]>
Commit | Line | Data |
---|---|---|
1 | 1.0 = 0.98 (just because of API changes the version number was bumped | |
2 | 0.98... | |
3 | small measurement fields added that only scan around mean | |
4 | diagnostics in show=1 and show=2 use lines now. | |
5 | fix zoom detection/treatment | |
6 | zoomspeed param for optzoom | |
7 | gaussian filter for cam path optimization | |
8 | 0.97 | |
9 | optzoom = 2: adaptive zooming added | |
10 | border bluring for interpolation>=bilinear | |
11 | spurious black pixels bug fixed | |
12 | 0.96 | |
13 | libavfilter plugins work | |
14 | flexible support of pixel formats, ala ffmpeg | |
15 | 0.95 | |
16 | virtual tripod mode | |
17 | ||
18 | 0.94 | |
19 | local motions are now stored in the file (file format changed, | |
20 | old one still supported) | |
21 | BUGFIX for zero-transforms | |
22 | ||
23 | 0.93 | |
24 | single pass version filter_deshake | |
25 | made stabilize part print the transforms to the file right away | |
26 | ||
27 | 0.92 | |
28 | boxblur added and unsharp filter removed for motiondetection | |
29 | ||
30 | 0.91 | |
31 | spiral search for matches in compareSubImg | |
32 | ||
33 | 0.90 | |
34 | complete restructuring into own library | |
35 | Fixed point arithmetic used for transform code | |
36 | motiondetection optimized with ORC: speedup by factor 4 | |
37 | ORC 0.4.14 is now required | |
38 | added tests for speed and functional tests | |
39 | removed biquadratic interpolation routine | |
40 | ||
41 | 0.80 | |
42 | keep borders with stabilized last frame (as it should be). | |
43 | Pointed out by Guido Torelli. | |
44 | ||
45 | 0.79 | |
46 | speed optimizations using SSE by Alexey Osipov | |
47 | search tree cut, spiral search and sse: | |
48 | together speedup factor ~8 of stabilize run | |
49 | stepsize is increased stepwise | |
50 | keep border at transform plugin improved | |
51 | ||
52 | ||
53 | 0.77 | |
54 | interpolation routines improved a lot, thanks to hints on ffmpeg list | |
55 | added bicubic interpolation that uses 4x4 pixel (useful for large zoom values) | |
56 | bilinear interpolation is now the default | |
57 | transform plugin uses last transform for the remaining frames | |
58 | -> this enables to use the transform plugin for constant transformations | |
59 | ||
60 | 0.76 | |
61 | BUGFIX in calcFieldTransYUV caused SEGFAULT | |
62 | ||
63 | 0.75 | |
64 | two meta parameter introduces: shakiness and accuracy and | |
65 | removal of fieldnum, fielsize, maxshift, maxfieldnum | |
66 | field placement changed: Now they fill the frame perfectly | |
67 | added bluring such that stepsize can be much larger: much faster now! | |
68 | linear interpolation is now the default | |
69 | uncertain angles are set to 0 (more robustness) | |
70 | ||
71 | 0.70 selects a maximal number of fields according to their contrast | |
72 | nicely distributed over the frame | |
73 | Todo: remove fieldnum, allowmax (now false), and compressed parameter | |
74 | Todo: add accuracy parameter instead of maxfields | |
75 | Bugfix of contrast routine (error in the calculation of maxi) | |
76 | ||
77 | 0.62 | |
78 | different interpolation functions. No interpolation is default and | |
79 | seems fine in most applications. | |
80 | field placement changed. Now they fill the frame better. | |
81 | "compress" option moves them more to the center (vertically) | |
82 | fields and their areas and transforms can be drawn on the frame. | |
83 | ||
84 | 0.6 | |
85 | new field placement allows for any number of fields | |
86 | and respects aspect ratio | |
87 | fields with low contrast are ignored | |
88 | non-symmetric field setups are possible | |
89 | ||
90 | 0.5 | |
91 | global and per-frame zooming supported | |
92 | optimal zoom value is calculated to avoid moving border effect | |
93 | add filter unsharp to sharpen the frames | |
94 | write parameters to transforms file | |
95 | ||
96 | 0.44 | |
97 | bugfixing | |
98 | finished RGB #support | |
99 | transforms file format supports comments | |
100 | ||
101 | 0.42 | |
102 | input filename and tc_list stuff used | |
103 | 0.41 | |
104 | code style from transcode used | |
105 | 0.4 | |
106 | moved to CVS version we use a cleaned median to select the right | |
107 | transform from the fields | |
108 | 0.3 | |
109 | Support RGB (untestet) | |
110 | maxshift default 50 | |
111 | allowmax added | |
112 | scan only every second shift and then make a second small | |
113 | scan around the best match | |
114 | removed black rectangle around transformed frame |