Vertical flip

[bitli]

Hello,

When I open a FITS image in DSS or AIP4WIN, it is mirrored vertically when compared with the same image opened in PI.

Who is right ? PI, of course :-)... but why is there this discrepancy ?

-- bitli

[Carlos Milovic]

There are two standards for image coordinates. Bottom-left and Top-left. It will depend on the data type you are using. You may use whatever you want, editing the FITS preferences (format explorer -> Fits -> ... ).

[Niall Saunders]

Try opening the image in (the free app) "FITSVIEWER" - that will display the image the way FITS was 'meant to be' (at least, I take a FitsView as my 'benchmark'), however the encoding of the image itself might be 'wrong' - that error is the responsibility of the software that 'captured' the data.

HTH

[Carlos Milovic]

No no. It is not an encoding error. It is just that the FITS format allows both type of coordinate definitions, and both are used in different scenarios. Top-left is more consistent with other formats, and is widely used in most image processing tasks. However, bottom-left is more used in scientific and pure algorithmic processing, since it is just a Cartesian coordinate system, with the right orientation of axis.

So, it is not big deal. Since vertical mirroing does not change pixel data (preserves all information), there is no problem to perform that operation. Anyway, as I said, you may set your favorite orientation with the Fits preferences.

[Niall Saunders]

Hi Carlos - sorry, 'wrong encoding' didn't really fit what I was trying to say.

Basically, I use FitsViewer to verify what the 'natural' data should 'look like', for an astronomical CCD image taken straight from whatever capture software that I am using. Then I would compare that orientation with what I see in the likes of PixInsight. And, I seem to remember that the 'default' behaviour of PixInsight conforms to FV.

Any other orientation flips that I encounter are then either attributable to 'other' processing software, to the capture software, or to the orientation of the CCD on the OTA, and the number of optical 'elements' in the OTA itself.

But, as you correctly state - there is nothing actually 'wrong', all a user needs to do is to be aware of 'where' any flips have occured, and to then take a final decision of how they want to portray the 'actual' image when they are finished (bearing in mind that all 'FLIPS' and 'FLOPS' are 'lossless transformations' in PI).

Cheers,

[Carlos Milovic]

[bitli]

Thanks everybody. Indeed I can change the orientation in the format explorer, so that it looks the same as the other programs I used (it is more comfortable to make comparisons).
-- bitli

[rol_ler]

Hi All!

Is it important when I doing calibrated frames?

I mean: when I capture frames of a galaxy for example I have to use default setting in Pi (have to use 3 different sw at capturing, so I have to synch them) to see proper picture.
Calibrating frames: I have to use "Lower left corner (bottom up)" setting in Pi.

Could anybody confirm that in that case I have to use different setting in Pi fits prefernces, please?

[jegsav]

Among the programs that I use, half use one convention, the other half uses the other. As far as I can see PixInsight is the only software that does not allow the image to be mirrored when viewed to accommodate the two conventions. I think this is poor and should be corrected. (And what about a decent manual too?)

[Niall Saunders]

Hi,

You state that certain software conforms to one "convention", whilst other programs conform to "the other convention".

As far as I am aware, there is no such "convention" defined anywhare - plese, someone correct me if I'm wrong.

The manufacturers of imaging devices, at component level, have some notion of where the (0, 0) position is, and in which direction things 'count up' (i.e. rows before columns, or columns before rows, etc.). Then a camera manufacturer selects that particular imaging device and has the freedom to impose whatever convention they choose to use - they do not have to adhere to any conventions whatsoever.

Now you add an external software package to control the camera and to download the (un-)conventional images taken by the imager. Guess what? Yep, there is, again, no convention whatsoever that specifies how that image should be stored on disc - or, rather there are a host of conventions that could be used, none of which has any greater standing than any other.

Foryunately PixInsight knows this, and gives the user the flexibility to apply an over-ride to the stored data such that the user is then given the power to impose their own 'convention' to the data typically acquired from their imager.

Of course PI provides the ability for the data to be 'mirrored' - in fact it even allows for 'both' types of miroring - in either the horizontal or vertical direction (these two actions are known as Flipping and Flopping - although I can never remember which does what !!!)

So, unless I have missed some crucial point in your brief reply, I hope that this has been of some help.

As to release of a manual, is this really necessary given the wealth of information already released and published about PI. If one were to be released, just exactly how many people would read it? And how many people would refer to it more than once? It would ultimately just be as useful as the manual for your PC, your Operating System, your Toaster, your TV and even your Car.