"Fossies" - the Fresh Open Source Software Archive

Member "blender-2.93.1/intern/libmv/ChangeLog" (22 Apr 2021, 16468 Bytes) of package /linux/misc/blender-2.93.1.tar.xz:


As a special service "Fossies" has tried to format the requested text file into HTML format (style: standard) with prefixed line numbers. Alternatively you can here view or download the uninterpreted source code file.

    1 commit efd7a93317e0278b99e66785f667823e451daef1
    2 Author: Sergey Sharybin <sergey.vfx@gmail.com>
    3 Date:   Tue May 9 10:16:42 2017 +0200
    4 
    5     Fix strict compiler warnings, unused variables
    6 
    7 commit 8efd47e13dfdd3f7209bc96f26d0b13127dd6376
    8 Author: Sergey Sharybin <sergey.vfx@gmail.com>
    9 Date:   Wed Dec 14 10:44:57 2016 +0100
   10 
   11     Fix T50243: libmv_panography_test is broken
   12     
   13     There was fully wrong logic in comparison: was actually accessing memory
   14     past the array boundary. Run test manually and the figure seems correct
   15     to me now.
   16     
   17     Spotted by @LazyDodo, thanks!
   18 
   19 commit 6dfb9cd1bd14669d84be789000ce234747fb00ff
   20 Author: Sergey Sharybin <sergey.vfx@gmail.com>
   21 Date:   Thu Jul 14 11:49:38 2016 +0200
   22 
   23     Fix some strict compiler warnings
   24     
   25     One of them was a real bug!
   26 
   27 commit f61adaecf7b29ebe6677be0e1c825f0a8d475e4b
   28 Author: Sergey Sharybin <sergey.vfx@gmail.com>
   29 Date:   Wed May 31 11:22:34 2017 +0200
   30 
   31     Enable explicit schur complement for BA step
   32     
   33     This is something we do in Blender and only reason it was not
   34     enabled for standalone Libmv is because we did not have fresh
   35     enough version of Ceres bundled.
   36 
   37 commit fc5d3a1d4880c6658aff693c1c1e8c10c96ce1a7
   38 Author: Sergey Sharybin <sergey.vfx@gmail.com>
   39 Date:   Wed Nov 2 15:32:11 2016 +0100
   40 
   41     Update tests to make tests pass after recent Ceres update
   42     
   43     Just a precision issue, difference is around 1e-7. Should be fine to
   44     simply update expected value.
   45 
   46 commit e1ac9f6124110c1a90d8e417bea47acfcbdcca42
   47 Author: Sergey Sharybin <sergey.vfx@gmail.com>
   48 Date:   Wed May 31 10:54:48 2017 +0200
   49 
   50     Update Ceres to latest release 1.12.0
   51 
   52 commit ac1571352b4962f110929b963f8616d7310ceea5
   53 Author: Sergey Sharybin <sergey.vfx@gmail.com>
   54 Date:   Fri Apr 7 17:10:44 2017 +0200
   55 
   56     Fix crash of keyframe selection on 32bit linux
   57 
   58 commit 5f8df3da965686df39a6ae5c9f17482075017bf4
   59 Author: Sergey Sharybin <sergey.vfx@gmail.com>
   60 Date:   Tue Jan 19 14:00:53 2016 +0500
   61 
   62     Solve some strict warnings in tests
   63 
   64 commit 8ea3a5d752a9ce3337ab7643897472a4d33747f1
   65 Author: Brecht Van Lommel <brechtvanlommel@gmail.com>
   66 Date:   Sat Feb 18 23:52:31 2017 +0100
   67 
   68     Fix a few compiler warnings with macOS / clang.
   69 
   70 commit ffbe81461770e70736e80b8cab8e6eb1f8b27160
   71 Author: Mike Erwin <significant.bit@gmail.com>
   72 Date:   Wed May 31 10:43:08 2017 +0200
   73 
   74     Fix comparison of identicals
   75     
   76     Some of these check that dimensions match before running code that
   77     assumes they do match.
