Hubble Space Telescope IC 434 repro

[admin]

Hey guys,

While testing StarTools 1.3.5.229 alpha, I decided to have a go at the HST's recently released magnificent infrared data on the Horsehead nebula. Not having anything in particular in mind, I was surprised by the result just following a basic 7-ish minute workflow and wanted to share. The colours came about by performing an auto-white balance (e.g. clicking sample in the Color module), which really helped making use of the full color spectrum, seeing as this is a false color image (2 IR spectrum lines).

For reference, this is the original rendition (Credit: NASA, ESA, and the Hubble Heritage Team (STScI/AURA)), which I also quite like.

More can be found on the NASA HubbleSite.

Cheers,

[Rowland]

Ivo. The detail enhancement is striking. It's a beautiful image.

[Epicycle]

Awesome rendition Ivo! Is there a link to the linear data so we can have a go at it ourselves? It should be fun to process this one. Thanks,
E.

[admin]

Awesome rendition Ivo! Is there a link to the linear data so we can have a go at it ourselves? It should be fun to process this one. Thanks,
E.

Yes, but (big but) only a 64-bit version of StarTools currently processes these image correctly (e.g. the MacOSX version will struggle). The reason for this has to do with ST's internal engine using 32-bit integers on these platforms (for purposes of speed, memory conservation and portability). Hubble's data comes in floats, but uses a huge dynamic range that does not fit comfortably in a 32-bit integer representation, even afater normalization. This huge dynamic range usage is partly because the stacking artifacts that some images contain (the IC434 is one of them), combined with a lack of information in the FITS files on how to deal with these artifacts and/or anomalous values. Hence ST has no choice but to take them as real values and allocate dynamic range for them.

A Mac OSX 64-bit version is in the works, but I've been also steadily working on ways around the problem, as I really want the 32-bit version to be a viable option for people on, for example, Android tablets and TV sticks. The latter devices are severely memory constrained and deal very poorly with floating point data and arithmetic, so I'm keen to find a workable solution for the sake of these users.

You can find the data on the Hubble Legacy Archive, or do a search for IC 434 on the HLA serach engine. Sort the results on exposure time and get the first entry.

Cheers,

[Epicycle]

Thanks Ivo!
E.

[admin]

Thanks Ivo!
E.

Made some progress with the 32-bit versions and devised a way to let the 32-bit versions process high dynamic range data like this while still keeping integer throughput speeds and conserving memory... stay tuned!

[luigi]

This is awesome.
I tried to download the .fit file linked but when opening it with starTools nothing is shown. Which option should I choose?

[admin]

This is awesome.
I tried to download the .fit file linked but when opening it with starTools nothing is shown. Which option should I choose?

Hi luigi, which platform and version of StarTools are you using?

[luigi]

This is awesome.
I tried to download the .fit file linked but when opening it with starTools nothing is shown. Which option should I choose?

Hi luigi, which platform and version of StarTools are you using?

The last 1.3.5 beta on Windows 64 bits.

[luigi]

I was able to open the file, it needed a crop it seems...
Of course I can'teven get close to your results with the image that shows how much I have to learn about ST.

I have problems with dynamic range (areas are overexposed others too dark) with color calibration (it's always too blue or too red) and with noise, contrast and sharpness. The rest (?) is fine

Do you have a step by step of how you processed the photo? Or a video? A video of this can do a lot of good to ST as many other HLA images can be processed as a way to learn the tool and processing of good quality data.