Images of Total Solar Eclipse in Aruba Island

MULTIPLE LAYER DIGITAL PROCESSING



From now, I reach the parameter by image processing on the computer, and following it, I did process in the darkroom. They are corona image of 1991 and 1994.
But recently an ability of our computer increases rapidly, so I decide that I make every process on desktop at this time.


This time, to making a corona7.jpg image, I did;

1

9 different exposure time transparency films taken by 6x6camera+1400mm lens

2

digitize those images using High end Scanner of Dmax4.5,RGB36bit ieach image is 4000x4000pixel

3

Using a software Photoshop4.0 on MAC computer, I adjust a tone curve of 36bit data and change to 24bit data from 36bit, then save it with an other mane.

4

Using Photoshop, 24bit 9 images open on each layer. At that time, lower image is shorter exposure image.

5

adjust a transparency and look a lower image through upper image then adjust the position of each image.

6

select white over exposure part and make a quick mask. shade off a outline of the mask using gausian blur filter and select it.then cut off it from RGB image. This process should be done from the top image to the bottom.

7

cut a white part where is saturated with signal, we can see useful part under image.

8

do this process to 8th image, we can see every useful signal parts of every images.++++(PhotoShop 4.0 desktop)

9

adjust a density of each image and check whether it is natural, and superimpose every layer to make one image.

10

use a weak unsharp mask to clear a structure.


I call this way " Multiple layer method"

This method is less a loss of gradation than previous SUM method that use unsharp mask many times, and more beautiful.
A using transparency way can make an image that has higher saturation, higher contrast and wider gradation than a using negative film way.(but in the case of transparency it is twice piece of negative images.)

1998.04.30 Shigemi Numazawa

CORONA PAGE

½ðÞVÌÐÈþ¤Î¤¹¤Ù¤Æ¤ÎºîÉʤϡ¢ ¢§atlas@atlasphoto.com¢§¤Ç´ÉÍý¤·¤Æ¤¤¤Þ¤¹¡£

Ì䤤¹ç¤ï¤»À袧jpl@jplnet.com¢§

+