Antialiasing with multiple cached resources
Combining multiple cached map services is one of the more powerful aspects of the ADF for ArcGIS Server. High performance mapping applications can be built off of multiple fused cached map services combined within the ADF. Lets build on the cached aerial photography example from the other day and add a cached dataset of a street network. In order to get the best performance when combining cached datasets in the Web ADF for the Microsoft .NET Framework you need to ensure that your cache configurations are identical (same scales, projection, tile size, etc) and that the map services share similar dataframe extents. When building a map cache you have the option of taking advantage of Antialiasing to smooth the edges of labels and lines for improved display quality). Of course you want to have smooth lines and labels! So what does antialiasing really do? Let's look at some comparisons of similar tiles with and without antialiasing.
The image on the left is without antialiasing enabled. The image to the right has been built using antialiasing. See the difference? The image with antialiasing of the 2000 Census Roadways dataset looks crisper, not as choppy. Antialiasing is designed to minimize jagged, blocky aliasing when representing higher resolution data at a lower resolution. So let's see what happens when we publish the data on top of our ortho cache...
Wait a minute...where did my roads go? Let's look at the same area without antialiasing...
That is what I get for implementing something without understanding the logic behind the process! The way we have implemented antialiasing from ArcGIS server is to generate tiles four times as large as the specified tile size so that we can average the RGB color of the four pixels that intersect the feature or label. So instead of a tiled pixel being completely black or not, it will be an average of the subpixels contained by the pixel. What does it average if it is just a line that is either black or not black? It uses the background color of the dataframe to compute the average pixel value. The background color is used as the transparent color by cache generation process and if you don't set the background color of your map the default is to set the background equal to 253,253,253 or almost white. So in my example I have ended up with gray roads (which don't show up well on a dark image) instead of what I originally designed as black roads. If I rebuild my map cache of my road dataset with a background color set to an average color within the image of the ortho tile (RGB: 129,129,116--pulled from this slick freeware tool) I end up with a tile that looks like this...
Try the application with each of the antialiasing options here: http://serverx.esri.com/antialiasingexamples/
Jeremy