Load frames

Reduc reads the following image formats:
FITS int 8, int 16, int 32, IEEE 32 and IEEE 64
FITS Cube in the same formats
BMP
VFW compatible AVI files

Converts an AVI file into single BMP files.
It is a rough routine that works well with the AVI produced by the old webcams.
Recent cameras sometimes use codings that are not recognized (and will likely not be supported).
Anyway your imaging software should be able to split the AVI files into single BMP or FITS that can be loaded into Reduc.
VirtualDub and VirtualDubMod are often used by observers to do the conversion.
If your imaging software is able to create FITS files or FITS cube, do it, for many reasons it is better!

Proceed as follows to convert AVI files:
- Select the AVI file to convert then click convert

- Choose the library where you want the bmp files to be created and give a generic name for the files.
The generic name will be indexed automatically by Reduc.
- The conversion tool close automatically after the conversion and you can now select the bmp files with File/Open or File/Load Folder

Save files

Reduc is able to write the following formats:
FITS int 16, IEEE 32 and IEEE 64
FITS Cube in the same formats
BMP
RED is an internal high precision format used by Reduc

Use the commands from :
- group 1 to save only the current image
- group 2 to save the whole series of images checked in the file list.
The files will be automatically numbered from 00001 to xxxxx in the order they appear in the list.
- group 3 to save the series of images checked in the current file list into a single FITS cube of the selected format

1 Title bar displays the current frame's name
2 The name of the loaded frames are displayed in the list. The background of the list is white when the original frames are loaded, in blue when the frames come from the temporary workspace of Reduc.
3 Files selection controls of the Work buffer
The tabs allow direct access to the inetrnal buffers containing the last stack operations or the last results of interferometry operations.
4 The current frame is displayed there
5 Some stats about the current frame. Bold, coordinates and intensity of the pixel under the mouse cursor
6 Level sliders for display
7 Zoom (4x) of the cursor area. The checkbox Inv inverses video of the zoom window
Click anywhere on this zoom area to get a huge zoom of the region of interest
8 Settings of centroid detection
9 Section views of the last centroid
10 Star model
11 AutoReduc, Surface and ELI procedures
12 Calibration procedures on star trails and drifts

Files selection

All automated function except BestOf apply only to the checked files.
The frames can be selected and deselected with a click in the checkbox.
The number of files currently selected appears in grey (Sel: xxx).

The Inv button inverts the selection of frames and the checkbox left of the Inv button selects/deselects all the frames.
The percentage dropdown list indicates how many frames are currently checked. You can also use it to quickly select the top x% files of the list.

Sort
During the observations it is recommanded to acquire many frames of the target.
At the time of reduction it could be better to select only the best ones. Reduc offers two features to sort the frames (Processing / BestOf and Bestof(Vis)) in the menu bar.

BestOf does the evaluation of the whole list whatever the image is checked or unchecked. However it doesn't change the state of the check marks. After BestOf the list is sorted from the best to the worst.
BestOf (Vis) allows to do yourself the evaluation. Reduc presents a series of close-ups of the region of interest of each frame. Then you can select or unselect them manually.
BestOf (x,y) does BestOf on a fixed area.

Crop and Align
Please note that Reduc works always on copies of your frames and it never modifies your original images.

This group of commands does what it says. It remains there more for educational purpose, when you will be familiar with Reduc you will use it very rarely.

Double stars observers often take images of the small ROI (Region Of Interest) surrounding the double star to be measured.
Depending on the camera/software it is sometimes impossible to select a ROI when imaging. Working with too wide images decreases the performance of the software and might lead to problems in some cases. It is wise to crop wide images in order to keep only the area of the target.
After a click on Crop Reduc shows a dialog box that allows you to fine tune the size and the position of the region of interest, then it processes the checked images.
The effects of the adjustements are materialized by the red/white dotted cross.
Align (pixel) uses a simple translation at the pixel scale, Align (sub-pixel) does a precise sub-pixel alignment.

Align (...) reframes the images. At the difference of the simple crop the cropped area is defined relatively to the brigthest component found on the image. The effects of the adjustements are materialized by the red/white dotted cross.
NOTES:
- Reduc can combine pictures of different sizes.
- As you can set the area at the place you want relatively to the brightest component, the latter is not necessarily in the final image.

