AN ILLUSTRATED ROADMAP FOR PROCESS DIGITAL PHOTO IMAGES
DEEP-SKY APPLICATION
Tutorial for IRIS V4.16 and upper


Pour une version en français clicker ici


Example of the Messier 57 field
Instrument: Canon EOS 10D images + Canon 200 mm f/2.8 II objective lens + Canon 2x extender (equivalent focal length of 400 mm at f/5.6)
Exposure time: 180 seconds - ISO 400 - May 20, 2004 - Castanet-Tolosan
 


 

SETTINGS (1/2)

From the file menu, open the dialog Settings. Enter the working path (your images location). For example "I:\image\". Select also the PIC or FIT option (Iris support 48-bit (3x16 bits true-colors) for the native PIC format and the standard FITS format).


 

SETTINGS (2/2)

Click on the Photo icon of the tool bar. Select your DSLR type (Canon 10D for example).

Select the linear method for convert CFA image to a true-color image (recommended for deep-sky imagery).


 

COMPUTE A MASTER OFFSET IMAGE

First select and convert the RAW files (CRW files for Canon EOS). Offset (or bias) are short exposure images taken in darkness.  Run the Decode RAW file from the Digital photo menu. A dialog box appear at the center of the screen (delete front dialog box if necessary). Drag and drop your offset RAW file from the Explorer. Here 9 files are selected.


 

CONVERT CRW  IMAGES (OR NEF IMAGES FOR NIKON USERS)

Enter the generic name of the converted images. Here the name "I" (for example, a short name). Then click on the CFA button for start the conversion.

The files I1.PIC, ...., I9.PIC are now  created in your working directory. Click the Cancel button for exit.


 

VISUALIZE ONE OF THE CFA OFFSET IMAGES

For control you can load in memory a file of your choice and display its content. Use the Load command of the File menu and choice for example the file I1.PIC.

For an optimal display adjust threshold.


 

MAKE THE OFFSET MASTER FROM INDIVIDUAL OFFSET FRAMES

From the Digital photo menu run the Make an offset command. Enter the generic name of the offset images  (here "i") and the number of images (here 9). Click OK/ The program compute now  the median stack of the images. The result is displayed on the screen.


 

STORE ON THE DISK THE MASTER FRAME (method 1)

From the file menu select the Save command and save the image onto the working directory. Select the name OFFSET (for example) (the file OFFSET.PIC is now  created in your disk).

STORE ON THE DISK THE MASTER FRAME (method 2)

Use an in-line command. Open the console dialog box by clicking the following icon of the tools bar:


Now  enter the command:

>SAVE OFFSET

Run the command with keyboard Return. Do not remove the ">". The PIC extension is not necessary.


 

COMPUTE A DARK MASTER (1/3)

Open the Decode RAW file dialog box from the Digital photo menu and drop the dark frame (i.e. thermal image of the detector). Multiple selection are possibles. A dark frame is taken in the darkness with a exposure time equivalent to the night images.

Convert the CRW  images  to a readable CFA . Give the name "I" for example. The offset individual images are replaced, but it is not a problem here. 8 dark frames are used in this example.


 

COMPUTE A DARK MASTER (2/3)

Run the Make a dark command of menu Digital photo, given  the generic name of the individual dark frames, the name of the offset master and the number of image to process. Choose the  mean stack method (recommended for prepare master dark with DSLR). The program compute the mean of the sequence I1.PIC, ..., I8.PIC and subtract the offset contribution. So, the resulting image is a pure thermal map of the detector.


 

COMPUTE A DARK MASTER (3/3)

Save the dark master of the disk. Use for example the console.
Note: for info about command line click here.


 

COMPUTE A FLAT-FIELD MASTER (1/4)

A method for obtain a flat-field consist to point the instrument toward à black screen. Were are a short exposure image (the thermal noise is negligible). Adjust the exposure for cover the 2/3 of camera dynamics (typ. 3000 ADU). Do not saturate any pixel.


 

COMPUTE A FLAT-FIELD MASTER (2/4)

Decode a sequence of flat-field frame. Use the now  well know  Decode RAW files  command of the Digital menu. We suppose 12 individual images with the name I1.PIC, I2.PIC, ... I12.PIC after decoding.

If you zoom on one of this image the Bayer matrix of the Color Filter Array is clearly visible (red, green and bleue pixels succession).


 

COMPUTE A FLAT-FIELD MASTER (3/4)

Invoque the Make a flat-field of the Digital photo menu. Enter the generic of flat-field frames, the offset master name, the normalization value (the median value of all green pixels is of 20000 ADU in this example - a typical good choice). Remember, the max value accepted by Iris is 32767.

The program compute the median of the 12 frames and subtract the offset.


 

COMPUTE A FLAT-FIELD MASTER (4/4)

Finally, save the master flat-field.

Here, the typical content of the cosme.lst file (X,Y coordinate of hot pixels). For display (and edit) this use a text editor or the Display data command of the Analysis menu.


 

FIND DEVIANT PIXELS IN THE THERMAL MAP

Some pixels are affected by a high value thermal parasitic signal. The rate signal of  "hot pixel" can be non-linear. A special procedure permit to erase bad effect of this pixels in the final result.

This cosmetic correction apply the local median to a set of pixels on the image of a CFA image, around each hot pixels  (a separate procedure is used for red, green and bleue pixels). The coordinate of the hot pixels are in an ASCII file. This file, known as cosmetic file, is then used by Iris to correct certain systematic defects during the preprocessing of the deep-sky images. The extension of the cosmetic file is .LST (list file).

For construct the cosmetic file use the in-line FIND_HOT command on the dark frame master image. So, the produced file will contain coordinates of the pixels affected by an abnormally high dark current, up to a given threshold.

