I've read in another post about inserting a texture that is a 16 bit image. I'm attempting to insert a floating point IR image instead of using a terrain cutout. There are some anomalies along the edges of a terrain cutout that may be avoided when using a texture insert instead. The values for my texture insert range in the 1e-4 values.
I read in the aforementioned post that GDAL will clamp those values between 0, 255 and that I should use the -scale option with gdal_translate, creating a VRT to tweak until it looks good before creating the GTiff. For us, the scale option will not work since tweaking the scale to make it 'look ok' won't do.
I can get the correct look for the texture by using the coverage=true option, because that reads the texture in as a float in ReaderWriterGDAL. However, the screen shot below (not the actual insert texture) demonstrates the problem with this. I've done my attempt to have ReaderWriterGDAL read the image as a float without the coverage tag with no success. Also, I really don't want to use the coverage option since I want these values to interpolate.
Is there a correction I can do within the shader to avoid the screen shot anomaly? Is there another method I'm not aware of where I can insert a georeferenced floating point texture without any special preprocessing?
Are you using an HDR display? A 16-bit RED channel would require an HDR-capable monitor and a full HDR rendering pipeline, which OE cannot help you with. Most displays are limited to 24-bit RGB in which case will have to remap (downsample) to an 8-bit RED channel for display.