Stacks

Stack 10%...100% : sums the selected images.
After the completion the file list is switched to the internal buffer named Stack Folder, it shows 20 files named as follows:
stack 10%: sum of the first 10%
stack 20%: sum of the first 20%...
stack 100%: sum of all files

step 10%: sum of the first decile
step 20%: sum of the second decile
...

Auto Align and Stack : does simultaneously Align (sub-pixel), BestOf and Stack 10%...100%

Directed Align and Stack : It is a fine tuned alignment. Particularly useful with components of close magnitudes and often when you must deal with inequal quality frames.
Follow the instructions given in the dialog box.
First, the images are selected regarding the settings then they are processed as in Auto Align and Stack.

These operations reframe the images. At the difference of the simple crop the cropped area is defined relatively to the brigthest component found on the image. The effects of the adjustements are materialized by the red/white dotted cross.
NOTES:
- Reduc can combine pictures of different sizes.
- As you can set the area at the place you want relatively to the brightest component, the latter is not necessarily in the final image.

Two clicks for a centroid and four clicks for a measure:
Left click on the main component. Reduc draws a detection box (green square) around the star and estimates its centroid.
If this deduction is ok for you, a right click will memorize it.
Do the same thing with the secondary component.

Note: When Reduc loads the frame it automatically detects the brightest component and the detection box is already set around this star.

Many things happen with these two clicks. Before all let's familiarize with the upper band containing the centroid detection settings.

- Left is the Zoom Area.
It zooms the area under the cursor. When you click and hold down the left button it centers on the current available centroid.
The colors of the zoom area can be inverted with the Inv checkbox.
If you have difficulties to read the zoom area, a click inside the area opens an independant and parametrable zoom window.

- Just right this is the Settings Area.
We saw that the detection box is showed as a green box. Its size can be adjusted with the horizontal slider .
Adjust it to encompass the star depending on its size. Reduc estimates a new centroid every time you change the size of the detection box.
When Automatic Centering is checked, you can click very approximately (in a range about one time the box size), Reduc will find the star centroid correctly.
Checking Manual Centering forces Reduc to search the centroid inside the box (and forces you to click on the star). Automatic is the default value, you might need the Manual Centering very exceptionally for very close double stars, it avoids Reduc to jump to the brightest centroid when you want to measure the very close secondary component.

- The Section Area shows a section view of the star. The colors match the two axis colors of the zoom window. Maxi, Mean, Mini are the maximum, average and minimum pixels values inside the detection box. Current is the level value from which pixels are considered as significant for Reduc's centroid algorithms. This value is depicted by an horizontal dotted line which can be dragged to change the threshold. Moving this line forces Reduc to re-model the star and reestimate the centroid.
When the model is ok for you right click in the yellow area and Reduc will save this custom centroid instead of its own automatic estimation.
For each new star Reduc reestimates its own value. Automatic estimation by Reduc is almost always the best one for most of images.
Vertical dotted lines are the limit of the detection box.
The dropdown list sets the base value of the curves, thus it modifies the vertical scale. The value is set to Mini everytime you click on a star or a new frame is loaded.
The value Fixed sets the base at the level of the horizontal dotted line (Current value).

- The Model Area shows the star as Reduc "sees" it for the computation.

Reduction window

The first right click on a star opened a window named Reduction.
The menu item Window/Reduction opens also this window.

1 Camera selection list. The last camera used is recalled at the next start of Reduc.
2 Pixels size. Pixel size is automatically selected when changing camera. The camera list can be customized for your needs by a click on the [...] button.
3 Quadrants orientation. Switch from one to others orientations by clicking the buttons labelled N,S,E,W in this section. Give here a rough image orientation. The important thing is the succession of the quadrants. Reduc saves this setting from one session to the other.
4 Coordinates and intensity of the current centroid.
5 Current mode (Calibration mode or Measurement mode)
6 D = image inclination regarding the celestial equator. Automatically computed in Calibration mode.
7 E = pixel sampling. Automatically computed in Calibration mode.
8 q= mean position angle of the double star. Automatically computed in Measurement mode.
9 r = mean double star separation in arcsec. Automatically computed in Measurement mode.
10 dM is a difference of magnitude estimation.
11 s= standard deviation for theta and rho (these values aren't significative in Calibration mode)
12 Tabs of the reduction sheets.
13 Menu options :
- Sort = Many sorting options of the reduction sheet and an automatic rejection option are available.
- Add to log = save the reductions into a log file (see Publishing)
- Clear = clear both reduction sheets (a confirmation message prevents an accidental click !). Must be used every time you change of target. Constant value (camera, pre-orientation, D, E) are preserved.
14 Disp Button: opens a window showing the spatial distribution of the relative positions of the secondary component. Works obviously when a series of measures is available in the reduction sheet.

