Radiance ambient and penumbra test v1.0

Mark J. Stock

Introduction

Both ambient accuracy and penumbral smoothness are qualities that make a raytraced image appear more realistic. We tested Radiance's rpict renderer, with some specific paramters set, to determine the optimal settings to achieve smooth gradations in the ambient calculation and smooth-appearing penumbras.

Because the scenes that we are interested in are composed of large numbers of primitives, both near the observer and far away, and commonly fill up the computer's RAM (1 GB, normally), it was decided that we should set rpict's -aa (ambient accuracy) paramter to zero. This forces an ambient calculation to be made for every intersection of a view ray with a surface (with a few minor exceptions, such as mirrors).

In addition, because of the inability of simply using the -ds (direct sampling) parameter alone to create smooth penumbras (we have it set at 0.1), we also set the -dj (direct jitter) paramter rather high (0.6).

With these paramters set, we rendered a scene of a modified Sierpinski gasket with a single area light and no sky dome while varying the following parameters:

-ad (ambient divisions) This is how many sample rays are sent from each ambient point into the hemisphere to determine the indirect incident light. We used values of 4, 16, 64, and 256,
-ab (ambient bounces) This is how many bounces indirect light is allowed in the ambient calculation. We varied this from 1 to 2 to 3 in order to change the ultimate number of indirect samples per pixel.
-x -y (resolution) We rendered at resolution multiples of 1, 2, 3, 4, 6, and 8, (and in the -ab 2 case, 12, and 16), in order to determine when the jittering in the ambient calculation and penumbras was filtered smooth.

Ultimately, we decided to filter all images down to 256x256 pixels using pfilt with the -r 0.7 option and save them as PNG files. We based that smoothing radius on some of our previous work.

The rpict statement used is the following:

rpict -vp 3.6 2 1.5 -vd -3.6 -2 -1.1 -vh 25 -vv 25 -ds 0.1 -dj 0.6 -aa 0 -ab N -ad N -x N -y N scene.oct

Of additional importance is the use of rpict's default values for the following parameters:

-pj 0.67 -ps 4 -pt 0.05 -aw 0 -as 0

The perl script used to make the images in this table is called renderall.pl, and the source files are in this directory.

NOTE: The Radiance distribution was compiled with the -DMC option, which enables true Monte-Carlo sampling of specular highlights, penumbras, and the like. This causes no repeatable change in render times.

NOTE: All reported CPU times are from an Athlon XP 2000 processor in a system with 1 GB RAM.

Using -ab 1

	-ad 4	-ad 16	-ad 64	-ad 256
1x final resolution	0.908s, 40971 bytes	2.943s, 38974 bytes	16.406s, 35837 bytes	47.939s, 34572 bytes
2x final resolution	3.561s, 39744 bytes	11.443s, 36973 bytes	1m3.332s, 35837 bytes	2m58.037s, 33240 bytes
3x final resolution	7.908s, 38143 bytes	25.465s, 35256 bytes	2m20.682s, 32811 bytes	6m31.549s, 32213 bytes
4x final resolution	13.988s, 36796 bytes	45.076s, 34042 bytes	4m8.195s, 31686 bytes	11m18.621s, 31136 bytes
6x final resolution	31.352s, 34701 bytes	1m41.002s, 31993 bytes	9m15.141s, 30050 bytes
8x final resolution	55.623s, 33100 bytes	2m59.893s, 30625 bytes	16m19.447s, 28839 bytes

Using -ab 2

	-ad 4	-ad 16	-ad 64	-ad 256
1x final resolution	1.840s, 45050 bytes	16.354s, 42099 bytes	3m13.193s, 37102 bytes	35m9.725s, 35087 bytes
2x final resolution	7.293s, 42468 bytes	1m4.271s, 38229 bytes	12m27.719s, 34491 bytes
3x final resolution	16.371s, 39996 bytes	2m24.789s, 35744 bytes	27m36.326s, 32763 bytes
4x final resolution	28.916s, 38034 bytes	4m15.781s, 34135 bytes	48m26.859s, 31620 bytes
6x final resolution	1m5.014s, 35360 bytes	9m35.699s, 31778 bytes	108m48.000s, 29655 bytes
8x final resolution	1m55.525s, 33465 bytes	17m3.891s, 30233 bytes
12x final resolution	4m19.605s, 31083 bytes	38m14.984s, 28405 bytes
16x final resolution	7m40.945s, 29470 bytes	67m58.619s, 27108 bytes

