 | Tangra 1.4 |
| TANGRA | VIDEO RENDERING | PHOTOMETRY | PRE-PROCESSING | TRACKING | LIGHT CURVES | ASTROMETRY | LINKS | LICENSE | DOWNLOAD |
Astrometric Calibration
For a successful calibration and astrometry it is important to read and understand the information given below.
Calibration is necessary for Tangra to be able to solve plates and run astrometry. Because different cameras, grabbers, drivers, software and accessories will generally result in a different size of the resulting pixels in the video frames, it is necessary to calibrate the different configurations you want to be using for astrometry. The only parameter that can vary without a need to recalibrate is the focal length of the telescope which usually varies slightly with the outside temperature. If you change only one other component of your configuration you will need to recalibrate before you can do astrometry. For example if you put a diagonal (which you didn't have before) you will need to recalibrate. If you change your frame grabber driver or grabbing software - you will need to recalibrate. If you change the parameters of your software grabber - you may need to recalibrate. If you change your frame grabber hardware - you will need to recalibrate.
When you attempt to run astrometry Tangra will ask you to choose a calibrated configuration (left below) or to choose a new one (right below). First you choose your video camera and then the name of the configuration. For each configuration it is important to choose a flip settings if applicable. If there is an odd number of reflections in the optical path then you'll need to choose a 'flip' settings. It is generally not important which flip you are going to choose because Tangra can align the image in any rotation as long as RA and DE grow in the correct directions.
 |
 |
However if you have an equatorial mount and you prefer to see North up and East to the right you will need to choose the correct flip and rotate settings. The limiting magnitude that is set for the configuration has the meaning of the faintest magnitude you can detect with this configuration. If this is an integrating video camera, then choose the faintest magnitude detectable with the longest integration. Tangra will not attempt to recognize/fit reference stars fainter than this configured magnitude.
If you have selected a calibrated configuration, when you press 'OK' Tangra will proceed with the astrometry. If you choose a non calibrated configuration be sure the "Calibrate Now" checkbox is checked and when you press "OK" Tangra will start the calibration process.
The first step of the calibration is to define the "on screen display" area. The purpose of this process is to tell Tangra if there is an area on the screen where additional information is displayed (such as GPS coordinates and time). If this area is not excluded in the field alignment then the OSD pixels will interfere with the alignment process and could make it run very slowly or could make it fail. The OSD area is only excluded during the alignment and not during the astrometric fit. Do define the OSD area you will first need to click on the OSD Area button on the top, right under the "Frame Actions" menu item (See below).
Once the tool is activated you can use the mouse to resize and effectively reposition the OSD area on the screen. The default OSD exclude area is at the bottom of the screen, which corresponds to where the KIWI OSD timestamp is located.
For a successful calibration your video should contain at least 20 not too faint real stars. If you have many bright hot pixels then it is recommended to subtract a dark frame. For best results it is recommended to record a special video for the calibration. It is best to select an area of the sky that is not too crowded but has a sufficient number of stars (30-100 stars will be an easy to calibrate field). If you have one or more bright stars in the field of view (FOV) this will make your life easier when you match the FOV in a planetarium program.
When you have finished defining the OSD area, to proceed with the calibration click "Calibrate Configuration". The purpose of the "Adjust Star Depth" checkbox will be discussed later. At this stage leave it unchecked. There are two ways of completing a calibration and first we are going to look at the three star identification, which is also the default option.
After you have clicked on "Calibrate Configuration" Tangra will ask you some initial information such as the approximate center of the field. During the calibration process Tangra uses stars in a square region of 3xFOV diagonals around the entered field center. The size of the FOV will be derived from the selected camera model (cell size) and the entered focal length.
After the calibration Tangra will keep the focal length you have entered and will adjust the size of the matrix cells accordingly. So you may want to put as the initial focal length the actual focal length of your system in order to not end up with strange cell dimensions. Even if you do put an incorrect focal length though this will not affect the calibration or the astrometry accuracy. A more important input parameter is the epoch of the observation which is used for applying the proper motion of the stars. You also have the option of re-defining the limiting magnitude for this calibration/fit only which may be useful for integrating video cameras not used in maximum integration mode.
After you confirm the input parameters Tanga will display the current video frame and will draw cross markers on some of the objects. Because Tangra doesn't know yet which objects are stars, which are hot pixels and which are OSD display pixels, you may end up having markers on objects that are not actually stars. By default you should have around 25 cross markers displayed. Those of the marked objects that are actually stars can be used for a three star identification. If you see that there are not enough real stars marked by Tangra (you will need at least 10 of them) you can start the calibration again and this time check the "Adjust Star Depth" checkbox. Then you can adjust the depth Tangra will go to and get more marked objects.
The idea behind the three star identification is that you manually match the FOV from the video to a planetarium program and then identify three of the stars. You can identify a star by clicking on the marked objects. As soon as you hover your mouse over an object that can be used in the calibration the cursor will change to a hand and when you click on the object you will be able to pick the star number from a list (see below). The star identification form can be used to filter the stars. The numbering of the stars used by Tangra may be different than what your planetarium program uses so pay attention to the expected format. As you type, Tangra will filter only those stars in the 3 FOV region around the previously entered field center that match the entered part of a designation. Once you see the star you want to identify the object to - you select it from the list and click "Identify". You will need to do this for 3 stars and if you get one of the objects wrong you can use the "Clear" button and start again.