   78     
   79     Found with PVS-Studio T48917.
   80 
   81 commit 206c01999cde16c1c6c43a8e13ffa86020821d98
   82 Author: Sergey Sharybin <sergey.vfx@gmail.com>
   83 Date:   Wed May 31 10:39:16 2017 +0200
   84 
   85     Add basic track masking API in place
   86     
   87     This brings back ability to mask non-interesting parts of
   88     specific track (the feature got lost with new auto-track API).
   89     
   90     Added it back by extending frame accessor class. This isn't really
   91     a frame thing, but we don't have other type of accessor here.
   92     
   93     Surely, we can use old-style API here and pass mask via region
   94     tracker options for this particular case, but then it becomes much
   95     less obvious how real auto-tracker will access this mask with old
   96     style API.
   97     
   98     So seems we do need an accessor for such data, just matter of
   99     finding better place than frame accessor.
  100 
  101 commit faa069cb826892780356477cc10602390fecf06b
  102 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  103 Date:   Wed May 31 10:36:26 2017 +0200
  104 
  105     Tests: Tweak epsilon to avoid what looks a false-positive failure
  106 
  107 commit 7c84e45c1d330871477ba3516f57178e5b9d101f
  108 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  109 Date:   Wed May 31 10:15:43 2017 +0200
  110 
  111     CMake: Fix mistake in closing branch
  112 
  113 commit cb769a0d319a8c95948153d78a4c3378a0142ece
  114 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  115 Date:   Thu Jul 21 12:52:33 2016 +0200
  116 
  117     Set of fixes for MSVC215
  118     
  119     - Move GLOG/GFLAGS defines to a more global scope,
  120       this way ANY of our own libraries will use proper
  121       declspec.
  122     
  123     - Compile png/zlib/openexif on Windows as well since
  124       those are required for a correct linking.
  125 
  126 commit bb95c8654fd2cea72d66ed04cd825cc3712ea804
  127 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  128 Date:   Wed Jul 20 18:14:46 2016 +0200
  129 
  130     Disable unexisting Ceres option
  131     
  132     Explicit Schur complement requires having
  133     newer Ceres than we currently have bundled.
  134 
  135 commit a2e12c959ef32cc9382244d1581992c2f7aa9c09
  136 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  137 Date:   Wed Jul 20 18:04:57 2016 +0200
  138 
  139     Various fixes for MSVC
  140     
  141     - Update Eigen to 3.2.7 since this brings crucial
  142       fixes for MSVC 2015.
  143     
  144     - Switch to STATIC build by default.
  145     
  146       There are issues building current sources as dynamic
  147       libraries with MSVC2015 and additionally building
  148       dynamic Ceres is not recommended anyway, so let's
  149       not do this for the time being.
  150     
  151     If anyone finds a way to make this all working --
  152     it'llsurely be a welcome addition.
  153 
  154 commit 7a676106720fb126a27ff010abdd8bb65d7e0d9a
  155 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  156 Date:   Mon Jan 4 18:30:12 2016 +0500
  157 
  158     Fix strict compiler warnings
  159 
  160 commit d3537e3709fe11f42312e82cb1c9837c9e742385
  161 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  162 Date:   Sun Jan 3 14:17:10 2016 +0500
  163 
  164     GLog/GFlags: Reduce difference between upstream and bundled versions
  165     
  166     Several things here:
  167     
  168     - Re-bundled sources using own fork with pull-requests applied on the sources.
  169     
  170     - Got rid of changes around include "config.h", it was needed by Blender to
  171       make it's include directories configuration to work. This could be addressed
  172       differently from Blender side.
  173     
  174     - Moved some customization to defines set by CMakeLists.
  175 
  176 commit 1ec37bba2cfbbf0d6568429fa3035ee2164c23e6
  177 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  178 Date:   Sat Jan 2 12:42:55 2016 +0500
  179 
  180     GFlags linking errors fix for MSVC
  181 
  182 commit df7642b270e8e43685e9ffb404b59d7b226a9f60
  183 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  184 Date:   Thu Dec 31 17:56:12 2015 +0500
  185 
  186     Alternative fix for missing prototype for a couple of functions
  187 
  188 commit 08f685797b7d776cdaa579136c82b15ddc6ffb30
  189 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  190 Date:   Thu Dec 31 17:33:05 2015 +0500
