Stereoscopic failure of  STEREO mission images

Good 3D images were available , (June) but the satellite separation now (August) is nearly too big for human stereoscopy. Unfortunately 3d vision will fail at solar maximum, so stereoscopic web sites like this need to accumulate stereo images from before now, because analysis will depend on mathematics rather than binocular vision in the future. (The STEREO mission is designed to only last for 2 years, but may go for 4. When the craft head towards the far side of the sun, it will be difficult and then impossible to pick up the signals). 

Placing two satellites from one launch vehicle in such different orbits required incredible skill and it would be churlish to get upset about some stereoscopic technical problems (which were predicted by NASA).

By June 2007, stereoscopy of objects at the east and west limits of the solar disc were becoming difficult, because one eye sees too far around the edge for the other eye to get a satisfactory view. Now, August, the east and west edges cannot be seen in 3D.  That is a pity because those are the edges which most clearly show prominences and prominence eruptions associated with CME's.

Alignment issues

The images from the A and B space-craft are very different in size, Ahead being bigger than Behind. This is expected, since the faster space craft (A) must orbit closer to the sun than the slower (B).

The difference in magnification introduces stereoscopic distortion not previously seen, since once the magnification is corrected, it is equivalent to taking each picture with a different focal length lens. Stereoscopic photographers avoid even slight differences in their two lenses. The result is different perspective in the two images.

The sun images are rotated slightly relative to each other and are rotated relative to the stereoscopic base-line. This can be corrected manually in Photoshop but auto alignment programs do not do such a good job.  Close observation of the images shows they are rotated before publication to make the N-S line vertical, but there is still about one degree difference between the images.

Once rotation difference between the two STEREO images is corrected, it is necessary to rotate both images until the stereo base-line is horizontal. The sun does not rotate at right angles to the ecliptic and the space craft are not quite in the ecliptic anyway, which means the stereo base rotation correction is best done manually.

Automatic alignment programs (like SPM) are causing difficulty. They take an average, by increasing the small picture (B) and decreasing the larger (A) to get a magnification match. This is not precise enough for comparison between different wave-length images.

For example, 304 Ångstrom images come from lower in the corona than 171 Ångstrom, but if stereoscopic views from the two wavelengths are combined, it is only too easy to have the 304Å features looking closer than the 171Å. This is exaggerated if correction is not made for images at different wavelengths being saved at different sizes in the archive. (This trap for young players was unexpected, since the same camera is used, but a different part of the filter system for each wavelength).

SOHO

SOHO space craft images can be combined with either of the STEREO images to reduce the baseline length. SOHO is in a stable, halo orbit at the Lagrangian point, between earth and sun, so it stays about half way between the two STEREO craft for stereoscopic base-line purposes. SOHO images are lower resolution and do not come at the high cadence of STEREO. The low frequency means SOHO is not so good for studying the details of solar flares, as this web site is attempting to do.

August 4th 2007 3D red/cyan anaglyph images

304Å, August 4th 2007

The stereoscopic distortion is now severe - Madonna distortion or bullet distortion. Instead of a spherical sun we seem to be looking at a blunt bullet coming at us. Inexperienced observers  may have difficulty fusing the stereo image at all, until they move back from the computer screen:

Or view a distorted image, in which the horizontal differences are reduced, which makes it easier to see in 3D.

There is a nice erupting prominence on the West (right) edge of the sun, but it cannot be seen in 3D because on the left eye image it is hidden behind the solar margin.

The hedge prominence on the upper left edge (10 o'clock) did not become nicely profiled from earth until 5th August, but it also cannot be shown in 3D as the right eye image has not yet profiled it.

Ng Wee Nghee of Singapore photographed this prominence in H alpha on 5th August (published on Spaceweather) NOT in 3D: Larry Alvarez, using a Coronado 90, saw it as a dinosaur eating.

Anaglyph at full resolution shows it is still possible to see the centre of the sun in 3D during August. But it is a very exaggerated, hyperstereoscopy. Even so, it does show a filament very nicely (left edge).  Exagerated depth also shows the transition zone columns. Spicules are revealed as small "sticks" in 3D but are hard to pick out on the original 2D images.
(304Å emission is at the temperature of the transition zone, hotter than the chromosphere H alpha image shown above).

September 1st 2007 (smaller scale)

Mathematical analysis

Just what mathematics the STEREO scientists plan to use for analysing 3 dimensional features is unclear. Standard photogrammetry operates on stereoscopic pairs, but once the stereoscopic base is too great for stereoscopy that becomes problematic. Difficulty can be expected with CME's (Coronal Mass Ejections) which look like clouds. The selection of homologous points is fairly easy for the human brain, even on clouds, but with separation too wide for stereoscopic fusion that might well become a problem.

This correspondence problem and other more advanced computation difficulties are discussed by Bernd Inhester of the Max Planck Institute here . You will quickly realise stereoscopic data reduction is not a trivial matter.

• For the purposes of this web site, it is a pity the space craft are going to separate so fast that full disc stereoscopy will soon be difficult.
• It is also sad that the best 3D images are occurring now, at solar minimum, while at solar maximum there will be no stereoscopy available.

Of course all this was known to NASA scientists. For example: William Thompson (Goddard Space Flight Centre) and Eric De Jong (Jet Propulsion Laboratory) who described some of the problems in "Stereo World," July/August 2006, Volume 32, Number 1

"Eventually, the separation will become too great for the eye to fuse the two images together. However the mission will be far from over at this point - in fact, it will have only just begun. In place of their stereo glasses, scientists will use sophisticated computer programs to derive three-dimensional information."

"One effect that must be taken into account when combining the two images is that each satellite will be at a different distance from the Sun, so the image sizes will not be the same. Another effect - one not normally a problem in usual stereo photography is the speed of light. It takes about eight minutes for the light from the Sun to reach Earth, somewhat less to reach STEREO-Ahead and somewhat more to reach STEREO-Behind. Because the Sun is constantly changing, the Ahead satellite must be commanded to take its image before the Behind satellite, sometimes by more than a minute, to take the difference in light travel time into account. That way, the images will appear to be taken simultaneously."

The speed of light effect has already shown up on the images turned into anaglyphs on this web site. The times in the NASA data base match when choosing images, but the times recorded on the images themselves have been slightly different.  The times quoted on this web site are the database times.

Index