Back to the Home page.

03/29/06 Total Eclipse of the Sun from Salloum Egypt:

   Click here for many more pictures and additional information about the eclipse.

   This will take you to a high resolution animation of the eclipse, ~14 MB.

   Go here to see images of the Sun.  Images are at 30 second intervals.

   This is a short photo galley of the trip to Egypt.

 

 

 

Image Information:

Date	March 29, 2006
Optics	Nikon ED IF AF-I f/4 600 mm focal length, 150 mm aperture lens
Mount	Orion SkyView Pro equatorial mount
Camera	Nikon D-70s DSLR
Filter	None
Location	Salloum, Egypt
Exposure(s)	28 exposures from 1/6th to 1/8000th second
Processing software	FITS Liberator V 2.0, Photoshop CS2 V 9.0

 

I took the pictures with a Nikon D-70s DSLR body and Nikon ED IF AF-I f/4 600 mm focal length, 150 mm aperture lens.  These were attached to an Orion SkyView Pro equatorial mount.  The shutter of the D-70s was automatically triggered by an external intervalometer made by Harbortronics, the DigiSnap 2100.  The exposure bracketing feature was used on the D-70s to vary the exposure one frame to the next.  Below is a list of the individual images that were captured as totality unfolded.  Images were collected at ISO 200 and as raw Nikon NEF files and later converted to 16 bit TIFF images in Photoshop CS2.  The images are sorted here by their exposure time and then selected to be used in the composite.  All images used for the composite were captured within 96 seconds of 2nd contact.  The closer together all the images are the better.  This helps reduce the effects caused by the "moving moon".  I used only 28 frames out of a total of 96 unfiltered images.  This means that I shot an average of one frame every 2.5 seconds during totality.

The frames used for the composite can be summarized as follows:

# of        Exposure

frames    seconds

1            1/8000

4            1/1500

3            1/750   (one 1/1000th and two 1/750th)

3            1/250  (one 1/350th and two 1/250th)

3            1/125 (one 180th and two 1/125th)

6            1/60

3            1/30

3            1/15

2            1/6

The 16 bit TIFF frames of similar exposure were stacked in RegiStax version 3.0 shown here.

I picked a feature from each exposure that RegiStax could use for accurate registering.  This was tricky but worked out in the end.  Stacking increases the quality of each exposure by reducing the grainy appearance as shown here.  On the left is just one exposure of 1/60th second.  On the right  is a stack of 6 1/60th second frames.  The corona looks smoother although is not as pronounced here in these 8 bit JPGS as in the original 16 bit TIFFs.  After stacking, I had 9 16 bit TIFF images that I labeled as 1-8000, 1-1500, 1-0750, 1-0250, 1-0125, 1-0060, 1-0030, 1-0015 and 1-0006 for each exposure. 

Although each exposure was accurately stacked, there were registration differences between exposures.  Photoshop CS2 was used for all the remaining processing steps.  The shortest exposure was loaded first making the base layer, as shown here.

All other exposures were added as new layers. Shown here, is the 1-0060 layer.

The inverse of the image at a 50% fill was used to help better align one frame on top of the base as seen on the left.  Moving the layer pixel by pixel gives "minimal color" when the layers are registered as seen on the right.

If all of the 9 layers are successfully aligned and animated, the full dynamic range of totality can be simulated.

This animation shows a series of 9 frames of long and short exposures.  The longest exposure is 1/6th of a second.  The shortest exposure is 1/8000th second.  It is just right to show the fire of the chromosphere.  The 1/6th second stacked exposure is shown below.  The inner corona is way over exposed while the outer corona at the edges of the image is just right or perhaps even slightly under done.  Surface features on the moon can also be seen with this long of an exposure.

Once all of the layers were registered in Photoshop, layer blending and layer masking were done.  Layer blending is much easier than layer masking.  I did my layer blending first and saved a flattened file for use later in the processing sequence.  Layer blending is nothing more than adjusting the fill percentage on each layer with the blending mode in the screen setting.  The fill percent for the the 1-0006 layer was set very low, down at the 5 % level.  This decreased the brightness of the "overexposure blow-out".  The fill percent increased 5 % for each layer up to 40 % for the 1-1500 layer and 100% for the 1-8000 base layer.  This gives the corona an almost uniform brightness.  This file was used later to remove artifacts induced by layer masking and radial blurring.  The image below is what the corona looked like with layer blending only.  There is a distinct lack of detail in the corona and in the chromosphere here.

The next step involved layer masking for each layer.  With the blending mode set to normal and the fill at 100 % on each layer, the magic wand tool is used to select the "good area" outside any overexposed areas.  Shown here is the selection on the 1-0060 layer.

A mask is made for each layer from this selected zone.  Standard layer masking just uses a circular mask,  I think the magic wand selected zone is better than a circle for masking.

This picture below shows the effect of the mask.  It removes all of the image where the mask is black and lets the image through where the mask is white.  The base layer shows through the "hole" made by the mask.

Once concentric masks are made for each layer, the corona begins to take shape.  Only the "correct" exposures show for each layer.  I call these hard masks because the edges of the masks are sharp.  This was a good time to do a color balance between all the layers so that color artifacts would not be introduced.

The next step is to make the masks "soft" so that they can all blend together.  This picture below shows how a Gaussian Blur of about 50 pixels is applied to the 1-0060 mask.

The effect of the hard and soft mask is seen here for the 1-0060 layer.  The hard mask (pre-color balance) is on the left and the soft mask (post color balance) is on the right.

When all the masks have a Gaussian Blur applied, the corona begins to take shape.  Hard (left) and soft (right) masks for all the layers can be seen below.

Layer masking introduced circular artifacts due to an edge that developed on the blurred mask.  This occurred even though the mask was at 16 bit.  The only way I could get rid of it was to combine it with the previously saved layer blended file.  The blending was done in the screen mode at a fill of 50%.  The results are seen below.  Note that the corona looks smooth with some detail and the chromosphere still shows.  Keep in mind that the images here, on the web page, are 8 bit JPGs.  The real 16 bit TIFFs show details much better.

Now is when the radial blur technique is applied.

The images were further processed in Photoshop CS2 by radial blurring and self multiplication layer blending techniques.  Some of these digital image processing techniques are outlined in Sky and Telescope May 2006.

Sorry, this Page is still under construction.  Come back later for more.