  191 
  192     Update GFlags to the latest upstream version
  193     
  194     Makes it easier to synchronize some compiler/warning fixes.
  195 
  196 commit e0ef5b09203e3906a555e6c2010f25cb667da9cd
  197 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  198 Date:   Thu Dec 31 16:15:59 2015 +0500
  199 
  200     GLog: Solve some compilation warnings
  201     
  202     Those are actually sent to a pull-request, see
  203     
  204       https://github.com/google/glog/pull/81
  205 
  206 commit 2072b213d4d3a55d099a063ed1e7331cc773454e
  207 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  208 Date:   Thu Dec 31 16:13:53 2015 +0500
  209 
  210     Add Martijn Berger to the AUTHORS file
  211 
  212 commit 4dd0770d98d67896e4f936356e281f63d927410e
  213 Author: Martijn Berger <martijn.berger@gmail.com>
  214 Date:   Thu Dec 31 16:13:08 2015 +0500
  215 
  216     Fix compilation error of Glog and Gflags with MSVC2015
  217 
  218 commit 2712f42be2ad79e7d3a6c9905f6d8d1e3b7133ac
  219 Author: Brecht Van Lommel <brechtvanlommel@gmail.com>
  220 Date:   Thu Dec 31 14:00:58 2015 +0500
  221 
  222     Fix OS X (with 10.11 SDK) glog build errors due to using deprecated code.
  223     
  224     Some values are now restored to the ones from before the upgrade today.
  225 
  226 commit d249280fdf7c937fd6ebbc465508843a70aafd4c
  227 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  228 Date:   Wed Dec 30 16:59:28 2015 +0500
  229 
  230     Tweaks to Glog to support building on all platforms
  231     
  232     This makes it possible to compile Libmv on all platforms,
  233     amount of hacks is lower, which could bring some warnings
  234     up, but those are better be addressed via upstream which
  235     is now rather active.
  236 
  237 commit 86c57750ddb857643fb5dd2c83b4953da83dd57d
  238 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  239 Date:   Wed Dec 30 16:15:47 2015 +0500
  240 
  241     Enable explicit Schur complement matrix by default
  242     
  243     Gives up to 2x speed up of camera solving process in average scene.
  244     In the really huge one it might be slower, but that we need to investigate.
  245 
  246 commit d6c52a70b5a0664b7c74bda68f59a895fe8aa235
  247 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  248 Date:   Wed Dec 30 16:13:03 2015 +0500
  249 
  250     Fix one frame memory leak when tracking last frame
  251 
  252 commit 6e2ac41d25d5923b2a62c96d27d919a36eff9b48
  253 Author: Brecht Van Lommel <brechtvanlommel@gmail.com>
  254 Date:   Wed Dec 30 16:11:24 2015 +0500
  255 
  256     Motion tracking not workig with Xcode 7 on OS X.
  257     
  258     Caused by use of the uninitialized shape_ variable in Resize().
  259 
  260 commit fc72ae06fb4ae559ac37d14d1b34d6669505cc86
  261 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  262 Date:   Wed Dec 30 15:56:40 2015 +0500
  263 
  264     Update GLog to latest upstream
  265     
  266     Should fix issues building with MSVC2015.
  267 
  268 commit d4b2d15bd3d195074b074331354de96a1b51042f
  269 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  270 Date:   Wed Dec 30 16:01:10 2015 +0500
