Radiance Render

The ClimateStudio Radiance Render workflow supports the creation of physically-based renderings using a path tracing version of the Radiance raytracer.

Simulation Setup


To prepare a model for rendering, work your way through the six subpanels labeled 1-6 in the figure above.

2 - Sky
4 - Luminaires (optional)

If you have not done any lighting simulations in ClimateStudio, it is recommended that you go through the Lighting Model Setup video tutorial (5 minutes). The Rhino file used in the tutorial is available for download.

Once all required inputs have been populated, click the Render Window button (7), which opens a form with a camera viewport on the left and camera settings on the right.

Camera Setup

  • Metric: ClimateStudio supports the creation of both luminance and illuminance images. The luminance option shows the photometric intensity of light arriving at the viewing position (per solid angle), as a theoretical physical camera would. The illuminance option shows the total photometric flux falling on surfaces visible in the scene. This is not something a camera (or human eye) could see from the stated vantage – but is useful for displaying light levels falling onto task surfaces, e.g. when placing luminaires.

  • View: The View dropdown contains a list of standard Rhino views, as well as several non-perspective/non-parallel lens types supported by Radiance.

  • Rhino views include Top, Bottom, Left, Right, Front, Back, Perspective, and Parallel, as well as any Named Views that are part of the document. These views are shown in a navigable viewport control, which supports pan (left-click) and rotate (right-click) actions. The control is limited vis-a-vis the native Rhino viewport in two important respects. First, it does not support ghosted display modes, so all surfaces are previewed as opaque. And second, the mouse wheel does not dolly the camera location as one might expect. As a result, it may be more convenient to drive the camera using the native Rhino viewport. To do so, select the Active Rhino view (one way sync) option.

  • Radiance views include Cylindrical and Equirectangular panoramas, as well as Angular, Hemispheric, and Stereographic fisheyes. For more details consult the Radiance rpict manual pages. When selecting one of these views, a non-navigable preview will appear for the current camera location and target.

  • Finally, a Fisheye (rotating) view generates a 360-degree panorama that is remapped onto an oriented 180-degree angular fisheye in real time. Unique to ClimateStudio, this option allows changes in view direction without having to start a new rendering.

  • Location and Target: For all views, the camera location and target can be altered manually by setting XYZ coordinates or picking points in the native Rhino viewport.

  • Width/ Height (pixels): Sets the pixel resolution of the image.

  • Lens Length (mm): Applies only to perspective projections. Sets the opening angle of the camera along the image’s shortest edge, which in turn determines the extent of the scene shown in the rendering.

Once the camera is set, a rendering is invoked by pressing the Start button. ClimateStudio uses a progressive path-tracing version of the Radiance ray tracer, plus an AI denoising technology to achieve faster convergence. While a rendering is in progress, pixels are sampled until the user-specified number of samples has been reached, or the Stop button is pressed. Details on the simulation settings can be found by clicking the Settings button. In contrast to other Radiance-based lighting softwares (including DIVA-for-Rhino), ClimateStudio’s default settings are high-fidelity. You do not need to tinker with the settings to produce accurate results.

Simulation Results


Radiance renderings are high dynamic range (HDR) images. For each pixel, an HDR image contains red, green, and blue color channels similar to a traditional bitmap, plus a luminance value that determines the pixel’s absolute brightness. Because it contains absolute luminances, an HDR image can record physical lighting conditions and serve as the basis for predicting human visual comfort responses, which low-dynamic range (LDR) images cannot. Another benefit of HDR images is that exposure levels can be re-adjusted to highlight different regions of the image in post-processing.

The Image Display section contains settings responsible for converting the HDR image into the LDR representation that appears on the screen:

  • Channel determines the type data displayed. Options include RGB and Greyscale, which mimic the response of a traditional photographic film, or Falsecolor, which maps luminance values onto a color scale. Also available are two auxiliary images (Albedo and Normal), which are used by the AI denoiser.

  • Exposure and Gamma work in concert to determine the brightness of pixels under RGB or Greyscale mapping. Exposure is a scalar that adjusts the overall brightness of the displayed image, while Gamma is a parameter that describes the nonlinearity of the tone scale. A gamma of 2.2 is a reasonable default for mimicking the power responses of photographic film and the human eye.

  • Glare Pixels may be enabled to flag all pixels above a user-defined luminance (by default 2000 cd/m2) with a distinctive color.

Luminance values for individual pixels or rectangular regions may be tagged by clicking or clicking-and-dragging (respectively) over the image. For rotating fisheye projections, right clicks are used (to distinguish from rotation); otherwise, left clicks are used.

For the rotating fisheye projection, ClimateStudio calculates the daylight glare probability (DGP) for the current view and classifies it as either imperceptible, perceptible, disturbing or intolerable. Details can be found under the Annual Glare workflow.


Once the simulation has been stopped, the rendering can be saved as a ClimateStudio result file and/or exported to HDR or LDR image formats.


ClimateStudio uses Intel’s Open Image Denoise (OIDN) technology to remove noise from the raw HDR rendering, dramatically reducing the number of samples required to converge on a smooth result. The technology is built on a deep-learning convolutional neural network (CNN) trained to handle a wide range of images generated through stochastic ray tracing.

The denoiser engages after the rendering has reached one sample per pixel, and re-runs periodically while the ray trace is ongoing. ClimateStudio saves both the raw and denoised images. You can toggle between them using the Denoising checkbox during or after the run.


Fig. 1 Unfiltered Radiance rendering @ 5 samples/pixel


Fig. 2 Result after OIDN filtering

Post-Processing of Lighting Zones

For scenes with luminaires, ClimateStudio is capable of storing separately the luminous contributions of different light sources. To set this up, create lighting zones and enable their Post-Process setting. This will cause multiple contribution images to be stored in the result, allowing sources to be switched on/off or dimmed after the rendering is complete. You may also adjust the lamp color, which will change the source spectrum without altering its luminous power. Two additional channels are reserved for daylight and any remaining non-adjustable sources in the scene, so you can easily isolate zones or flip between day and night conditions. Post-render adjustments can be made freely before or after saving the result, and all image-processing features (falsecolors, pixel tags, etc.) are valid for any adjusted image.