> > Also, for what it's worth, we at Gemini are currently deciding what
> > display tools we'll support for observing and data reduction in the
> > future. I'd very much like Gaia to be a serious contender in this,  
> > but I
> > do think the majority opinion will be that an option to display
> > north-up, east-left with a single click and arbitrary rotation angles
> > has to be a hard requirement.
> 
> Does it matter if it takes a while for very large images? 

Depends on "a while". I think it would need to be able to handle
something of order 6000x4000 pixels (ie a GMOS image) in not more than
say 4 seconds on a reasonably modern machine.

> Do people  
> want control over the rebinning/resampling scheme used 

Nope...

> or is it  
> understood that you might not be able to get completely accurate  
> photometry with optimal PSFs from such images?

I think that would be fine.

Our use case isn't really to be doing detailed analysis on the rotated
image, it's mostly simply for visualisation and comparison with finding
charts and other data.

Of course, ideally one would have the situation where it purely is
simply a display mapping and if you do photometry etc within Gaia it
simply takes the location of the appertures or sources etc and does the
actual analysis on the original data, much in the way as if you analyse
data that's zoomed back to 1/4 scale or something.

One thing that might be an annoyance with doing with an actual kappa
regrid call is that if you mouse over the rotated image, then the X and
Y pixel co-ordinates would presumably be those of the re-gridded image
rather than the original data file, making it difficult to measure say
original pixel co-ordinates of a feature in the image from the rotated
view... Hmm, now that I think about it this might be a significant
problem wrt our target acquisition mechanisms... I wonder if it would be
possible to propogate a co-ordinate system that represents the original
detector pixels into the rotated image and then have Gaia display that
in the X and Y fields?


> It should be pretty trivial to implement this (says the man who is not  
> doing it) since GAIA could just call astRebin itself just like KAPPA  
> REGRID does (or even in the short term simply call KAPPA REGRID to do  
> it - but that might end up in a menu rather than a button on the main  
> gaia display).

Cool! Do let me know when you're looking for Beta testers... :-)

Cheers,
 Paul.