  271 
  272     Fix wrong README file reference
  273 
  274 commit 2b4aa0b7720cae9a408284834559bea9960157ee
  275 Author: Keir Mierle <mierle@gmail.com>
  276 Date:   Mon May 11 02:16:53 2015 -0700
  277 
  278     Make README more informative for GitHub viewers
  279     
  280     Reviewers: sergey
  281     
  282     Reviewed By: sergey
  283     
  284     Differential Revision: https://developer.blender.org/D1295
  285 
  286 commit 514e4491aea655d20be047ed87f002fb7854d5c9
  287 Author: Keir Mierle <mierle@gmail.com>
  288 Date:   Mon May 11 01:54:09 2015 -0700
  289 
  290     Simplify the modal solver Ceres cost function
  291     
  292     Fix test by flipping the quaternion.
  293     
  294     Reviewers: sergey
  295     
  296     Reviewed By: sergey
  297     
  298     Projects: #libmv
  299     
  300     Differential Revision: https://developer.blender.org/D756
  301 
  302 commit e55fafd31f7d53d42af7c6b7df2eebe3c2568da9
  303 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  304 Date:   Wed Dec 31 19:05:51 2014 +0500
  305 
  306     Synchronize MSVC compilation fixes from Blender
  307 
  308 commit 7d6020d2ec42c6cb2749bc891186b4880d26d40b
  309 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  310 Date:   Wed Dec 31 15:32:07 2014 +0500
  311 
  312     Update GLog to latest upstream revision 143
  313     
  314     Mainly to solve compilation error with demangle.cc.
  315 
  316 commit 5dc746700eaf85cb674f0fb73ff3c1b49a7f6315
  317 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  318 Date:   Fri Dec 12 14:59:55 2014 +0500
  319 
  320     Update GFlags to latest release 2.1.1
  321     
  322     Main purpose of this (andsome of upcoming) update is to check if the
  323     upstream sources are useable without any modifications for us. And if
  324     not, then we'll need to consider moving some changes into upstream.
  325     
  326     This commit contains an one-to-one copy of the upstream GFlags library
  327     and also changes namespace usage since it's changed in the upstream.
  328 
  329 commit 6fe6d75f7e90e161b44643b953f058a3829a5247
  330 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  331 Date:   Sat Nov 1 02:53:36 2014 +0500
  332 
  333     Libmv: Code cleanup, mixed class/struct in declaration/definition
  334 
  335 commit d2a5f7953812d2d09765431b59c6c4ac72faf35b
  336 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  337 Date:   Thu Oct 30 23:13:53 2014 +0500
  338 
  339     Libmv: Support disabled color channels in tracking settings
  340     
  341     This was never ported to a new tracking pipeline and now it's done using
  342     FrameAccessor::Transform routines. Quite striaghtforward, but i've changed
  343     order of grayscale conversion in blender side with call of transform callback.
  344     
  345     This way it's much easier to perform rescaling in libmv side.
  346 
  347 commit d976e034cdf74b34860e0632d7b29713f47c5756
  348 Author: Keir Mierle <mierle@gmail.com>
  349 Date:   Sat Aug 23 00:38:01 2014 -0700
  350 
  351     Minor keyframe selection cleanups
  352     
  353     Reviewers: sergey
  354     
  355     Reviewed By: sergey
  356     
  357     Differential Revision: https://developer.blender.org/D757
  358 
  359 commit bc99ca55dadfca89fde0f93764397c2fe028943d
  360 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  361 Date:   Sat Aug 23 01:55:32 2014 +0600
  362 
  363     implement backward prediction
  364     
  365     The title actually says it all, just extend current implementation
  366     of PredictMarkerPosition() to cases when tracking happens in the reverse
  367     order (from the end frame to start).
  368     
  369     it's still doesn't solve all the ambiguity happening in the function
  370     in cases when one tracks the feature and then re-tracks it in order
  371     to refine the sliding. This is considered a separate TODO for now and
  372     will likely be solved by passing tracking direction to the prediction
  373     function.
  374     
  375     Reviewers: keir
  376     
  377     Reviewed By: keir
  378     
  379     Differential Revision: https://developer.blender.org/D663
  380 
  381 commit 5b87682d98df65ade02638bc6482d824cf0dd0b3
  382 Author: Keir Mierle <mierle@gmail.com>
  383 Date:   Thu Aug 21 22:45:22 2014 -0700
  384 
  385     Make libmv compile on Ubuntu 14.04
  386     
  387     Reviewers: fsiddi
  388     
  389     Reviewed By: fsiddi
  390     
  391     Subscribers: sergey
  392     
  393     Differential Revision: https://developer.blender.org/D755
  394 
  395 commit 0a81db623c458e0384b4f7060d1bcff8993fb469
  396 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  397 Date:   Wed Jul 23 00:42:00 2014 +0600
  398 
  399     Fix wrong residual blocks counter
  400     
  401     This happened in cases when having zero-weighted tracks
  402     and could lead to some assert failures on marking parameter
  403     block constant.