Using -ab 3

	-ad 4	-ad 16	-ad 64
1x final resolution	3.225s, 45514 bytes	37.916s, 42016 bytes	22m35.180s, 36871 bytes
2x final resolution	12.783s, 42286 bytes	2m29.344s, 38084 bytes	86m37.094s, 34316 bytes
3x final resolution	28.756s, 39643 bytes	5m35.080s, 35639 bytes	193m8.916s, 32686 bytes
4x final resolution	50.875s, 37680 bytes	9m54.410s, 34049 bytes
6x final resolution	1m54.518s, 34981 bytes	22m15.584s, 31666 bytes
8x final resolution	3m23.006s, 33016 bytes	39m33.496s, 30033 bytes

Discussion

As is readily apparent, smooth penumbras require high oversampling. Less oversampling would be required for smoother appearance of both penumbras and normal surfaces, but that would lower the overall level of detail in the image, a comprise we chose not to make.

The effect of changing the ambient bounce setting was somewhat predictable. Noticeable differences exist in the shadow regions between otherwise-equivalent images when the paramters is changed from 1 to 2. Much less difference is seen as the paramter is raised to 3, definitely not enough difference to warrant the increase in render time. This paramter was not the main focus of this test, though. It was mainly used to try to affect the influence of -ad on the images.

The effect of the -ad is surely the most interesting result from this test. At 1x resolution, obviously, the low sampling used for the -ad=4 and 16 cases causes clear speckling of the final image; while -ad 64 and -ad=256 create images with very smooth intensity gradations. Obviously, if penumbras were not called for, one would need no more than 3x oversampling and -ad of 64.

It would be important to note that the PNG file size seems to be a fairly good rating of the overall quality of the image. Any images with file sizes less than 32000 appear to be very high quality, while file sizes under 30000 appear nearly perfect. We shall use this system as a quality paramter later.

Notice that when oversampling is doubled (1x to 2x, or 3x to 6x), four times as many view rays are shot through each pixel window, making the primary contribution of ambient rays equivalent to the next level tested (-ad 4 becomes -ad 16, etc.), but the render time for these "equivalent" renderings strongly favors the one with higher oversampling and less ambient divisions! This is more evident for the cases with more ambient bounces, so let's start with one ambient bounce, and look at the numbers:

-ab 1

Oversampling	-ad	CPU time	file size
1x	256	47.939s	34572
2x	64	1m3.332s	35837
4x	16	45.076s	34042
8x	4	55.623s	33100

As you can see, the render times are roughly equivalent, as would be expected with only one ambient bounce allowed. Additionally, the file sizes decrease slightly, most probably due to the smoother penumbral shadow.

When we look at images rendered with two ambient bounces, we should expect some differences. Whereas the first bounce should shoot an equivalent number of rays into the hemisphere, the second bounce should shoot fewer rays as -ad goes down. Theoretically, this should lead to decreased rendering times and reduced image quality.

-ab 2

Oversampling	-ad	CPU time	file size
1x	256	35m9.725s	35087
2x	64	12m27.719s	34491
4x	16	4m15.781s	34135
8x	4	1m55.525s	33465

Instead, what we see is sublinear decrease in render times (meaning that the reduced weight of the second bounce rays allows more at -ad 64 to be ignored), and a slight increase in image quality (though, again, that could be caused by the smoother penumbras). The effect of this method is to create a system by which we can force more ambient rays per pixel at the first bounce (where their visual effect is greater), and fewer for the second bounce and beyond (where visual effect is calculable, but less important).

Conclusion

For the renderings we desire---with huge geometries, smooth penumbral shadows, and proper ambient lighting---we recommend -ab 2 to capture enough interreflection to be interesting; oversampling of 8 to 12 to smooth the penumbras, and -ad 4 or 8, whichever turns out to provide sufficient smoothness on surfaces for available rendering time.

Mark J. Stock, Aerospace Engineering, The University of Michigan