Once you have clicked "Identify" on the third star, Tangra will go an attempt to do an automatic calibration. For this to work it is important that you have identified the stars correctly and that all of the 3 stars have RA/DEC that are different. It is best to select 3 widely separated stars. Sometimes two stars may have very close RA or DEC or there may be a different issue with the positions of the stars that may prevent Tangra from successfully completing the calibration. In such cases you should try with a different set of stars or follow the instructions from Tangra. Also if the total number of stars in the FOV is quite large (say more than 200) the calibration may take half a minute or more (depending on your computer speed) so be patient. Tangra will always return back to you with a positive or a negative result. If the operation takes too long after the configured timeout Tangra will get back with a negative result. So be patient and wait for Tangra to respond.
If the calibration is successful you will see green circles around the successfully identified stars. Your configuration will be now marked as calibrated and the calibration data will be saved. The image below contains more than 500 recognized stars down to magnitude 15 and as you can also see stars close or inside the OSD area have been also identified correctly regardless that this area has been excluded for the initial alignment.
An alternative way to complete the calibration is to manually match the stars to the objects on the video frame by rotating, scaling and shifting the plane with the expected star positions. This may be challenging but could be also easy if there are some very bright stars in the FOV. An example is given below. To switch to a manual calibration mode you will need to check the "Do a manual fit" radio button. When you do this Tangra will now plot the stars from the selected field based on the initial focal length, video camera cell size, video frame size and with a rotation of 0 degrees and aspect of 1. If North is up on your image then you should not need to rotate. The example below has been recorded with a polar mount and we will need to rotate to complete the calibration.
The first task is to actually recognize some of the bright the stars (the blue circles) and match them to the objects in the video frame. You can adjust the limiting magnitude so fewer stars are plotted and can then try to identify the bright stars in the field. On the example above we have marked the stars and the video frame objects that match to those stars. The first task will be to pan the star field until the bright star is positioned on the top of its matching object from the video frame. Simply click on the frame, hold the mouse button down and then move around the star plane until the bright star is positioned correctly.
After that we'll need to rotate the star field until the second star (see above) is also positioned correctly. Because the rotation is done relative to the center of the FOV, while doing the rotation the previously correctly positioned first star will get misaligned. So while rotating our goal should be to make the line between the two stars parallel to the line between the matching objects on the video frame. Once we have achieved this we can now pan the star field up again so the two stars are aligned.
The next task will be to find the correct focal length and because of the way the Aspect works in Tangra, it is going to be less work if we first fit the focal length vertically ignoring any horizontal misalignment. For this we will need to change the Focal length until the vertical distances between two well separated stars becomes the same as the vertical distance between the corresponding objects in the video frame. In our case we will change the focal length until the distances A and B become the same (see below). While doing this the two initially aligned stars will get misaligned again.
However what is really important here is that vertically most of the stars and their corresponding object on the video frame align. Once we have a good vertical alignment we should not change the focal length any more. Pay attention to the horizontal green lines (see below) that show that the correct vertical alignment has been achieved across the entire video frame. The next step would be to also get the horizontal scale and we do this by adjusting the aspect.
We should only need to change the Aspect and pan horizontally at this stage until part of the image gets a relatively okay match - as the green area shown below. You don't have to fit the full image and the match doesn't have to be perfect. Tangra will be able to work out the calibration even with an approximate fit. Once you get some alignment you can press the "Solve Plate" button to see whether Tangra will work it out.
If the calibration fails you can tinker with the fit a little more and then give it another try. And to give you an idea of how good/bad the manual fit should be before you try to "Solve Plate", the image below shows a pretty rough fit which Tangra did not have problems calibrating. As you can see the manual fit doesn't have to be that good at all and all you need is to have the stars (blue circles) to be closer to the correct objects from the video frame than to any other objects for 7 to 10 stars in the field. If you can do this the calibration will be successful.
Once Tangra has worked out the calibration for you the calibration data will be automatically saved in the configuration.
Test videos from the examples above containing only 3 frames and with a size of 3Mb each are available below.
Example 1Video File: Example1 - (18h18m46s -18d13'12'').avi (right click and choose "Save Target As...")Size: 3 Mb Calibration Type: 3 Star Identification Video Camera: WAT-120N+ (PAL) Flip Horizontally Focal Length: 1000 Limiting Magnitude: 15 Field Center: RA=18h 18m 46s DE=-18d 13' 12" Identified stars to the right |
|
|
Example 2Video File: Example2 - (19h43m01s 11d50'41'').avi (right click and choose "Save Target As...")Size: 3 Mb Calibration Type: Manual Fit Video Camera: WAT-120N+ (PAL) No Flip Focal Length: 1000 Limiting Magnitude: 14 Field Center: RA=19h 43m 01s DE=+11d 50' 41" |
||