[column width=377px margin_left=120px margin_top=80px]
Figure 1: A scene with three cards renderer with a virtual camera using a virtual fish eye lens.
Simulating a Physical Camera in Computer Graphics
This thesis is about creating photo-realistic images, using physically based rendering. The process of physically based rendering can be divided into three parts, the problem of transporting the light, creating functions that determine how the light interact with the materials and a model of the camera. A lot of research has gone into the first two problems, but not much work has gone into creating more realistic camera models. The most commonly used camera model, in physically based rendering, is the perspective camera model, which is based on a pinhole camera.
The cameras photographers use, have a lens instead of a pinhole. The lens distorts the light in different ways, which sometimes creates artifacts in the image, such as the distortion caused by a fisheye lens (Figure 1). We are used to seeing images taken with modern cameras and we therefore expect to see the artifacts created by the lens. One of the most visible artifacts is depth of field, where objects near and far from the lens are out of focus, while objects in the middle are in focus (Figure 2).
Other artifacts are known as aberrations. The aberrations are sometimes divided into monochromatic aberrations and chromatic aberrations. The monochromatic aberrations appear for all light, while the chromatic aberrations only appear if we have incident light of multiple wavelengths.
Figure 2: When the image plane is positioned at position a or c, the point will be spread out on the image sensor and it will be perceived as blurry. When the image plane is positioned at position b, then the point will be perceived as being in focus.
[column width=377px margin_left=46px margin_top=80px]
In this thesis we try to make the rendered images appear more as the ones taken with a physical camera. To archive, this I have looked at how the camera works and different camera models for simulating the camera.
Figure 3: A Nikon DX AF-S Nikkor 55-200 mm 1:4-5.6 G lens.
The first model we have looked at is the perspective camera model. The main drawback of the perspective camera model, is that everything in the image is in sharp focus. This gives the images a synthetic and unrealistic look. The perspective camera model can be extended with a thin lens, which enables the model to create a depth of field effect. To create a more realistic model of the lens we look at a method described in . This method can simulate all the monochromatic aberrations, but it can’t simulate the chromatic aberrations. We have therefore extended this method, to create chromatic aberration. Finally we compare a rendered image to an image taken with a physical camera, using a Nikon lens (Figure 3).
By simulating the way light is dispersed into different colors, when it’s refracted by the lens, I have successfully rendered an image with chromatic aberration (Figure 4).
Figure 4: Chromatic aberration.
 Kolb, C., Mitchell, D., and Hanrahan, P. (1995). A realistic camera model for computer graphics. In SIGGRAPH ’95: Proceedings of the 22nd annual conference on Computer graphics and interactive techniques, pages 317–324, New York, NY, USA. ACM.