Tuesday, July 13, 2010

The seeds are distributed uniformly.

In parallel way, around 900 stipples as starting points.

In the parallel way and based on stippling results

It takes around 1mins to get this.

Grid-based method to fix up the mosaic

1. Find the particle trails for each grid we specify. 
2. Store the particles.
3. After the process, deal with the stored particles and build a network.

Similar result from the dense lines as previous blog, but the coarse one is still not good. But this one may refuse very long strange line, which is what we try to do. But if we take a closer look at it, there are some artifacts from the grid.

Two levels for generating mosaic incrementally

1. Let the maximum length in the first few steps a smaller value which not allowing strange long lines. 
2. Store those particles from these few steps.
3. After the whole process is finished, deal with those particles from the first few steps to obtain a connected network.

Good:
The method is good for creating dense or slim patches. 
Bad:
1.But if wanting large patches, the mosaic is not good. You can see the quality from the comparisons of two results for the old man.
2.There still are some unconnected lines from this method.
3.Some disturbing shapes from the network should be removed later or transformed.
Left:
1. If we want the stylized trails? How to force the stylized trails to be honest with the field.
2. hatching
3. rendering effects

Saturday, July 10, 2010

Without CAH scheme, based on raster scanning order

A little lower quality, a few seconds faster

Directional mosaic based on CAH scheme

Next, I am going to focus on those wrong lines from the first few placement and small details. Remove them first and connect those broken lines intersected with those first wrong line in the first few steps.

An acceptable drawing


(87mins)(36mins)
First place stipples and then grow each stipple in parallel based on particle dynamics. It takes very long time to process. I have to think about the acceleration and the effects or rendering.Also there are some issues with current results.

Friday, July 9, 2010

Process each stipple in parallel

A result from 20 iterations(There is a bug for the use of direction).
Have to find a way to control the density and think about the first placement.

Thursday, July 8, 2010

After improving the quality of the trails

After improving the quality of the trails, it looks much better. Even the swirling area is not a bothering problem now.But the order of the lines is very important now.It seems useful for mosaic effects and closer to our pattern ideas.(Details?)