| Advancements Made |
Cassidy Curtis's basic model of loose and sketchy rendering was presented in a technical sketch at SIGGRAPH '98.
Using this model as a starting point, the loose and sketchy renderer was built and several advancements were made.
The Basic Model:
Advancements Made to the Basic Model:
| The User Interface |
Using FLTK, a graphical user interface was created to house the loose and sketchy renderer. Much research was done in the area of graphical user interface development in an attempt to create an interface that was as user friendly as possible. The interface greatly extends the usability of the renderer, making it easy for someone with very little knowledge of the technique to dive right in and create a loose and sketchy render. Also, the interface was greatly needed since the addition of the parameters and keyframing would have made command line renders next to impossible.
|
|
| Additional Parameters |
First the interface was built containing only the two parameters described in the basic model of loose and sketchy rendering.
Then, one by one possible parameters were added to the renderer and tested to see if they improved the technical performance of the renderer or aesthetic appeal of the output without adding unnecessary complication to the performance of the renderer or to the user friendliness of the interface.
The parameters can be divided into the following categories:
Each of the parameters are listed below. For a more detailed description, see the user's manual or the thesis.
| Color Options |
 |
Background
|
| Line Style Options |
 |
|
| Drawing Style Options |
 |
|
| Edge Detection and Forcefield Options |
 |
Forcefield
|
| Animation |
The addition of keyframable animation was very important. Many non-photorealistic rendering techniques only allow animation as a batch process in which all of the frames are output using the same exact parameters. This doesn't give the animator enough control. By allowing the keyframing of the parameters, the animator has control over how the output changes over time. By allowing test frames to be rendered, the results can be previewed before the entire animation is output. To cycle through the frames, a frame counter was added.
To determine the inbetween values of parameters, value parameters are linearly interpolated between the keyframes and toggle parameters have the constant value of the earlier keyframe between the keyframes. For more on how the animation is handled, see the user's manual.
 |
|
| Image-Based Edge Detection |
The form of edge detection used in the basic model of loose and sketchy rendering was geometry-based. The input image had to be representative of the depth of the geometry in a 3D modeled scene.
By creating an intensity image from the input image before doing the edge detection, loose and sketchy rendering was extended to allow any type of image input.
Using images other than depth maps allow the inner edges of objects to be detected, not just the silhouette edges.
Depth maps can still be used in the renderer in the same manner as before. Because the depth map is a greyscale image, its intensity image is identical to its original image.
The only additional work done by the renderer to make this extension is to create the intensity image. The edge detection process and creation of the forcefields does not change.
The intensity image can be created using the equation intensity = (30*r + 59*g + 11*b)/100, at each pixel in the image and replacing the pixel's r, g, and b color values with the intensity.