Hi David, Thanks for the reply! Ok, I tried multiblend with my another panorama on several systems. Maybe you will find this useful:
Ubuntu x64 - works with any number of levels Ubuntu x32 - works with 1 level, doesn't work with 2 levels. Windows Vista x64 - works with any number of levels Windows 7 x32 - cannot load even half of tiff images. The utility you suggested http://www.techpowerup.com/forums/showthread.php?t=112556 didn't bring any difference. Every machine has 4 GB RAM in total, on every system (with the exception of Windows 32), I had about 2.1 GB RAM given to the process. I am not sure about the exact peak or the maximum allowed by a system - I used the simple system monitor to track the memory. If this is important, the uncompressed resulting panorama tiff weights 628 MB. The total of overlapping for every source image was about 100% - 30% from one side + 30% another + ... Evgeny On Thu, Apr 12, 2012 at 6:00 PM, Monkey <[email protected]> wrote: > Hi Evgeny, > > Memory usage is definitely an area where multiblend needs improvement. > > At the point it's failing, it's trying to malloc 164mb to store the one of > the channels of the output image. This isn't much compared to the ~1.4gb > it's already using, which makes me think you're using a 32-bit build. If > you're running a 64-bit OS, you should be able to complete the blend with a > 64-bit build of multiblend. > > If you're using my Windows builds, I'm not certain the 32-bit exe is large > address aware (I'll make sure it is on the next build). You could try this > utility to make the exe large address aware (able to use 3gb/4gb depending > on your version of Windows, instead of just 2gb): > > http://www.techpowerup.com/forums/showthread.php?t=112556 > > Finally, you *might *have some success with splitting your TIFFs into two > sets and blending them separately, then blending the two results together - > but multiblend doesn't preserve TIFF coordinates (something else to fix!) > so you'd have to realign the two halves in Photoshop (or similar) before > the final blend. > > The amount of extra memory required for each extra level is only 1/4 of > the memory required for the level above, so it shrinks pretty quickly as > you increase the number of levels. To go into more detail, there will be X > mb assigned for the input images, Y mb for the output image, and Z for > working space - it's only Z that increases with the number of levels, and Z > for all levels will never exceed 1.333 times the amount of memory required > for just a single level. Additionally, Y does not begin to be malloc'd > until some of X is discarded. > > The next version will hopefully have some rudimentary disk caching (and > the ability to blend much larger mosaics with the 64-bit build, if you have > plenty of RAM). > > David > > > On Thursday, April 12, 2012 4:47:46 PM UTC+1, Evgeny Toropov wrote: >> >> Hi, David, >> >> How much memory does it need for blending with L levels? >> I have a huge wide cylindrical panorama 72076x2278. >> When all the images are loaded, multiblend consumes up to 1.4 GB out of >> my total 4 GB. >> The program only works with 1 level (-l 1). Otherwise, I get the >> following error: >> >> " >> 72076x2278, 8bpp, 2 levels >> split done >> seaming... >> masks... >> blending... >> Error: not enough memory for output channel! >> " >> >> When 1 level option is set multiblend takes ~4 minutes to finish (almost >> all is time to load nona tiffs), while enblend (also with 'levels' set to >> 1) takes 140 min. >> >> >> >> >> On Tue, Mar 27, 2012 at 2:11 PM, Monkey <[email protected]> wrote: >> >>> Some pseudowrapping bugs cleared up (including one that *created* a >>> seam in the middle - oops!) - now at version 0.31beta. >>> >>> Windows binaries: >>> http://horman.net/multiblend/**multiblend0.31beta.zip<http://horman.net/multiblend/multiblend0.31beta.zip> >>> Source for Linux/Mac/FreeBSD etc: http://horman.net/multiblend/** >>> multiblend0.31beta.tar.gz<http://horman.net/multiblend/multiblend0.31beta.tar.gz> >>> >>> David >>> >>> >>> >>> On Tuesday, March 27, 2012 12:01:56 AM UTC+1, Monkey wrote: >>> >>>> Hi