From a calibration star :
- Reduc computes image orientation
- Reduc computes the sampling

You must give :
- Characteristics of the calibration star

HOW TO PROVIDE CALIBRATION PARAMETERS


Select Calibration to switch Reduc to Calibration mode
Provide Theta and Rho of the calibration star (Reduc recomputes on the fly the new parameters as soon as you introduce new values).

When using several calibration stars for an imaging session (highly recommended), repeat the entire procedure of calibration for every calibration stars and compute manually the means of D,E. These means will be reintroduced after having switched to Measure mode.

Calibration with a calibration star

It is the same procedure than a simple reduction.
- Load the set of images
- Reduce it manually or automatically

When the reduction is completed :
- Select Calibration in the reduction sheet
- Type the theta and rho values of the calibration star

The parameters are computed and introduced in D and E.

Don't forget to switch to Measure mode before measuring !

NOTE : The second reduction sheet (Reduced tab) can't be displayed when in Calibration mode

Instrumental constants already known

You can enter D,E at any time when you are in Measure mode.
Reduc recomputes all that is necessary immediately when you enter new values.

Others calibrations

Reduc implements also two methods to calibrate the orientation on star trails or drift images.

Manual measurements

After the calibration the precise image orientation and the pixel sampling are known. So you can switch to Measurement mode.
If you got D,E by other means than by Reduc calibration, you can provide them at any time in the corresponding areas.
The Measurement session is the easiest phase of Reduc's learning !!!
- Load the set of images
- Sort it
- Click on the main component of the first frame to obtain its centroid, if detection is ok, right click to memorize.
NOTE that usually Reduc detects automatically the first component when loading, so if detection is ok you can right click directly.
- Repeat the same for the secondary component (click,click !)
- Select the next frame in the list and so on ... you stop when you want.

Each frame processed causes a complete refresh of the reduction. The results are displayed in the Reduction window. Both reductions sheets can be displayed in Measurement mode.

Remember to press Clear before reducing a new target.

AutoReduc: Automated Reductions

This method does the reduction of a set of frames automatically. It mimics a manual measurement on each frame, it reduces the frames independently.

Select a good frame in the file list (BestOf might help)
- Click on the brightest component and adjust if necessary the size of the search box like in manual reduction
- Click on 'Comp A'

- And voila ! After a while the set of frames is reduced

Depending on the seeing and the difference of magnitude, the brightest component may change from one frame to the other.
If strict match is checked, Reduc uses only the frames where the rank of the magnitudes matches your choice (mag compA < mag compB).
After the reduction the frames kept for the reduction are checked in the file list, the other frames are unchecked.

If strict match is unchecked, Reduc uses all selected frames and defines the right orientation when the rank of magnitudes is inverted.

After the completion the Disp. window shows the spatial distribution of the B component. The black square is the size of one pixel.

ELI: Easy Lucky Imaging

The basic principle of lucky imaging is to take many very short exposure frames of the target with the hope that a significant number (several %) of frames will be at least very sharp, at best diffraction limited. After a severe sorting the combination of the sharpest images gives a near diffraction-limited image.
ELI (Easy Lucky Imaging) works in this way but furthermore it uses an original algorithm to find the best correlated images. The result is that a much larger number of frames is used to provide a high SNR image.
The resulting image can be measured either by Surface or by the click-click procedure.
ELI can be used with severely speckled frames and very low SNR frames.

Set of short exposure frames

ELI result

The procedure is similar to AutoReduc.

