Table Of Contents
|
Russian-Roulette is an technique that changes which type of ray is shot in scenarios with multiple ray types like reflection and refraction, it allows the renderer to shoot only one kind of ray once the importance of another ray type is deemed low enough.
For example, if a glass object is transparent and only has a very faint amount of reflection, the renderer will mostly choose refractive rays versus reflective ones. . Choosing which kind of ray to shoot (reflection or refraction) is driven by shader parameters.
Controls how aggressively Russian Roulette skips rays below the threshold. High values skip more rays resulting in faster renders that are less accurate and may look incomplete, low values shoot more of the rays necessary to complete the look of the image while rendering more slowly.
The Importance Threshold affects both the final look and the look of the earliest passes during scene update.
In the example below, the snake statue has a shader with green transmission and blue reflections, when the Importance Threshold is higher the refractions fall below the threshold and are skipped while only the reflection rays remain. Note that when Importance Threshold is higher the render time is lower at the expense of accuracy since more of the refraction rays are skipped, all examples were rendered with 500 passes.
|
|
|
| Importance Threshold: 0 Render Time: 19 seconds |
0.5 14 seconds |
1 11 seconds |
| Glycon model from ThreeDScans | ||
Controls how aggressively Russian Roulette skips rays below the threshold but only during scene updates and interactivity like camera and object movement. High values skip more rays resulting in faster interactivity while the render momentarily looks less accurate, low values shoot more of the rays necessary to complete the look of the image which will match the final resolve more closely.
Falloff does not effect the final look after scene updates have stopped, however, the Importance Threshold value is always in effect.
In the example below notice how the snake model loses its green refractive contribution during interaction and fades in after the camera settles, this is the effect a higher Falloff value makes during interactivity, you lose render accuracy temporarily but gain interactive speed.
| Falloff: 1 |
When enabled, it forces Redshift to load and process all geometry upfront before any actual rendering starts. For certain scenes this can have a positive performance impact.
By default this is disabled and Redshift loads and optimizes geometry data during the rendering process. This has the benefit of only loading/processing data that is actually visible by rays. The drawback is that, in certain cases, the CPU utilization might be less than ideal which means that geometric data processing might take longer than needed.
Controls the accuracy of the ray tracing acceleration structures.
Lower values can speed up difficult-to-ray-trace scenes such as ones that contain lots of hair, leaves or grass.
Please note, lower Max Leaf Primitive values also increase the GPU memory requirements. We, therefore, advise using lower values (such as 4, instead of the default 8) on fairly lightweight scenes or if your GPU already has plenty of VRAM.
You can diagnose the ray tracing acceleration structure memory requirements by looking in the Redshift log file, at the memory usage statistics printed at the end of each frame.
When in use, it allows Redshift to update scenes more quickly by aggressively optimizing/limiting what is processed from the scene before starting a render. This is ideal for maintaining high interactivity during Interactive Rendering but it comes at the expense of some ray tracing performance.
By default Redshift only uses Fast Preprocessing during IPR rendering to ensure that everything is appropriately updated during Final Renders.
If you notice differences between interactive rendering and final renders try setting Fast Preprocessing to None to see if that fixes the inconsistencies at the expense of responsiveness during Interactive Rendering.