compatibility of ISIS images with Visual Spec

[Robin Leadbeater]

The fits images produced by ISIS are not compatible with VSpec even after conversion to 16 bits. Currently I am converting to 16 bits and then producing .pic files using IRIS. It would be nice to have a VSpec compatible format conversion tool in ISIS.

Cheers
Robin

[Martin Dubs]

Hello Robin,

I have no problem opening ISIS image files in VSpec, both 16 and 32 bit (maximum file size 24 MB with an SBIG ST8300). However I cannot see the header in VSpec. I checked both with echelle files and files from LISA. Also spectra are displayed correctly.
I am using VSpec 4.0.0 and ISIS 5.0.2 on a Windows7, 64-bit computer.
This will probably not help you, but you should not despair. For looking at the images I use Audela, which has a nice zoom feature and also shows the fits header.

Regards,

Martin

[Dubreuil Pierre]

Hello Robin

temporarily:
For an image, it is possible in the opening (16 bits converted ISIS) with IRIS software.
Must then bring negative pixels to 0 by the menu "treatment -> clipping min" and save the new image therefore you should be able to open in Vspec.
Is this what you are looking for?

cordially

Pierre

[Robin Leadbeater]

Hello Pierre

ISIS fits files can be opened in Visual Spec but it does not interpret them correctly (values over 16 bit in 32bit files are clipped and values <0 in 16 bit files are not displayed correctly. (See the attached examples)

To correct this I convert 32 bit ISIS files to 16 bit using the tool in ISIS ( using the scaling option to avoid clipping)
I then convert the 16 bit fits files to pic files in IRIS (The special VSpec compatible format which can handle +- 16bit)
This works but a single step in ISIS would be useful.

Cheers
Robin

[Robin Leadbeater]

Hello Pierre,

Must then bring negative pixels to 0 by the menu "treatment -> clipping min" and save the new image

It is important not to clip negative values at any stage when processing spectra. If you clip negative values you will get false zero points which will give errors in normalised spectra.

Cheers
Robin