all, >>>> >>>> Previous discussion: http://groups.google.com/**group** >>>> /hugin-ptx/browse_thread/**threa**d/b08211b2659a7eab<http://groups.google.com/group/hugin-ptx/browse_thread/thread/b08211b2659a7eab> >>>> >>>> I've implemented a feature I'm calling "pseudowrapping" to blend >>>> around the left/right borders for 360 degree panoramas, a much- >>>> requested feature. >>>> >>>> There are no command-line switches to access this; pseudowrapping is >>>> enabled any time multiblend is run with only a single uncropped TIFF >>>> (such as that already created by a normal multiblend run) as its >>>> input. >>>> >>> -- >>> 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<http://wiki.panotools.org/Hugin_FAQ> >>> To post to this group, send email to [email protected] >>> To unsubscribe from this group, send email to hugin-ptx+unsubscribe@** >>> googlegroups.com <hugin-ptx%[email protected]> >>> For more options, visit this group at http://groups.google.com/** >>> group/hugin-ptx <http://groups.google.com/group/hugin-ptx> >>> >> >> > On Thursday, April 12, 2012 4:47:46 PM UTC+1, Evgeny Toropov wrote: >> >> Hi, David, >> >> How much memory does it need for blending with L levels? >> I have a huge wide cylindrical panorama 72076x2278. >> When all the images are loaded, multiblend consumes up to 1.4 GB out of >> my total 4 GB. >> The program only works with 1 level (-l 1). Otherwise, I get the >> following error: >> >> " >> 72076x2278, 8bpp, 2 levels >> split done >> seaming... >> masks... >> blending... >> Error: not enough memory for output channel! >> " >> >> When 1 level option is set multiblend takes ~4 minutes to finish (almost >> all is time to load nona tiffs), while enblend (also with 'levels' set to >> 1) takes 140 min. >> >> >> >> >> On Tue, Mar 27, 2012 at 2:11 PM, Monkey <[email protected]> wrote: >> >>> Some pseudowrapping bugs cleared up (including one that *created* a >>> seam in the middle - oops!) - now at version 0.31beta. >>> >>> Windows binaries: >>> http://horman.net/multiblend/**multiblend0.31beta.zip<http://horman.net/multiblend/multiblend0.31beta.zip> >>> Source for Linux/Mac/FreeBSD etc: http://horman.net/multiblend/** >>> multiblend0.31beta.tar.gz<http://horman.net/multiblend/multiblend0.31beta.tar.gz> >>> >>> David >>> >>> >>> >>> On Tuesday, March 27, 2012 12:01:56 AM UTC+1, Monkey wrote: >>> >>>> Hi all, >>>> >>>> Previous discussion: http://groups.google.com/**group** >>>> /hugin-ptx/browse_thread/**threa**d/b08211b2659a7eab<http://groups.google.com/group/hugin-ptx/browse_thread/thread/b08211b2659a7eab> >>>> >>>> I've implemented a feature I'm calling "pseudowrapping" to blend >>>> around the left/right borders for 360 degree panoramas, a much- >>>> requested feature. >>>> >>>> There are no command-line switches to access this; pseudowrapping is >>>> enabled any time multiblend is run with only a single uncropped TIFF >>>> (such as that already created by a normal multiblend run) as its >>>> input. >>>> >>> -- >>> 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<http://wiki.panotools.org/Hugin_FAQ> >>> To post to this group, send email to [email protected] >>> To unsubscribe from this group, send email to hugin-ptx+unsubscribe@** >>> googlegroups.com <hugin-ptx%[email protected]> >>> For more options, visit this group at http://groups.google.com/** >>> group/hugin-ptx <http://groups.google.com/group/hugin-ptx> >>> >> >> > On Thursday, April 12, 2012 4:47:46 PM UTC+1, Evgeny Toropov wrote: >> >> Hi, David, >> >> How much memory does it need for blending with L levels? >> I have a huge wide cylindrical panorama 72076x2278. >> When all the images are loaded, multiblend consumes up to 1.4 GB out of >> my total 4 GB. >> The program only works with 1 level (-l 1). Otherwise, I get the >> following error: >> >> " >> 72076x2278, 8bpp, 2 levels >> split done >> seaming... >> masks... >> blending... >> Error: not enough memory for output channel! >> " >> >> When 1 level option is set multiblend takes ~4 minutes to finish (almost >> all is time to load nona tiffs), while enblend (also with 'levels' set to >> 1) takes 140 min. >> >> >> >> >> On