This is an animation showing Io's weird photometric behavior:

Click to view attachment

Note added later: A significantly improved version of this animation is available later in this thread. See this message for details.

To model the brightness changes as a function of phase angle I used a global map from a JGR paper by Simonelli et al. published back in 2001. The map shows the Henyey-Greenstein g parameter and provides a measure of how strongly backscattering or forward scattering individual regions are. The bright terrain near the equator is strongly backscattering while the dark terrain is far less backscattering. The result is that as the phase angle increases the equatorial terrain appears no brighter than the terrain that appears fairly dark at low phase.

The animation should be reasonably accurate, except that the brightness of the equatorial terrain relative to darker terrain may be a bit overestimated at very low phase angles. Another problem is that the bright terrain is "overexposed" at low phase - dynamic range is frequently a problem when you want high photometric accuracy (I really need a monitor that can get as bright as the sun ;-) ). Also beware that the images are "normalized", meaning that the brightest point in each image has a fairly constant intensity throughout the animation. What this really means is that the high-phase images appear too bright. Despite these "errors" I think this animation gives a fairly good idea of how Io changes with phase angle and this certainly turned out better than I expected when I decided to use the Henyey-Greenstein parameter map I mentioned.

I might do a more realistic version of this in the future - this really is work-in-progress.

That is just amazing. Add any sort of motion from an observer's perspective and I'd insist that you'd filmed a shiny globe.

Do you take into account differences in g between the different Galileo filters in that animation?

My only other comment is that I wish I had that normal map in Celestia. My version of Tohil is pretty good (based on topo data from Paul Schenk) and some of my other mountains and paterae look good, but it only covers so much:

Click to view attachment

Not yet - it's one of the things I want to do next. I'm using the green filter values for not only green but also red and blue (I'm using synthetic blue images). The Simonelli et al. paper includes maps of g for green and violet images and also notes that red is similar to green. I don't think using separate values for violet (or in my case, blue) would make a big difference.

Regarding the normal map/DEM I'm using, it's largely fictional so I don't think the paterae look realistic - the map is created by processing the texture map in Photoshop. However, following this I also manually 'painted' in all of the mountains listed in a paper published in Icarus or JGR years ago using Galileo images as a guide. I added approximately 100 mountains this way. Inaccurate and crude but looks great.

Well, it turns out I was wrong there (I should have remembered the slightly bluish color of Io's equatorial terrain in some of the Galileo images). Here is a greatly improved version of the animation:

Click to view attachment

In this new version I'm using separate values of the Henyey-Greenstein g parameter for the blue channel. The map of g I'm using in the blue channel is an average of the green and violet maps in the Simonelli et al. paper. Variations in the value of g are more subdued in violet than green. Thus the equatorial terrain is a bit less backscattering in blue than it is in red/green. The result is that the equatorial terrain gets a bit bluish at high phase angles. This effect is visible in the Galileo color images.

Another change from the previous version of the animation is that I adjusted the "normalization". The result is that the low-phase images are no longer "overexposed" whereas the high-phase images are now brighter. This is actually less realistic but looks better and makes it easier to visually compare how the relative brightness and color of various terrain units changes with phase. Despite the brightness normalization an "opposition flash" can be seen but it should probably be brighter if you want more realism.

One problem is that the texture map I'm using was made from low-phase images. This means that directly applying g from the maps of that parameter would result in incorrect contrast and color at low phase. So I changed how my renderer handles g at low phase - it now reduces the effects of g as the phase angle approaches 0. What this means is that the animation is probably less accurate at very low phase angles than it is when the phase angle increases.