On Mon, 30 Nov 2009, Tim Connors wrote:
>> On Tue, 20 Oct 2009, Tim Connors wrote:
>>
>>> On Mon, 19 Oct 2009, Peter W. Draper wrote:
>>>
>>>> On Mon, 19 Oct 2009, Tim Connors wrote:
>>>>
>>>>> I'm trying to feed realtime 16 bit unsigned data into skycat and
>>>>> gaia (the skycat window displays bitpix=-16), but have encountered a
>>>>> probable bug where if the min cut value is set to < 0 with the max
>>>>> value > 0 (such as by right-dragging the mouse in gaia to decrease
>>>>> the contrast and increase the brightness - which is what most of our
>>>>> telescope operators have been trained to do), the display blanks
>>>>> completely when it is loaded with unsigned data as soon as min<0.
>>>>>
>>>>> Up until now, for our aquisition cameras, we've been feeding 8 bit
>>>>> unsigned data into a plugin that told skycat/gaia it was dealing
>>>>> with 8 bit signed data, and this has strangely worked even when the
>>>>> the data input is > 127 (I don't understand why it works, but let
>>>>> that slide for now; our code is a mass of spagetti). Now that we
>>>>> have moved to 16 bit for some of our cameras, the data values >
>>>>> 32767 now get mapped to "blacker than black" as I would have
>>>>> expected, but I can't simply get gaia to display unsigned 16 bit
>>>>> values as I would desire, because of the aforementioned min/max cut
>>>>> bug.
>>>>
>>>> Hi Tim,
>>>>
>>>> if you're using the rtdevt interface to do this (I expect so, since
>>>> you're sending -16 bitpix), then there are two elements in the
>>>> rtdIMAGE_INFO struct, "highCut" and "lowCut" that you can define to
>>>> set the initial limits used in the display. That should stop the
>>>> going negative issue.
>>>
>>> We're using something called dserv ("Drama Interface to pskycat",
>>> pskycat helpfully being described as the "pskycat image viewer"),
>>> which may or may not be AAO specific things, and rtdserver ("Files
>>> extracted from ESO RTD for use in AAO2 CCD software"; sounds like
>>> rtdevt to me, especially since it has the rtdIMAGE_INFO struct in
>>> there) but I haven't yet worked out how they interface together.
>>>
>>> Nevertheless, on our monitors, we'd want to be able to freely adjust
>>> the brightness in software, which would be more difficult if the
>>> minimimum cut levels were set to 0 (although the workaround would at
>>> least be useful in order to get the telescope operator out of
>>> trouble).
>>>
>>> Also, don't highCut and lowCut in rtdIMAGE_INFO just apply to the
>>> initial limits (that's what the description says anyway)? The
>>> autoadjusted initial limits are (usually - I haven't tested for cosmic
>>> rays yet) fine here, it's just when we right click-drag the mouse to
>>> set new limits that the image suddenly disappears if lowCut<0.
>>
>> Hi Tim,
>>
>> I've come across this limitation myself now, so have made some changes
>> to Skycat so that the cuts can now be outside the data range of short
>> integers. This allows the lower limit for unsigned shorts to be less
>> than zero, with the desired effect of brightening the background.
>>
>> Don't know where you go to with all this, but the changes are committed
>> to the JAC repository:
>>
>> http://starlink.jach.hawaii.edu/git/?p=skycat.git
>
> Ah thanks. Knew there had to be a way since it worked for datatypes.
>
> Have you got hints as to how you compile skycat? As far as I can tell,
> there is not a single mention of "libBLTX30.so" anywhere in the world
> apart from the skycat sources :)
>
> I have libBLT24 and libBLT on various machines, but with
> ./configure --with-tclinclude=/star/include/ --with-tkinclude=/star/include/
>
> I only get
> ...
> checking for BLT library...
> configure: error: could not find libBLTX30.so: Please use the --with-blt=DIR option.
> configure: error: /bin/sh './configure' failed for tclutil
>
> libBLTX30 is not even in the tcltk that I eventually found supplied on
> http://archive.eso.org/cms/tools-documentation/eso-skycat-download
Hi Tim,
this version of BLT is used after the recent upgrade to Tcl/Tk 8.5 that I
did for GAIA and Skycat. It's part of the source tree, so if you build the
complete Starlink you'll get it, but it's also at:
http://starlink.jach.hawaii.edu/git/?p=blt.git
I've not tried to build this outside of the Starlink tree, no don't have
any idea what it's status is wrt to other Tcl/Tk builds.
> The next battle if I get this to work will be to try to compile gaia again
> on a minimal installation with a similar fix, but do I undestand
> correctly that gaia will just be linked against the object files compiled
> for skycat?
No, it also links against a lot of other Starlink libraries, for NDF
access for instance, so you'd need a good chunk of things in /star/lib.
Cheers,
Peter.
|