Gas Giant Texture Maps

Various freely available texture maps of solar system bodies can be found on the World Wide Web (WWW). However, most of these are of fairly low resolution (1440x720 pixels is common) and sometimes not of very high quality. Several years ago I had reached the point where this was inadequate for what I wanted to do. It soon became obvious that to get what I wanted I would have to make my own maps. This was the start of a huge amount of work much of what was spent developing and perfecting my methods. This included developing specialized software. What follows is a general description of how I would create a texture map of bodies with no solid surface (e.g. Jupiter). This is more complicated than creating texture maps of solid bodies (e.g. Ganymede). The main reason is that I almost always use 'raw' spacecraft images and not the usually colorful 'PR images'. The raw images are more accurate and using lots of them (often more than 100!) yields global coverage. For something like Ganymede I can use preexisting maps to determine the spacecraft's viewing geometry if it is not available (a common problem) but for the gas giants this is impossible because their appearance is constantly changing.

The first step is to determine how the planet reflects light, e.g. how it darkens near the terminator, limb darkening etc. This varies with wavelength.

The next step is to process the raw images to remove any artifacts and correct for uneven camera sensitivity across the image, a common problem with old images like the Voyager spacecraft images but a minor problem in e.g. the Galileo images. Once this is completed the viewing geometry needs to be determined if it is not known. This requires images where the planet's limb is visible. Images where the limb is not visible can be used if they overlap images where the limb is visible (this is usually the case). Once this is completed the images can be reprojected to simple cylindrical projection, correcting for the varying illumination across the image on the fly using information from the first step as input. This I do using software I developed for this purpose.

The spacecraft images are acquired through color filters so to get a color image three images are needed. Commonly only two of the three required images exist since once a spacecraft is close to a planet no time may be available to image it globally in three colors. Frequently green is missing. In this case the missing color channel must be synthesized from the other two. This is often done by averaging the other two but I have found this to be inaccurate so I use images obtained from a greater distance (where all three color channels are available) to determine exactly how to combine the two available color channels.

This results in lots of full color 'maps' each covering a small area of the planet. These are then mosaicked together into a global map. A section from such a map can be seen below.

The map has some seams since the varying illumination is not fully corrected for although the seams are far less prominent than they would have been without any correction during the reprojection step. Also there are some blank areas. The seams must be manually removed using mainly Photoshop although I sometimes use software I developed for partially fixing the seams. In Photoshop I mainly use gradients and the rubber stamp plus a few other tools. I work on both the full color map and the individual color channels when doing this. During this step I also fill any gaps using 'cloned' data. For fixing the poles I reproject the map to polar projection, fix the poles and then reproject the polar areas back to cylindrical projection.

The final step is to fix the color balance. The spacecraft images are usually not obtained through red, green and blue filters, for example orange, green and violet is common. This may involve replacing violet (which until this step was used instead of blue) with synthetic blue created by combining green and violet. An example of color balance experiments can be seen below. I ended up using a color balance similar (but not identical) to the average of the images at lower left and center right.

The end result is a high resolution, seamless global map with a fairly accurate to very accurate color. In Jupiter's case the result was a 5040x2520 pixel map made from almost 100 Voyager 2 images. A sample rendering can be seen below.