CCDProc Release Notes Notes are stored in reverse chronological order V2.0 (2011 Nov 08): [rwp/osu] ------------------ Revamped the function prototypes - still repeating across the various routines, sloppy, but no time to fix that. Compiling with g++ for better compile-time fault detection (much more unforgiving compiler, and catches missing system includes that gcc inexplicably gives us a pass on). Multiple changes to many of the low-level and stats routines to fix bugs that have emerged. Works well with the latest revisions of cfitsio, but we're not yet exploiting some of the newer and better capabilties. We have also renamed CCDProc.h -> ccdproc.h, more in keeping with our current coding standards. For the future, we are moving more to python, specifically using PyFITS and Numpy, as much more powerful for many kinds of pipeline building applications, however there are still issues of performance of compiled C vs. python we have not yet come to grips with. A prediction is that development will slow with this release and we'll starting putting more effort into python tools, but there is as yet no organized effort beyond basic reconnaissance as of yet. V1.5 (2005 Jan 21: [rwp/osu] ------------------ Added the imstat function to compute min, max, mean, mode, sigma for an image. The mode is crude (peak of histogram), nothing fancy like fitting a parabola to a histogram peak or other, so the routine should be viewed as only for quick-n-dirty analysis of an image, not science analysis. Sigma is similarly an approximation form, not a 2-pass calculation. Stats can be computed in the whole image (default) or a subframe. Type imstat for the syntax. V1.4p3 (2004 Sept 23): [rwp/osu] --------------------- Added the normalize and dicer programs to the suite. Documentation will be forthcoming, but in brief: normalize - divide all pixels in an image by the mean intensity across the entire image (or in an image subset), to create a normalized image (e.g., for making normalized flats) dicer - dice a larger image into multiple subimages (much more efficient than running ccdproc with different trim boxes multiple times). Collectively released as patch 3 of the package flt2int v0.1 (2003 Aug 4): [rwp/osu] ------------------------- Converts BITPIX=-32 to BITPIX=16 with data editing to [0..65534] to avoid discretization errors on coersion into 16-bit signed integers. Should be used only to convert single, uncombined images that have been output as BITPIX=-32 into BITPIX=16. V1.4p2 (2001 Oct 18): [rwp/osu] -------------------- Stephen Kane found that the -a (average) option was giving incorrect numbers. Running it down, I found an error in the addition routines whereby imcombine was not adding the last image in the stack. This was fixed and v1.1 of imcombine was released, constituting "patch 2" to V1.4. V1.4p1 (2001 Jan 22): [rwp/osu] -------------------- Bug in ImDiv whereby we were getting the memory registers out of synch has been fixed. This only affected cases where the flat field images had pixels with data values = 0.0. Andrew Williams (Perth Obs) reported this bug to use, and we fixed it (albeit differently than his suggestion). V1.4 (1999 Aug 8): [rwp/osu] ----------------- Added code to do dark subtraction. In the processing model adopted here, dark frames are preprocessed either OT or OTZ prior to applying them to object frames. Dark subtraction comes between zero subtraction and flat-field correction (you don't want to subtract dark counts after the flat-field division has "flattened" the pixel-to-pixel gain variations across the detector). Code has been added to include the "DARK" function in proc files. V1.3p2 (1999 July 2): [rwp/osu] -------------------- Fixed odd bug in which the FWHMSKY would cause a segmentation fault crash. The cause appears to be a bad function prototype in CCDProc.h (actually, no function prototype to speak of for gparam). Unclear why this never surfaced before. It was causing bad values of the gain to be passed to gparam. Also instituted full Y2K-compliant date tags. Kind of ugly, but does the job (ISO8604 is not always pretty). V1.3p1 (1999 June 1): [rwp/osu] -------------------- Fixed a voodoo bug with writing SHORT images that seems to be related to stale header info in CFITSIO. Didn't see it in earlier versions, but I can't figure out what. Fixed it with a hack (open then close the input file after reading the data) that seems to do the trick. Weird stuff. Also added an extra argument to FITSError to let me get a little more debug info on FITS crashes. Thanks to Andy Stephens for sniffing out this one. V1.3 (1999 May 27): [rwp/osu] ------------------ ccdproc and imcombine now have a Makefile instead of the gccfits script for compilation. Make is the right thing to do, and should make adding things more transparent to the casual user. 00README file has been updated accordingly. Fixed minor bugs in getparam.c, in which the modal sky values were being systematically underestimated, and subject to 5.333 (16/3) ADU discretization noise due to using very coarse bins. This was accomplished by changing nhist to 32000 instead of 6000. Only makes a minor performance impact (same time to build the histogram, but takes a little longer to search it). The major time hit is still in the PSF fitting, so this is incremental at best. Tested this version of the code with CFITSIO v2.031 (released 1999 March 31) and found no problems. People trying to build CFITSIO on SunOS with mixed gcc and sun compilers *may* need to tweak the makefile as there is some include file conflicts in poorly configured systems (based on experience with the SunOS systems at OSU, your mileage may vary). ---------------------------------------------------------------- V1.2 (1998 Nov 23): [rwp/osu] ------------------ Fixed bug in getfwhm (in gparam.c) in which if a star used to estimate the FWHM is within 80 pixels of the image boundary it overwrites the image. Also added a MAXSTARS parameter to gparam.c to allow us to set the max number of stars used for FWHM calculations at compile time (rather than hardwiring 20). Recast routines in gparam.c slightly to use ANSI-style function prototypes, and to isolate printf's inside of #defs. This lets us reinstate the diagnostic printouts by defining __DEBUG at compile time. Tested this version with v2.024 of CFITSIO, which includes support for the new Year2000-compliant FITS date/time format, and fixes some bugs over version 1.42. No problems adopting the latest version of CFITSIO for ccdproc or imcombine. Added imcombine program to this release. ---------------------------------------------------------------- V1.1 (1998 June 2): [rwp/osu] ------------------ That didn't take long. Checking it out myself, I found a bug in the module of ccdproc.c (main) that writes the FITS header cards for the FWHMSKY estimation. TSMAX was mis-spelled as TMSAX, hence subsequent programs could not find the header card when correctly spelled. Dyslexia really sucks. ---------------------------------------------------------------- V1.0 (1998 May 27): [rwp/osu] ------------------ First non-beta release, with HTML documentation http://www.astronomy.ohio-state.edu/~pogge/PLANET/CCDProc/ distributed to PLANET consortium for testing and evaluation.