Copied from email list conversation: To all, I am wanting to move forward with creating a proposal for an implementation to extend functionality for HueRotate and Saturation for rasters. As stated by Torben the basic idea would be to take an image and convert it to the HSV or HSL color space and rotate the hue of each pixel by the same degree across the entire image, keeping the relative appearance in tact. A good example can be found here: http://crawlingrobotfortress.blogspot.ca/2013/06/optimizing-hue-rotation-operator.html In this discussion I believe there are two feasible approaches and was looking for any additional input in deciding the best way to move forward with the proposal. 1. Create a new rendering transformation option. This approach would be considered a raster-to-raster transformation and would require creating and executing a new process and operation/function and extending its use to the <Transformation> SLD. 2. Create a new normalization process for colors similar to the Contrast Enhancement normalization, extending the RasterSymbolizer elements. I feel that this option could be implemented in a couple of different ways as well. a. Extend the algorithm vendor option to include HueRotate and Saturation algorithms for Normalize within Contrast Enhancement. b. Create a new SLD element "ColorEnhancement" to follow in a similar fashion as Contrast Enhancement and the Normalize, but specifically altering the colorspace. Any additional thoughts would be appreciated, Thanks, John |