As many know, working with Captain John Dennis, pilot in command of QF Flight 2901 (the Cryodon/QANTAS eclipse flight) I had the unique privilege of defining the requirements, developing the lunar shadow airborne intercept scenario, and defining in detail the executable flight plan that lead to our immersion in the umbra for 2m 30s centered on 22:44 UT.  Preliminary information about the flight, in general form, was made available pre-flight here.  Now that the flight is history, the previously embargoed Mission & Planning Definition Overview which was presented to QANTAS airlines, is now publicly available and those who were on the flight may find the report itself and attached appendices A and B of interest.  Additional post-flight information will be posted on this web site in the weeks and months ahead.

An additional benefit of occupying the seat immediately behind Captain Dennis on the flight deck, was the unfettered use of the high quality 16 inch x 28 inch window and open floor space between those two seats to observe, photograph, and digitally image the eclipse. To make the best use of this opportunity, a collaborative imaging program was devised, and a four camera "common mount" three-axis gyro stabilized platform was constructed and used during the flight.  The camera complement and imaging program was as follows:

1) SBIG ST-2000XM CCD Camera, graciously on loan from the Santa Barbara Instrument Corporation. The imaging program for the CCD camera was devised by J. Pasachoff and S. Sousa (Williams College) and was autonomously executed under computer control during the flight.  This program acquired green-filtered coronal images, executing a repetitive sequence of 10, 20, and 40 millisecond exposures with an inter-exposure cadence of 4 seconds throughout the total phase of the eclipse.

2) NIKON F5 35mm film camera, 400mm f/5.6 Vibration Reduction Lens, Fuji NPZ 800 film (push processed 2 stops to ISO 3200). This camera would provide color imagery of the mid to outer corona to about 5 solar radii.  A uniformly spaced series of exposures, with a cadence of 5.36 seconds, was executed between contacts 2 and 3. A log-normal series of exposure durations 40 milliseconds after contact 2, ramping up to 160 milliseconds at mid eclipse, then ramping back down to 40 milliseconds prior to contact 3 were taken.

3) PENTAX 35mm film camera, 500mm f/8 Tamaron Lens with 1.4x focal extender, Kodachrome K-64P film.  This camera would provide color imagery of the inner to mid corona from just above the occulted photosphere to about 1.5 solar radii.  The exposure sequence with this camera was executed synchronously with Camera 2, but the log-normal exposure durations began and ended with exposure times of 10 milliseconds.

Both 35mm film camera were autonomously controlled with Umbraphile on a Macintosh Powerbook.

4) SONY DCR TRV 900 NTSC, three CCD Digital Video Camera.  This camera, primarily, was intended as a platform guide camera streaming real-time video to the Macintosh screen display, but also was set to record the video imagery during the eclipse.

The optical axes of the four cameras were co-aligned and mounted on a common platform.  The common camera platform was inertially stabilized by two orthoganally mounted Kenyon KS-8 dual-gyro stabilizers.  The platform, balanced to point upward at 14.9 degrees (the altitude of the Sun at totality) was critically adjusted to null out any torques, and suspended by a bungie cord to allow free reaction against any change in aircraft attitude, orientation and also body jitter.  The bungie cord was secured to the wall of the flight deck above the window by means of a 70 lb. static load 6-inch diameter vacuum suction cup.  The whole rig hung immediately behind the pilot's seat.  Batteries, inverters, power supplies, computers, cabling, and other support equipment were velcroed to the back of the pilot's seat, to the flight deck floor, and anywhere else where room could be found.

In situ photographs of the equipment as integrated on the flight deck will be added later.

Notes on above Image:

1) The above image is a mosaic made from five (5) of the 40ms exposures taken as part of the CCD imaging program,

2) The camera platform pointing changed between images, thus providing a wider effective field-of-view.  The original raw images are shown in thumbnail form below:

3) The projection of the rotationally and translationally co-registered fields onto the sky is illustrated in two figures below.  The images are critically sub-pixel co-aligned on the features in the corona, not on the lunar disk. The apparent position of the moon moved significantly w.r.t. the Sun between exposures.
 

