Irradiance map examples |
The following examples show the main differences between a blurry interpolation method (Least squares fit) and a non-blurry one (Delone triangulation). Notice how the images in the first column are more blurry, while the images in the second column are sharper.
CommentBlurry method (Least squares fit) Non-blurry method (Delone triangulation) The scene is a simple cube on a sphere as seen from above, lit by an HDRI map. Low hemispheric subdivs and low irradiance map rates were used intentionally so that the difference is more obvious. Both images were rendered with exactly the same irradiance map. This scene shows the ability of the Delone triangulation method to preserve detail. Notice that the shadows in the right image are sharper. Both images used the same irradiance map. A close-up of the previous scene. The irradiance map is exactly the same as for the two previous images (it was saved and then loaded from disk).
The following examples show the differences between the three sample lookup methods and more specifically, their behaviour in areas with changing sample density.
This is the test scene, the left image shows the final image and the right image shows the samples in the irradiance map (it was rendered with the Show samples option checked). The scene itself is a sphere on a plane, lit by a V-Ray area light and a little skylight. The area light had the option Store with irradiance map checked.
Test scene
The samples in the irradiance map
As one can notice, the density of the samples is quite different in the uniformly lit areas and in the shadow transition area. The following three images used exactly the same irradiance map with the Least squares fit interpolation method.
Nearest lookup method
Quad-balanced lookup method Overlapping method Density-based method
You can see the ringing artifacts (the white halo around the shadow) caused by the different sample density in the first two images. The third image, rendered with the Precalculated overlapping method is free from those artifacts. It also rendered faster than the previous two images. The last image was rendered using the Density-based method. This method yields the best result but is slower than the Overlapping method.
As a comparison, here is the same image rendered with the Delone triangulation interpolation method.
Nearest lookup method
Quad-balanced lookup method Overlapping method Density-based method
The images are nearly identical. This is because the Delone triangulation method, being a non-blurry method, is less sensitive to the samples that are being looked up, so long as the delone trianglulation can be performed successfully from them.
The following examples demonstrate the effect of the Check sample visibiliy parameter. The scene is a thin wall lit on the two sides by two V-Ray area lights with different color. Both lights had the Store with irradiance map option checked. The two images are rendered with the Medium irradiance map preset.
Check sample visibility is off
Check sample visibility is on
Notice the light leak in the first image. This happens because near the thin wall V-Ray will use samples from both sides. When Check sample visibility is turned on, V-Ray will discard the samples from the wrong side.
As a comparison, here is the same image rendered with the High irradiance map preset and Check sample visibility turned off.
High irradiance map preset, Least squares fit
High irradiance map preset, Delone triangulation
The light leak effect is negligible in the left image, and completely absent in the right one. This is because the High irradiance map preset will cause V-Ray to take additional samples at the base of the thin wall, thus decreasing the leaking effect. Using a non-blurry interpolation method (Delone triangulation) further limits this effect.
The conlcusion is that turning on Check sample visibility is only useful for low irradiance map settings. Also note that this option may not work very well for curved objects.