First, find an image where you can locate approximatively the both components. The goal is just to identify the components.
If you can not find easily a good frame, one of the following methods might be helpful:
- run Auto Align and Stack. Close magnitudes components can provide a false triple star, do not worry about the possible 180° ambiguity
- run an autocorrelation if the images are mostly speckled. Do not worry about the 180° ambiguity

- Click on the brightest component and adjust if necessary the size of the search box like in manual reduction
- Click on 'comp A'

- Using the same way, select the secondary component and adjust if necessary the size of the search box.
- Click on 'comp B'

During the procedure, Reduc might seem to freeze if you use several thousands of frames. Do not panic, let it run !

After the completion of ELI:
- the final image is displayed
- the file list is 'best of' ordered and shows only the remaining frames selected by ELI
- the buffer Stack contains the combinations of the remaining frames
- the Disp. window shows the spatial distribution of the B component on the individual frames if they were taken independently.
- you can measure the final images with Surface or with the manual click-click
After the completion of ELI the frames are aligned so:
- you can regenerate the ELI image at any moment (Processing/Stack 10%...100%)
- you can generate a sub-ELI image at any moment (select a new percentage or select manually in the list then Processing/Stack 10%...100%)

SURFACE : Adjustment of a tridimensionnal surface

This feature uses the Surface algorithm especially designed to measure couples where the light of stars is overlapping. It was developed by Guy Morlet and Pierre Bacchus to measure the images acquired on the 50cm refractor of the observatory of Nice. It is integrated in Reduc with the authorization and the courtesy of the authors.

Surface runs at full power on images with a high signal to noise ratio. Therefore it is not usable in the automatic reductions.
After having shift-and-added the image, the measure is made in two steps:
1 - click on the brightest pixel of the main star and click 'comp A' button
2 - click on the brightest pixel of the secondary star and click the 'comp B' button.
If the secondary star doesn't show a visible peak, click on what seems to be the center of the star.
At the end of the calculations, a window with the internal elements of Surface's reduction appears.The measurement is taken automatically into account in the reduction window.

Below is a practical case somewhat complicated in order to show all the possibilities of Surface combined with other functions of Reduc:

Most of the time Surface does the reduction and we will see that just after. Here Reduc returns an error message (6). Surface cannot compute the positions of the components.
The probable reason can be:
- the stars are too tight and do not offer enough surface (under-sampling)
- the signal is too weak
- it is simply impossible to measure

Here the stars are too 'small' to allow Surface to work.
Never mind, we have a very useful feature in Reduc and we will resample the image with QuadPx (7).

Our stars now are of reasonable size.
It remains only to restart with the operations already described: identify the main and secondary components.

Math Image

The Math Image option allows you to see how Surface has modelized the stars.

Look at this example at right with an image of STF 644.

- left the original image
- right the mathematical model after by Surface's computation

(16) (17)

Resampling with QuadPx

QuadPx resamples the displayed image and enlarges it by a factor two while keeping constant the distribution of the light by unit of area.
This method of enlarging is perfectly adapted to the measurement by Reduc because it works on this distribution.
If needed QuadPx can be applied twice on the same image.

Note for old users of Reduc: since V3.71, Reduc takes automatically into account the scale change and adjusts the reduction parameter with QuadPixed images.

The command QuadPx Series does QuadPx on the Work buffer

This chapter and the corresponding features of Reduc are dedicated to speckle interferometry technics.

Warning

The speckle interferometry has long been used by professionals to restore the resolution capability of large telescopes. It can also be effective on smaller instruments (d ~ 30-40 cm) for measuring very close double stars.
It requires the use of very short exposure times to freeze the seeing and the use a (very) long focal length. The Airy disk must spans several pixels.
Moreover better results are obtained with filtered images.
In this manual I assume that the user is familiar with these techniques and that he is able to define if these methods are suited to his own situation.

Autocorrelation

Obtaining autocorrelogram with Reduc is extremely simple. Just click on Autocorrelation!

The algorithms use the fast Fourier transform (FFT). It requires square images whose dimensions are a power of 2. If your images do not follow this requirement, Reduc handles resizing before creating the autocorrelogram.

Click OK on the error message and then choose a size compatible with your pictures in the next dialog box and then Ok

Reduc then generates the autocorrelogram.

In the final computation, the file list shows the content of the interferometry buffer with 10 solution files named AC0_xxx to AC9_xxx.

AC0 is the straight autocorrelogram. Autocorrelation peaks generally are embedded in noise and difficult to measure.

