2000-02-11 Lecture 15: Colour and transparency Texture mapping - - PowerPoint PPT Presentation

Mats Nyl´ en February 11, 2000 Slide 1 of 24

2000-02-11

• Lecture 15:

– Colour and transparency – Texture mapping

• Lecture 16:

– Volume visualization – Volume visualization examples

• Lecture 17:

– Stereo rendering – Aliasing – Camera tricks VIS00

Mats Nyl´ en February 11, 2000 Slide 2 of 24

2000-02-11

• Textbook: 7.1 – 7.22
• Exercises: 7.1, 7.2, 7.5, 7.6, 7.7 and 7.9

VIS00

Mats Nyl´ en February 11, 2000 Slide 3 of 24

Transparency and α-values

In order to specify the transparency of surfaces we add an α-value to the colour model so that we have C = (R, G, B, A) where A = 1 is completely opaque and A = 0 corresponds to completely transparent. The resulting colour value is then calculated as C = As(Rs, Gs, Bs, 1) + (1 − As)Cb where s subscript refers to surface and Cb is the colour behind. VIS00

Mats Nyl´ en February 11, 2000 Slide 4 of 24

RGBA mixing results

This illustrates the fact that rendering order now matters VIS00

Mats Nyl´ en February 11, 2000 Slide 5 of 24

More about RGBA mixing - 1

300 slices rendered with α = 0.1 Note that the colours get very staurated leading to “flat” areas. VIS00

Mats Nyl´ en February 11, 2000 Slide 6 of 24

More about RGBA mixing - 2

This is also 300 slices, but now with α = 0.025. Here the saturation is less evident. VIS00

Mats Nyl´ en February 11, 2000 Slide 7 of 24

Texture Mapping

Map images onto polygons. This can be used for example to wrap a map

• nto a sphere or to increase image detail

VIS00

Mats Nyl´ en February 11, 2000 Slide 8 of 24

Volume visulizations

Volume visualization are visualization techniques applied to entire volumes

• f data. Usally wewill consider a 3D scalar datasets. Typcial examples are
• Volumetric data from MRI cameras, ultrasound etc.
• Density data from a computation, e.g., electron density in a protein.

Before the volume can be rendered the data must be classified. VIS00

Mats Nyl´ en February 11, 2000 Slide 9 of 24

Volume rendering

Methods for rendering entire volumes of data belongs to two basic classes:

• Image order methods, and,
• Object order methods

We will discuss these in turn. VIS00

Mats Nyl´ en February 11, 2000 Slide 10 of 24

Image order volume rendering

In principle these methods work by casting a ray through the volume (x, y, z) = (x0, y0, z0) + (a, b, c)t and accummulating the pixels along the line. In practice 3D scan converting algorithms that works on a voxel by voxel basis might give better results. VIS00

Mats Nyl´ en February 11, 2000 Slide 11 of 24

Object order volume rendering

When using object-order rendering we have to use either front-to-back or- dering or back-to-front ordering. We will discuss the following methods

• Splatting,
• 3D texture maps.

VIS00

Mats Nyl´ en February 11, 2000 Slide 12 of 24

Volume classification

For volumetric datasets we classify data according scalar value, or combi- nations of scalar and gradient values. After the classification we need to specify transfer functions, one each for Red, Green, Blue and Opacity. VIS00

Mats Nyl´ en February 11, 2000 Slide 13 of 24

Volumetric illumination

Volumetric illumination is fairly complicated. Several different processes contribute

• absorbtion,
• emission,
• scattering (including multiple scattering),

VIS00

Mats Nyl´ en February 11, 2000 Slide 14 of 24

Regions of interest

We like to be able to define clipping regions, one way would be to use some implicit function classification to make regions of voxels completely transparent. VIS00

Mats Nyl´ en February 11, 2000 Slide 15 of 24

Mixing volumes and geometry

We might wish to do this for various reasons, here’s an example just for fun (from the vis99 conference) VIS00

Mats Nyl´ en February 11, 2000 Slide 16 of 24

Efficient and interactive volume rendering

Various methods to increase the speed in volume rendering

• Space-leaping,
• various decimating techniques,
• splatting approximations
• etc.

VIS00

Mats Nyl´ en February 11, 2000 Slide 17 of 24

Volume rendering future

Currently hardware volume rendering is supported on two types of plat- forms

• Special purpose volume rendering chips/chipsets, typically with fairly

stringent limitations,

• High-end graphics systems (i.e.

SGI Onyx2 Infinite Reality engines) which supports 3D textures. The future of volume rendering will be along these paths. VIS00

Mats Nyl´ en February 11, 2000 Slide 18 of 24

Example Volume renderings

Several examples. VIS00

Mats Nyl´ en February 11, 2000 Slide 19 of 24

Stereo rendering

Adding a binocular parallax gives a very effective technique to simulate 3D viewing. Two basic classes of techniques exists

• time multiplexed,
• time paralell

VIS00

Mats Nyl´ en February 11, 2000 Slide 20 of 24

Aliasing and anti-aliasing

Aliasing is casued by interference between the rendering process and the raster, anti-aliasing is techniques used to minimize these effects VIS00

Mats Nyl´ en February 11, 2000 Slide 21 of 24

Anti-aliasing techniques

There are three basic classes of anti aliasing techniques

• modify the rendering primitives,
• do sub-pixel rendering (i.e. renderata higher resolutin),
• accumulate images from different cameras

VIS00

Mats Nyl´ en February 11, 2000 Slide 22 of 24

Camera Tricks

By accumulating images rendered with slightly different cameras we can simulate various effects

• focal depth
• motion blur

VIS00

Mats Nyl´ en February 11, 2000 Slide 23 of 24

Mouse Interaction

Rotating things with the mouse is convinient. Only consideration is wheter to keep the view-up constant, or orthogonal to direction of projection. VIS00

Mats Nyl´ en February 11, 2000 Slide 24 of 24

Summary and Outlook

This lecture was about some more advanced topics in computer graphics. On Tuesday, we willl talk more about data. VIS00