  404 
  405 commit 2824dbac54cacf74828678be7a5c9fd960ce83e2
  406 Author: Sergey Sharybin <sergey.vfx@gmail.com>
  407 Date:   Fri Jul 18 12:52:03 2014 +0600
  408 
  409     Fix search area sliding issue
  410     
  411     The only way to do this is to store search region in floats
  412     and round when we need to sample it. Otherwise you'll always
  413     have sliding effect caused by rounding the issues, especially
  414     when doing incremental offset (thing which happens in the
  415     prediction code).
  416     
  417     Pretty much straightforward change apart from stuff to be kept
  418     in mind: offset calculation int should happen relative to the
  419     rounded search region. This is because tracker works in the space
  420     of the search window image which get's rounded on the frame access,
  421     
  422     This makes API a bit creepy because frame accessor uses the same
  423     Region struct as the search window in Marker and ideally we would
  424     need to have either IntRegion or Region<int> in order to make
  425     Libmv fully track on what's getting rounded and when.
  426     
  427     Reviewers: keir
  428     
  429     Reviewed By: keir
  430     
  431     Differential Revision: https://developer.blender.org/D616
  432 
  433 commit 04862c479332308be47a0f27361402444ace8880
  434 Author: Keir Mierle <mierle@gmail.com>
  435 Date:   Fri May 9 23:00:03 2014 +0200
  436 
  437     Start the automatic 2D tracking code
  438     
  439     This starts the 2D automatic tracking code. It is totally unfinished.
  440     
  441     Reviewers: sergey
  442     
  443     Reviewed By: sergey
  444     
  445     Differential Revision: https://developer.blender.org/D523
  446 
  447 commit be679f67d807a2139c1f7d7e2ca45141940b30d5
  448 Author: Keir Mierle <mierle@gmail.com>
  449 Date:   Fri May 9 14:36:04 2014 +0200
  450 
  451     Also shift the search window
  452     
  453     Reviewers: sergey
  454     
  455     Reviewed By: sergey
  456     
  457     Differential Revision: https://developer.blender.org/D520
  458 
  459 commit 66b8f5eef2633ebcde32a388fc14c60171011821
  460 Author: Keir Mierle <mierle@gmail.com>
  461 Date:   Fri May 9 13:06:28 2014 +0200
  462 
  463     Change the search region to absolute frame coordinates
  464     
  465     Smarter Eigen usage
  466     
  467     Better error logging
  468     
  469     Reviewers: sergey
  470     
  471     Reviewed By: sergey
  472     
  473     Differential Revision: https://developer.blender.org/D519
  474 
  475 commit a08193319ae409fad8f08887eae1f79f02e91eaa
  476 Author: Keir Mierle <mierle@gmail.com>
  477 Date:   Fri May 9 12:02:47 2014 +0200
  478 
  479     First cut at predictive tracing
  480     
  481     This adds a Kalman filter-based approach to predict where a marker
  482     will go in the next frame to track. Hopefully this will make the
  483     tracker work faster by avoiding lengthy searches. This code
  484     compiles, but is otherwise untested, and likely does not work.
  485     
  486     Fix else branch
  487     
  488     Add some tests
  489     
  490     Update patch coordinates as well (and test)
  491     
  492     Reviewers: sergey
  493     
  494     Reviewed By: sergey
  495     
  496     Differential Revision: https://developer.blender.org/D518
  497 
  498 commit 607ffb2f62b56e34a841abbb952d83e19cd1e23c
  499 Author: Keir Mierle <mierle@gmail.com>
  500 Date:   Thu May 8 16:05:28 2014 +0200
  501 
  502     Add constructor to AutoTrack
  503 
  504 commit c39e20a0c27da3733804c3848454b5d4c4f0e66b
  505 Author: Keir Mierle <mierle@gmail.com>
  506 Date:   Thu May 8 16:04:20 2014 +0200
  507 
  508     Fix GetMarker compilation issue
  509 
  510 commit 8dd93e431b6e44439c803bfd26ec2669b656177e
  511 Author: Keir Mierle <mierle@gmail.com>
  512 Date:   Thu May 8 15:50:26 2014 +0200
  513 
  514     Expose GetMarker() in AutoTrack
  515     
  516     Reviewers: sergey
  517     
  518     Reviewed By: sergey
  519     
  520     Differential Revision: https://developer.blender.org/D516