The figure on the left is a simple co-addition of the offset pointed raw images after co-registration.  As is apparent there is some amount of field overlap in portions of the wider mosaiced field.  The figure on the right illustrates the number of frames which contribute to a later done median collapse of the five images.  Black = 0, Dark Gray  = 1, Middle Gray = 2, Light Gray = 3.

4) Prior to combining the registered frames a calibration dark/bias frame was subtracted from each.  This calibration reference file was made from a median combination of 25 dark/bias frames taken on the aircraft immediately prior to the start of the eclipse imaging sequence.

5) NOTE: At this point the individual frames have *NOT* been flat-fielded, so the intra-frame photometric calibration is not correct, and they cannot acdtually be properly combined without introducing artifacts in the combination.  A flat-field reference file will very soon be provided by Steve Sousa and a proper image combination, devoid of the artifacts apparent in the topmost image can then be done.

6) Ahead of obtaining and applying a reference flat, the registered images were nonetheless median collapsed (though that will have to be done over again after flat fielding).  Because the CCD has a very large non-saturating dynamic range the information recorded cannot be shown in a single linear image.  Below, the inner regions of the corona captured in this image are shown:

To step through a linear display series, increasing the display gain by factors of 2 in each step, click on the above image. (This is a QuickTime movie; if your browser does not display it in-line then download and use a stand-alone QuickTime viewer).

7) More of the corona can be seen in a single image in a square root display.  Below are two flavors: Above inner and mid corona, and below mid and outer corona.
 

Again, the "seaming" artifacts will go away after flat-fielding.

7) More of the corona l structure can be seen in a single image in collapsing the dynamic display range by taking the Logarithm (base 10 in this case) of the image. Note, for example, the coronal streamer at the lower right which can be traced from the occulted solar limb (0.5 Rsun) out to 5 solar radii.

8) For better show the azimutally differentiated coronal features (streamers and holes) the underlying radial brightness gradient is partially subtracted.  This was done by measuring the radial brightness profile, on the azimuthal median, in all zonal regions one pixel wide, then subtracting 30% of the intensity of that radial profile at every corresponding radial zone.  30% is subjective, but the criteria is to not oversubtract any local feature to a negative value, which could then not be represented in a log display.  This "final" result is what was shown at the top of this page, and repeated here for contextual continuity:

The vertical striated "features" introduced in the corona on either side of the vertical axis of symmetry is due to display aliasing of the quantized subtracted radial profile.  This can be easily eliminated by simple 2 (or 3) point boxcar smoothing of the radial profile.  I certainly will do this after repeating this process with the data after it is flat-fielded.
 

Lastly, for now, on the CCD images, the same can be done for the 10ms and 20ms CCD images.  And then, when all are photometrically calibrated, all cam be combined onto a single final image to improve both the radial display range and the signal to noise.

As a FINAL note here, the actual data images are larger in linear scale by 4x, i.e., have 16x more pixels than shown above.  A smaller format was used just for web viewing.

Well, the hour got very late, so I won't describe this now in detail as I did the CCD images, but thought some would like a quick-look at the first results from the wider (400mm EFL on 35mm film) format film camera.  Don't worry about the color balance, that will get fixed in final processing.  The above image also has a radial median intensity profile subtracted (separately for a 3 color plane R, G, B separation, then recombined after subtraction) but shown in a linear display.  I also show this with only a 25% radial intensity filter (below, top), and also ORIGINAL image, without any such radial intensity filtering (below, bottom):
 

Here is a thumbnail gallery of the full set of "useful" coronal images obtained with this camera.  Some, unfortunately, not so well pointed, as you will see.  None-the-less, the intention is to combine all this information into a final high S/N color image in a process similar to that described for the offset dithered CCD images. To see this twice as large click on the gallery of images.