within the same software. This tutorial uses the new RIS renderer. RIS is highly-optimized for rendering global illumination, specifically for ray tracing scenes with heavy geometry, hair, volumes, and irradiance using single pass workflows. Because of differences in their architectures, it's important to be in RIS mode when following the tutorial. While the Reyes mode adds additional functionality, RIS is completely new technology that has been developed to take full advantage of modern hardware for physically based ray tracing, featuring intuitive tools, simplified workflows, and photorealistic results.
PART C - LOOK DEVELOPMENT
Introducing RenderMan's Versatile Shading System
Step 8: The Teapot - Subsurface Scattering
The main quality of the ceramic Utah Teapot is its translucent surface, which we can replicate with
. In this still life subsurface scattering was used as the base material for the teapot, teacups, grapes, pears, candles, and wooden barrel. To simulate any of these surfaces, just use RenderMan's Layered Material System to create a “LMSubsurface" material. Three controls are provided for subsurface scattering (near, middle & far). With the addition of two specular lobes, with their own independent bumps, the LMSubsurface is capable of all sorts of looks. With interactive rendering, dialling in the right look is fast and intuitive.
The teapot is shown with increasing amounts of subsurface scattering.
Step 9: Wall Rug – Rendering Maya Fur
First of all this is faux fur … no cartoon characters were harmed in the creation of this fur. RenderMan for Maya automatically renders Maya Fur and Hair, but what makes this fur special, is that it is rendered with the new
Hair shader created for Pixar production. That's correct, this is the first shader to ship from Pixar that was developed for an upcoming Pixar feature film. Marschner Hair delivers movie quality results, and setting it up is simple, just attach the Custom Shader attribute to any Maya Fur description and add Marschner Hair.
A Maya Fur system shown in Maya and rendered with RenderMan's Marschner Hair.
Step 10: Wine Barrel – Layering Materials
Here we'll use a
to create our wooden barrel bound together with metal bands. The geometry itself is simple, just a lightweight cylinder which flares out in the middle. First, a displacement shader is used to create the additional features using a texture map. Next, the base layer for the wood is created using an LMSubsurface node … and it's given just the right amount of subsurface scattering and specular highlights to simulate wood. Finally another layer is added on top of the wooden material to simulate metal and masked so the bands appear metallic.
On the top, the wooden and metal materials are shown. On the bottom the textures used.
Step 11: Candle Flame – Emissive Volumes
The candle flame was created using nested volumes and emissive geometry. To create a candle flame, first create an elongated sphere and attach a Volume Shader (pxrVolume), then on the Shading Group add a custom “Area Light Shader" attribute and connect a Mesh Light. Now the elongated sphere will behave as any other light source, even though it is a volume with distorted topology. Next place a slightly larger volume sphere around the first sphere to create the orange glow. Finally add a sphere near the bottom of the flame and attach a blue volume shader. It's a quick method of building a flame.
The geometry for the candle flame is shown on the left and the final result on the right.
PART D - CAMERA & RENDERING
Using RenderMan's Physical Camera and Depth of Field
Step 12: Get Some Focus
With the scene nearly ready to render, it's time to add depth of field. Just enable depth of field on the Maya camera and RenderMan for Maya will automatically render it. Special RenderMan Aperture controls can also be added to any camera via the attribute editor, and these provide support for features like faceted bokeh and oval shaped defocus. Tip: When dialing in depth of field, it can be useful to use the interactive renderer, and for heavy scenes it can be fastest to use the Visualizer Integrator mentioned above to set the focus. For the still life, it was important to keep the teapot in focus, and slightly blur the background.
The interactive workflow for creating depth of field.
Step 13: Create a Physical Camera
In the real world cameras are imperfect and can cause artifacts like vignetting. However, these imperfections can also give an image character. In the upcoming release, RenderMan now has a true Physical Camera which supports Vignetting, Chromatic Aberration, Lens Distortion, and Tilt Shift photography. For this still life the physical camera was an essential tool that was used to give the final render a look that is closer to a real photo. To set up the effect, simply enable the “Physical Camera" in the Features tab of the globals, and adjust the settings as you see fit.
Adjusting the physical camera during an interactive rendering session.
Step 14: Add Color Grading
RenderMan has a class of shader called “imager" which can be run at the end of a render, across the entire image. A common use for imager shaders is color grading, for those times when it is desirable to render everything in camera. This somewhat advanced step involves loading an imager shader (slo) into RenderMan for Maya using a “Frame RIB Box." In the case of the still life the imager shader was created in Pixar's Slim, and was used to push the shadows toward blue and the highlights toward orange. With the imager set up correctly, we're ready to render.
Imager shaders can be used to add color grading to images, as shown on the right.
Step 15: Render with the Denoiser
Finally before pressing render, enable “Denoise" in the Sampling tab in the globals.
was developed to address the issue of noise in physically based rendering (where images can take a long time to converge to a completely noise free image). Now renderings can use fewer samples for faster renders, and when the rendering is complete, the Denoiser is run on the image to remove any artifacts. RenderMan for Maya writes out additional information (AOVs) into a large multi-channel image (exr) for the Denoiser, which allows the Denoiser to only remove the noise. The Denoiser might just be the “make pretty button" we've all been waiting for.
On the top, before Denoise. On the bottom, after Denoise.
This tutorial is just a start. A lot of ground has been covered to introduce some of the features in RenderMan's RIS technology, and yet we've only scratched the surface. There simply wasn't enough space to thoroughly show each and every workflow, and the focus was placed instead on conveying a breadth of techniques. For further exploration, please refer to the scene files of the still life where you can closely examine the setups shown in this tutorial. RenderMan is also free for non-commercial use, so it's easy to begin exploring on your own. For more RenderMan resources please see
About the Author
Dylan Sisson combines fifteen years of experience in 3D animation and VFX with a traditional background of painting and illustration. He's created paintings, illustrations, and vinyl toys that have been shown in galleries around the world, and his independent animated shorts have won several awards. Dylan currently resides in San Francisco.