Tue, Mar 27, 2012 at 2:11 PM, Monkey <[email protected]> wrote: >> >>> Some pseudowrapping bugs cleared up (including one that *created* a >>> seam in the middle - oops!) - now at version 0.31beta. >>> >>> Windows binaries: >>> http://horman.net/multiblend/**multiblend0.31beta.zip<http://horman.net/multiblend/multiblend0.31beta.zip> >>> Source for Linux/Mac/FreeBSD etc: http://horman.net/multiblend/** >>> multiblend0.31beta.tar.gz<http://horman.net/multiblend/multiblend0.31beta.tar.gz> >>> >>> David >>> >>> >>> >>> On Tuesday, March 27, 2012 12:01:56 AM UTC+1, Monkey wrote: >>> >>>> Hi all, >>>> >>>> Previous discussion: http://groups.google.com/**group** >>>> /hugin-ptx/browse_thread/**threa**d/b08211b2659a7eab<http://groups.google.com/group/hugin-ptx/browse_thread/thread/b08211b2659a7eab> >>>> >>>> I've implemented a feature I'm calling "pseudowrapping" to blend >>>> around the left/right borders for 360 degree panoramas, a much- >>>> requested feature. >>>> >>>> There are no command-line switches to access this; pseudowrapping is >>>> enabled any time multiblend is run with only a single uncropped TIFF >>>> (such as that already created by a normal multiblend run) as its >>>> input. >>>> >>> -- >>> 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<http://wiki.panotools.org/Hugin_FAQ> >>> To post to this group, send email to [email protected] >>> To unsubscribe from this group, send email to hugin-ptx+unsubscribe@** >>> googlegroups.com <hugin-ptx%[email protected]> >>> For more options, visit this group at http://groups.google.com/** >>> group/hugin-ptx <http://groups.google.com/group/hugin-ptx> >>> >> >> > On Thursday, April 12, 2012 4:47:46 PM UTC+1, Evgeny Toropov wrote: >> >> Hi, David, >> >> How much memory does it need for blending with L levels? >> I have a huge wide cylindrical panorama 72076x2278. >> When all the images are loaded, multiblend consumes up to 1.4 GB out of >> my total 4 GB. >> The program only works with 1 level (-l 1). Otherwise, I get the >> following error: >> >> " >> 72076x2278, 8bpp, 2 levels >> split done >> seaming... >> masks... >> blending... >> Error: not enough memory for output channel! >> " >> >> When 1 level option is set multiblend takes ~4 minutes to finish (almost >> all is time to load nona tiffs), while enblend (also with 'levels' set to >> 1) takes 140 min. >> >> >> >> >> On Tue, Mar 27, 2012 at 2:11 PM, Monkey <[email protected]> wrote: >> >>> Some pseudowrapping bugs cleared up (including one that *created* a >>> seam in the middle - oops!) - now at version 0.31beta. >>> >>> Windows binaries: >>> http://horman.net/multiblend/**multiblend0.31beta.zip<http://horman.net/multiblend/multiblend0.31beta.zip> >>> Source for Linux/Mac/FreeBSD etc: http://horman.net/multiblend/** >>> multiblend0.31beta.tar.gz<http://horman.net/multiblend/multiblend0.31beta.tar.gz> >>> >>> David >>> >>> >>> >>> On Tuesday, March 27, 2012 12:01:56 AM UTC+1, Monkey wrote: >>> >>>> Hi all, >>>> >>>> Previous discussion: http://groups.google.com/**group** >>>> /hugin-ptx/browse_thread/**threa**d/b08211b2659a7eab<http://groups.google.com/group/hugin-ptx/browse_thread/thread/b08211b2659a7eab> >>>> >>>> I've implemented a feature I'm calling "pseudowrapping" to blend >>>> around the left/right borders for 360 degree panoramas, a much- >>>> requested feature. >>>> >>>> There are no command-line switches to access this; pseudowrapping is >>>> enabled any time multiblend is run with only a single uncropped TIFF >>>> (such as that already created by a normal multiblend run) as its >>>> input. >>>> >>> -- >>> 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<http://wiki.panotools.org/Hugin_FAQ> >>> To post to this group, send email to [email protected] >>> To unsubscribe from this group, send email to hugin-ptx+unsubscribe@** >>> googlegroups.com <hugin-ptx%[email protected]> >>> For more options, visit this group at http://groups.google.com/** >>> group/hugin-ptx <http://groups.google.com/group/hugin-ptx> >>> >> >> -- > 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 > -- 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