Cross-correlation

The cross-correlation is another way to interpret the images of speckles. It has a main advantage because it allows to remove the ambiguity of 180 °

Transactions are strictly the same as for the autocorrelation described in the previous chapter.

Load the images then click Cross-Correlation.

Again the FFT is used and the warning about the dimensions of the images may appear => proceed as in the previous chapter.

It remains only to measure the peak corresponding to the secondary component. It is the brightest peak (*)

(*) Actually the position of the secondary should be designed by the faintest peak. The internal process of Reduc reverse the image by 180° in order to show the secondary as the brightest peak. Just a question of easiness for the user :-)

Images measurement

The choice of the best solutions among the ten solutions computed must be adapted to your equipment and you'll have to test.

However, here is a rule of thumb that can help to begin:
Airy disk ~ 2 or 3 pixels => use preferrably S2/S1
Airy disk ~ 5 pixels => use S3/S4
and so on...

Smaller the sampling, greater the number of the solution.

Warning:
Using visualization sliders, make sure that the mask does not modify the shape of the secondary peaks, this leads to less accurate distance measurements.

The measurement is the same way that you measure a double star:
- either you measure the central peak and one one of the secondary peak
- either you measure the two symmetrical peaks (take care because the distance between peaks is twice the actual distance)

However Reduc proposes something simpler with the Interferometry Fast Measurement option

Interferometry Fast Measurement

Activating the Interferometry Fast Measurement option, we can now measure the correlogram with a single click.

Select the peak to be measured and then right click

Reduc computes immediatly the position angle and distance of the system and inserts them into the measurement window.

Obviously the measured peak must match the B component

Think of discarding the function if you measure later 'classic' images!

Star trail Calibration
A star trail is a good way to evaluate the East/West axis. The trail is however sensitive to the turbulence, if one wishes to get an information of quality it is necessary to apply some rules at the time of the capture and of the exploitation of the image:

- Aim at a bright star that does not saturate the sensor
- Bring it on the east side of the sensor
- Start the exposure
- Stop the RA motor
- Stop the exposure when the star has crossed the sensor.

Three factors are important, the length of the trail, its time duration and the intrinsic quality of the picture.
The length must be as longer as possible, a complete crossing of the sensor is the best but you can deal with a few less if you are afraid about loosing the star.
The duration: a trail of two seconds doesn't bring any valid information, it is necessary to be the longest possible, about ten seconds can be sufficient in many cases. In case of uncertainty catch several trails and use the mean of the results.
The quality of the frame:the star must be bright enough in order to let a straightforward trail on the sensor. Avoid the weak stars and the double stars oriented to 90° or 270°!
The duration of exposure is long enough to spoil the frame with noise. It is imperative to preprocess the frame, the hot pixels might distort heavily the reduction. Subtraction of the offset and dark are imperative (flat field correction might be useful but is not absolutely necessary).

CALIBRATION :

- Load the correctly preprocessed frame and adjust the levels in order to see easily the trail.

- Click a point close to an extremity of the trail. Choose a place where the trace is well established (fig. 1)

- Adjust the size of the detection box to a value a bit larger than the width of the trail. (fig. 2)

- Click on the 'comp A' button (fig. 3)

- The points of calculation are marked with green crosses and the curve of regression is in red. It is time to give to Reduc the orientation of the quadrants. Click the corresponding button of the dialog box. (fig.6)

- The calculated values are propagated automatically in the reduction sheet. (fig. 7)

- The quadrants orientation is put up to date too. (fig.8)

It is wise to capture several trails and to keep the mean value of their reductions as the final calibration value.

Synthetic Drift : drift calibration on multiple frames
This method is dedicated to the users of cameras without long exposure capability. The remarks about the trail method remain valid. It is essential to choose a bright star in order to have a good S/N ratio during the movement of the star. Obviously it must not saturate the sensor.
The method of capture is straightforward, put the star near the East edge of the sensor, launch a series of short exposures and stop the RA the motor until the crosses the sensor.

Dark and Bias preprocessing

Recall and basic principles:
Raw image = Science image + Bias + Thermal Noise
When acquiring the images, take care to acquire also some dark frames. These frames are acquired with a closed obturator and an exposure time equal to exposure time of the raw frames.They contain the bias and the thermal noise. Reduc can synthesize a median frame of the set and subtract it to the frame when measuring so you do not need an external software to prepare your images. All is done in Reduc's workspace and your original frames aren't modified.

