Harry, I have that new version downloading and will give it a try. Two things: 1. With your 6 May2009 version I have successfully stitched smaller panoramas from the problematic project and of a simple (not bracketed exposure) 6xHorizontal + 1xSky + 2xGround type panorama. That "simple" project did include Alpha Channel masks on two out of the nine images. As I type this the 6 May2009 version is running on a 12,000x6,000 version of that panorama.
So it (6 May2009 version) does seem to work as well as the Beta-4 version did - and crashes in the same way that the Beta-4 version did when asked to stitch the full-size version of the problematic project. So your 6 May2009 version is not any worse than the Beta-4 version. 2. two people took me up on the request to download the 2GBs of files and try the problematic project on another platform. So far Rogier Wolff has reported back on having completed that test using an Ubuntu based system and he said: > It crashed for me too. > Loading next image: t2_exposure_layers_0025.tif > ... > Creating blend mask: 1/6 2/6 3/6 4/6 5/6 6/6 > enblend: out of memory > std::bad_alloc > make: *** [t2_exposure_00.tif] Error 1 > (near image 25) So I am returning to my original view that this may not be a Mac OSX specific problem. My thinking is that there is a bug in there somewhere associated with masking that causes demand for memory allocation to run away and exceed what is possible, no matter how little or how much RAM or VM is available. So while the symptom is "running out of memory" the actual bug is is not specifically a memory-allocation problem but is more to do with an algorithm running out of control and memory is just the thing that runs out first. But I should say that is just my guess - in the context of Hugin - I am a user not a programmer and I don't know the internal structure or workings of enblend. Rogier did lots of other testing monitoring memory allocation and I probably has lots of useful debugging data. all the best George On May 7, 5:11 pm, Harry van der Wolf <[email protected]> wrote: > 2009/5/7 grow <[email protected]> > > > > > Harry, > > I ran the problematic project at 12,000 x 6,000 and it looks to me as > > if I got the usual crash and error message ... see below > > > all the best > > > George > > Hi George and other Mac users, > > Thanks again for testing. > I now have a last final test option. I compiled Christoph Spiel's branch > against dmalloc[1] instead of the standard OSX malloc. OSX's malloc doesn't > seem to be too good (see several of the previous posts). (The official > enblend sourceforge can't be compiled against dmalloc) > dmalloc is supposed to be a very clean implementation of malloc and, as an > extra bonus, it should also outperform MacOSX's own malloc [2] > > @George: Please download the RC1-dmalloc version and please give it another > (last?) try. > > @All other mac users: Please also download this version to check whether > dmalloc does function correctly. > > Information and binaries via my website > <http://panorama.dyndns.org/index.php?lang=en&subject=Hugin&texttag=Hugin>. > (The binaries itself are, as always, served from hugin.panotools.org who > kindly provide the disk space and bandwidth). > > Harry > > [1]http://www.dmalloc.com > [2]http://sekhon.berkeley.edu/macosx/ --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "hugin and other free panoramic software" group. A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/hugin-ptx -~----------~----~----~----~------~----~------~--~---
