From mihaiv at gmail.com Tue Nov 4 10:23:22 2008 From: mihaiv at gmail.com (Mihai Varzaru) Date: Tue Nov 4 10:23:41 2008 Subject: [ale] Ale compilation Message-ID: <8a5752f30811041023n71ccf27g3ee894e95ac4ee15@mail.gmail.com> I compiled ale 0.8.11 on Ubuntu 8.10 (gcc 4.3.2) and I encountered a compilation error: In file included from ui/ui_log.cc:22: > ui/ui_log.h: In member function 'virtual void ui_log::printf(const char*, > ...)': > ui/ui_log.h:41: error: 'strlen' was not declared in this scope This was easy to fix by including string.h but it is a problem that in a standard configuration like this ale 0.8.11 does not compile by default. I also tried ale 0.9.0.1 on the same machine. All the help pages do not show up and I receive instead an error message: > exec: 1: -a: not found This happens both with the distribution package and a local compiled version. From dhilvert at gmail.com Tue Nov 4 12:55:01 2008 From: dhilvert at gmail.com (David Hilvert) Date: Tue Nov 4 12:55:41 2008 Subject: [ale] Ale compilation In-Reply-To: <8a5752f30811041023n71ccf27g3ee894e95ac4ee15@mail.gmail.com> References: <8a5752f30811041023n71ccf27g3ee894e95ac4ee15@mail.gmail.com> Message-ID: <20081104215501.7728c694.dhilvert@gmail.com> On Tue, 4 Nov 2008 20:23:22 +0200 "Mihai Varzaru" wrote: > I compiled ale 0.8.11 on Ubuntu 8.10 (gcc 4.3.2) and I encountered a > compilation error: > > In file included from ui/ui_log.cc:22: > > ui/ui_log.h: In member function 'virtual void ui_log::printf(const char*, > > ...)': > > ui/ui_log.h:41: error: 'strlen' was not declared in this scope > > > This was easy to fix by including string.h but it is a problem that in a > standard configuration like this ale 0.8.11 does not compile by default. > > > I also tried ale 0.9.0.1 on the same machine. All the help pages do not show > up and I receive instead an error message: > > > exec: 1: -a: not found > > This happens both with the distribution package and a local compiled > version. I've pushed possible fixes to branches 0.8.11-branch and 0.9.0-branch at http://repo.or.cz/w/Ale.git The -a failure is a bashism. The fix in 0.9.0-branch is to explicitly call bash, but better solutions probably exist. (A similar, but slightly different, patch had already been applied to the master branch.) From dhilvert at gmail.com Wed Nov 5 03:28:09 2008 From: dhilvert at gmail.com (David Hilvert) Date: Wed Nov 5 03:28:40 2008 Subject: [ale] 0.8.11.1 and 0.9.0.2 Message-ID: <20081105122809.181aa582.dhilvert@gmail.com> Releases 0.8.11.1 and 0.9.0.2 are now available for download. URLs: http://auricle.dyndns.org/ALE/download/ale-0.8.11.1.tar.gz http://auricle.dyndns.org/ALE/download/ale-0.9.0.2.tar.gz Overview: 0.8.11.1 fixes certain compilation errors under recent GCC versions, and includes other minor fixes in the documentation tree. 0.9.0.2 has been updated to fix a bashism in the wrapper script, and includes other minor fixes to the documentation tree. 0.8.11.1 Changelog: o Fix build for Ubuntu 8.10, after this was reported by Mihai Varzaru as broken. o Update documentation Makefile for better Links compatibility. 0.9.0.2 Changelog: o Update documentation makefile for better links compatibility. o Fix 'exec -a' bashism in ale script, after this was reported by Mihai Varzaru as being broken in Ubuntu 8.10. From mihaiv at gmail.com Wed Nov 5 03:33:59 2008 From: mihaiv at gmail.com (Mihai Varzaru) Date: Wed Nov 5 03:34:14 2008 Subject: [ale] Super resolution problems Message-ID: <8a5752f30811050333o7477b29dg12705fd943296c6e@mail.gmail.com> I've taken photos of the moon (with Canon Powershot A710 camera) and used ale 0.8.11 to generate a 2x scaled image. I noticed very good noise reduction but I can't see additional detail. I used 64 different images of the moon and --scale=2 --ips 4 as options. All images are slightly displaced as I made them hand-held. I also tried just 10 different images and had an identical result. Here is one of the original images: http://picasaweb.google.com/mihaiv/Moon#5265128137336941266 Here is the resulting image as generated by ale: http://picasaweb.google.com/mihaiv/Moon#5265128136949737730 What can I do to get additional details? From dhilvert at gmail.com Thu Nov 6 02:15:47 2008 From: dhilvert at gmail.com (David Hilvert) Date: Thu Nov 6 02:16:10 2008 Subject: [ale] Super resolution problems In-Reply-To: <8a5752f30811050333o7477b29dg12705fd943296c6e@mail.gmail.com> References: <8a5752f30811050333o7477b29dg12705fd943296c6e@mail.gmail.com> Message-ID: <20081106111547.161dcaa5.dhilvert@gmail.com> On Wed, 5 Nov 2008 13:33:59 +0200 "Mihai Varzaru" wrote: > I've taken photos of the moon (with Canon Powershot A710 camera) and used > ale 0.8.11 to generate a 2x scaled image. I noticed very good noise > reduction but I can't see additional detail. Using a --bayer value appropriate for your camera (and image orientation) will probably help. If this option is not used, then the interpolated values imagined by your camera will be treated as actual values. Also, please ensure that all of the images have been well aligned; this can be particularly challenging in the case of the moon (due to its shape), which is partly what motivated the changes (multi-alignment) in the 0.9.0 series. --lod-max=-3 (or more positive; e.g., --lod-max=-2) may help alignment in the case of 0.8.11, which is configured by default to use a very optimistic level-of-detail (--lod-max=-4). (One way to check alignment is to use the --visp option to output a sequence of images, perhaps using a renderer such as last:triangle:2 to capture roughly the position of each frame.) > I used 64 different images of the moon and --scale=2 --ips 4 as options. All > images are slightly displaced as I made them hand-held. I also tried just 10 > different images and had an identical result. Additional images provide diminishing returns. There are graphs illustrating the general idea here (but whether the metric chosen is exactly right, I don't know): http://www.macs.hw.ac.uk/texture2005/programme/papers/014.pdf