First, reload the dark frame:
>LOAD DARK
then:
>FIND_HOT COSME 150

The first argument of FIND_HOT is the cosmetic file name (here the file COSME.LST is created onto the working directory). The second argument is the threshold. The command return the number of hot-pixels found up to 150 intensity value in the example. Try successive values for this threshold and examine the result in the final image. The typical useful number of  hot pixels listed is between 0 and 500.

 
 

PREPROCESSING (1/3)

Decode the RAW  images of the object, here a sequence of 10 images of the M57 field (the files name after decoding the CRW are I1.PIC, .... I10.PIC).

Were are 16-bits gray CFA images for the moment.

For control,you can display the converted images. For example with  the Load command of the File menu or from the console. Example:
>LOAD I1
>LOAD I2
>LOAD I10


If necessary, adjust the threshold. You can also use a command line, for example:
>VISU 1400 200
>VISU 1500 -100
>VISU 100 1000
    (a negative display)


 

PREPROCESSING (2/3)

Load in-memory the first image of the sequence to process. Use the mouse pointer for select a rectangular area of 200 to 300 pixels wide (drag with the right button of the mouse pressed). Choice a zone without bright stars if possible.

The program use this selected zone for compute the optimal dark to be subtract to the individual M57 images (minimization of the noise criteria).


 

PREPROCESSING (3/3)

Now, run the Preprocessing command of the Digital camera menu. Enter the master images names in the appropriate filed. Select the Optimize option for optimal dark signal removal. Click here for detail about this procedure.

Subtraction of the offset, the dark, division by the flat-field and cosmetic corrections are now  easy and automatic.

In the example we produce the "k" sequence of preprocessed image (K1.PIX, ...., K10.PIC).

 


 

CONVERT THE CFA SEQUENCE TO A TRUE COLORS SEQUENCE

Call the dialog box Sequence CFA conversion of the Digital photo menu.
Give the generic name of the preprocessed CFA image, the generic name of the true-colors sequence (48-bits internal coding), and the number of the images.

You can choice a color output or a black&white output (in this last case Iris add the red, the green and the bleue channel, and produce a gray 16-bits images).

 
 

DISPLAY A COLOR IMAGE

For example for see the first true colors image of the sequence execute the command:
>LOAD KK1

Note the 48-bits indication at the bottom of the window.


 

REGISTRATION (1/2)

The next operation is the registration of the component of the sequence. All the images are aligned relative to the first image. The more simplest method consist  to use a commun star for this (the method is appropriate here because the shift between the images  is moderate).

First, with the mouse, define a rectangle around a bright isolated and unsatured star. Use the first image of the sequence.

 

REGISTRATION (2/2)

Second, use the dialog box Deep-sky registration of the Processing menu.

The registered sequence image is KKK1.PIC, .... KKK10.PIC.

Note, from the command line you can also enter:
>
REGISTER  KK  KKK  10

 
 

STACK OF THE SEQUENCE IMAGES

We now  add the 10 aligned  images. Run the Add a sequence of the processing menu.

The result is displayed at the screen (adjust the thresholds if the image is  too white). The equivalent exposure time is of 10 x 3 minutes = 30 minutes.

 

 WHITE BALANCE ADJUSTMENT(1/3)

Store the result on the disk. Use the file menu or a command like:
>
SAVE  I

The sky background is very red because the observation is realized in severe suburban conditions. This problem is now  treated.

Define a rectangle in the image (drag with the mouse and press the right buttom). Then enter this line command (no argument):
>
BLACK
 
Iris equalize the sky background for the three colors channel, and the median background into the select zone is now of zero intensity.


 

WHITE BALANCE ADJUSTMENT(2/3)

The result appear probably very dark (remember, the sky background is near zero level after the BLACK command). OK, use the VISU command with negative value. For example:
>VISU 500 -150
 

WHITE BALANCE ADJUSTMENT(3/3)

Now  adjust correct gain of the refd/green/bleue channel. The operation consist to multiply each pixels channel by a distinct and appropriate constante for given a white aspect to a solar type star.

The typical appropriate constant for the Canon 10D/300D are:
Red : 1.96
Green: 1.00
Blue: 1.23

For this operation open the RGB balance of the Digital photo menu. Enter the coefficient, then OK.



Before the subsky procedure (full image - note the presence of severe parasitic ligh) :


After the subsky procedure :
 
 

THE FINAL TOUCH (1/3)

If the sky is no very flat use a SUBSKY procedure. For example from the Processing menu call the Background fit command, and enter the following paramètres :

Sigma = 4.0   (background automatic detection parameter)
Degree = 5     (polynomial fit degree parameter)

After the sky fitting is done, cross appear into the image. Do not panic, adjust once the visualisation threshold and the result appear.


 

THE FINAL TOUCH (2/3)

For crop the image you can use the contextual menu and the Window  command (it is necessary to define a selection rectangle before). You can use also the WIN console command and after that the two next click on the image define the cropping zone.

For orientate the image (north up), enter into the console the two commands : 
>MIRRORXY
>MIRRORX


 

THE FINAL TOUCH (3/3)

Convert the 48-bits images  to a B&W 16-bits images. Use the 48-BITS to 16-bits command (Digital photo menu). Iris compute the simple addition of the red, green and bleu channel.

A negative view  for facilitate observation of faint details. Simply, inverse the trheshold order. For example:
>VISU -250 3000

You can also add a constant to the image :
>OFFSET 250
>VISU 0 3250

 

EXPORT THE IMAGE
 
For export the image in the form of JPG, TIFF, ... file you can use the Save command of the File menu, or from the console :
>SAVEJPG M57 1

You can also export a 48 bits images toward Photoshp (PSD format) :
>SAVEPSD M57

Click here for more informations about Digital-SLR image processing

GOTO THE IRIS PAGE