Open topic with navigation
PMC is a custom mutating unbiased kernel written for GPUs. It helps resolve complex caustics and lighting (Figure 1).
Figure 1: The PMC Kernel settings
PMC Kernel Parameters
- Max. Samples - Sets the maximum number of samples per pixel before the rendering process stops. Higher numbers of samples per pixel results in cleaner renders. For quick animations and scenes with predominantly direct lighting, a low amount of samples (500-1000) will suffice. In scenes with lots of indirect lighting and mesh lights, OctaneRender® may require several thousand samples to get a clean render.
- Depth - Gives the maximum number of diffuse reflections if you set GI Mode to Diffuse.
- Depth - The maximum depth to trace rays (reflections/refractions) passing through specular objects. This prevents transparent objects from causing dark outlines in scenes.
- Ray Epsilon - The distance to offset new rays so they don't intersect with the originating geometry. If the scale of a scene is too large, precision artifacts like concentric circles appear. In that case, increasing this parameter makes these artifacts disappear.
- Filter Size - Sets the pixel size for filtering the render. This improves artifact aliasing in the render, and it also reduces noise. However, if you set the filter too high, the image becomes blurry.
- Alpha Shadows - Allows any object with transparency (Specular materials, materials with opacity settings and alpha channels) to cast a shadow instead of behaving like a solid object.
- Caustic Blur - Approximates caustics on rough surfaces, and increases or decreases caustic noise sharpness. A 0 value provides the sharpest caustics, and increasing this value increases the blurring effect that makes caustics appear soft.
- GI Clamp - Controls indirect light bounces, and provides a facility to balance out global illumination and noise produced by scene caustics. Reducing he GI Clamp lowers noise contributed by strong light passing through clear or specular objects.
- Irradiance Mode - Renders the first surface as a white Diffuse material. It's similar to Clay Mode, but it only applies to the first bounce. It disables the Bump channel and turns samples transparent when back faces block them.
- Alpha Channel - Removes the background and renders it transparent (zero alpha). This composites the render over another image without the background being present.
- Keep Environment - Allows the background to render with zero alpha, but is still visible in the final render. This works with the Alpha Channel setting, allowing more flexibility when compositing images.
- AI Light - Enables AI lighting.
- AI Light Update - Enables dynamic AI Light Update.
- Light IDs Action - Selects the action taken on selected Light IDs.
- Light IDs - Specifies the Light ID.
- Light Linking Invert - Inverts the light linking behavior for the selected Light IDs.
- Path Term. Power - Provides a system for tweaking samples/second vs. convergence (how fast noise vanishes). Increasing this value causes the kernels to keep paths shorter and spend less time on dark areas (which means they stay noisy longer), but increases samples/second. Reducing this value causes kernels to trace longer paths on average, and spend more time on dark areas. In short, high values increases the render speed, but also generates higher noise in dark areas.
- Exploration Strength - Specifies how long the kernel investigates good paths before it tries to find a new path. Low values create a noisy image, while larger values create a splotchy image.
- Direct Light Importance - Makes the kernel focus more on paths with indirect light. For example, imagine sunlight through a window that creates a bright spot on the floor. If Direct Light Importance is 1, then the kernel samples this area a lot, and it becomes clean fast. If you reduce this parameter, the kernel samples the area less and focuses on areas that are hard to render.
- Max. Rejects - Controls the render bias. Lower values produce results with more bias, but shorter render times.
- Parallel Samples - Controls how many samples are calculated in parallel. If set to a small value, OctaneRender requires less memory to store the samples state, but renders are slower. If set to a high value, then OctaneRender needs more graphics memory, but renders are faster. The change in performance depends on the scene, the GPU architecture, and the number of shader processors available on the GPU.
- Work Chunk Size - The number of work blocks (of 512k samples each) done per kernel run. Higher values increase the memory requirements on the system, but it doesn't affect memory usage. Render speed also increases.
- Toon Shadow Ambient - Ambient modifier for Toon Shadowing.
Open topic with navigation