3/ Bias appears at the right of the button
Since that every frame loaded is automatically preprocessed.

If the size of the loaded image is incompatible with the one of the dark, Reduc does not preprocess it.
It is useless to deactivate the preprocessing if the list of files is in blue because Reduc can recognize the frames that it already preprocessed.

3/ Preprocessing is stopped

The batch mode allows to run several commands automatically.
Select the options then click Run it.

The left board plays the following sequence of menu commands:
1-File/Open or File/Load Folder or File/FITS Cube
2-Processing/Align (pixel) or Processing/Align (sub-pixel) (Aligns on the brightest centroid and resizes the images to the selected default size or prompts to enter a custom size or skip the step)
3-Processing/BestOf (does BestOf before or after the alignment or skip the step)
4-Processing/Stack List if the checkbox Stack is checked

The right board is dedicated to cubes pre-processing. It allows processing several cubes at a once.
The option Save as gives the result as new cubes or as expanded cubes in individual folders.
The output format is either 16 bit int or 32/64 bits float.
Aligning can be done at sub-pixel scale or without interpolation (the latter is just a centering on the brightest pixel and keeps the original information of your images).
If Prompt for the size is selected, you will be prompted for each cube to define the size, if not the frames are normalized with the selected size.
Do bestOf allows the resulting cubes or the resulting folders to be BestOf ordered, the FITS files being numbered from the best to the worst frame.
Finally Save autocorrelations generates folders containing the autocorrelations as if you had run Interferometry/Autocorrelation manually.

- If Save as Cubes is selected
New cubes are created in the same folder than the original, they are named as follows: OriginalName_ppFORMAT_SIZE_ALIGN_BESTOF.fits where
FORMAT can be i16 (for Int 16) or f32 (for float 32) or f64 (for float 64)
SIZE is the value of size (sz128, sz256 or sz and a custom value if you select Prompt)
ALIGN alPix if no interpolation or alSubPix if interpolated
BESTOF if a BestOf option is selected
For example if you load a cube named MyCube and run the batch as shown in the illustration, the resulting cube will be named: MyCube_ppi16_sz128_alPix_BestOf.fits

- If Save as Files is selected
New sub-folders containing the expanded cubes are created in the folder of the original cube.
Each folder is named as follows: OriginalName_ppFORMAT_SIZE_ALIGN_BESTOF
For example if you load a cube named MyCube and run the batch as shown in the illustration, the resulting subfolder containing the expanded cube will be named: MyCube_ppi16_sz128_alPix_BestOf

In every case when Save autocorrelations is selected a sub-folder containing the autocorrelations is created. It is named with the same rule and followed by _AC.
Hence if you load a cube named MyCube and run the batch as shown in the illustration, the resulting subfolder containing the correlations will be named: MyCube_ppi16_sz128_alPix_BestOf_AC
Please note that the correlation files bear the extension .redc. This type of file is an internal high precision format used by Reduc. They can be loaded and used only by Reduc.

Language
Reduc interface is available in French, English, Spanish and Italian.
You can switch by the menu ?/Langage/... in the main window.

I thank very much Edgardo Ruben Masa Martin, Gianpiero Locatelli and Antonio Adrigat for the spanish and italian translations.

Horizontal and Vertical Swap
You can change the orientation of FITS files when they are loaded by Reduc. Since it is checked the option remains active for every FITS files loaded .

Cursors Set
The menu Options/Cursor allows selecting various cursors accordingly to your preferences.

Image Header
The menu Window/Image Header shows the header of the current frame loaded

Click on [...] button to customize the list.

Add a camera : type its name and the pixel size.

Delete a camera : blank all informations about it.

ABOUT THE PIXEL SIZE
the important thing is the relative size on the final image. Generally with square pixels there is not any problem. With rectagular pixels, two cases must be taken into account :
- the final image is a strict raw image of the sensor (1 pixel on the sensor -> 1 pixel on the image)
- the final image is interpolated by the camera or the capture software in order to simulate square pixels.
In the first case you enter the values corresponding to the features of the camera, in the second it will be necessary to enter an identical value in Horiz and Vert.