1 Movit 1.5.3, August 10th, 2017
6 Movit 1.5.2, July 5th, 2017
8 - The texture generation for ResampleEffect (when the size is changed)
11 - Work around an NVIDIA driver threading bug by aggressively caching VAOs.
14 Movit 1.5.1, May 29th, 2017
16 - Allow YCbCrInput to change input format after finalize.
18 - Some minor YCbCrInput bugfixes.
21 Movit 1.5.0, March 21st, 2017
23 - Support interleaved Y'CbCr input (4:4:4 in a single texture).
25 - Support 10-bit and 12-bit Y'CbCr, both for input and output. For planar,
26 these are supported packed in 16-bit ints; for interleaved, 10:10:10:2 is
27 supported. (Efficient conversion to and from v210, ie. 10-bit 4:2:2,
28 is possible using compute shaders, but Movit does not include support
29 for them at the current point.) Note that this now means the num_levels
30 flag in YCbCrFormat actually matters, although 0 will be interpreted
31 as 256 (8-bit) for the benefit of older applications.
33 - Limited support for having multiple Y'CbCr outputs from a chain.
35 - Allow changing the Y'CbCr output coefficients runtime, ie., after finalize.
37 - Fix an issue where the last pass would have been rendered with the sRGB
38 flag set, which confused Qt applications running in certain NVIDIA
42 Movit 1.4.0, November 5th, 2016
44 - Allow setting the intermediate format for chains, instead of hardcoding
45 it at 16-bit RGBA; advanced users can use this to e.g. ask for 8-bit
46 sRGB intermediates, reducing the amount of memory bandwidth needed
47 at the cost of reduced precision. Whether this is a good tradeoff or not
48 depends on the exact chain and your requirements.
50 - Fix an issue where a (cached) shader program could be used from multiple
51 threads at a time, causing the uniforms to contain unpredictable values.
53 - Make the error printed on check_error() slightly friendlier: Include the
54 enum if possible, and print it to stderr instead of stdout.
57 Movit 1.3.2, February 23rd, 2016
59 - Fix an issue with initialization in certain locales. Patch from
60 Jean-Baptiste Mardelle.
63 Movit 1.3.1, February 15th, 2016
65 - Fix an issue where certain effect chains (particularly involving
66 out-of-tree effects that return only a constant color) could cause
67 texture coordinates not to be set properly. (The new code probably
68 also has slightly lower OpenGL driver overhead.) Reported by
72 Movit 1.3.0, January 31st, 2016
74 - Movit now requires GLSL 1.30 (so a driver from 2008 or later);
75 before, it claimed to support 1.10, but actually used 1.30 features.
76 Note that some OpenGL drivers, in particular on OS X, only supports
77 GLSL 1.30 (actually, 1.50) if you have an OpenGL core context.
79 - Add a deinterlacer, based on YADIF.
81 - Allow parallel output to RGBA and Y'CbCr textures at the same time.
83 - Make FlatInput and YCbCrInput support taking in external OpenGL
84 textures. Also allow them to change width/height after instantiation.
86 - Various compatibility and performance fixes. In particular,
87 fp16 conversion on older (non-Haswell) CPUs is much faster, due to
88 new conversion code by Fabian Giesen.
91 Movit 1.2.0, September 24th, 2015
93 - Movit now ships a version.h with a #define MOVIT_VERSION that
94 increases on every API change, including in git. (The number
95 itself carries no semantic meaning beyond this.) Movit 1.2.0
96 is not API or ABI compatible with 1.1.x, so there has been a
97 soname bump to libmovit.so.3.
99 - More flexible Y'CbCr input; in particular, a special input
100 for 4:2:2 interleaved data (UYVY), and support for semi-planar
101 input (Cb and Cr in the same texture, like in NV12). Note that
102 you should now also set the new “num_levels” member of
103 YCbYCbCrFormat to 256 (signifying 8-bit input), although
104 it is not used yet; this is a stepping stone towards supporting
105 10- and 12-bit Y'CbCr.
107 - Basic support for Y'CbCr _output_. Currently only 8-bit,
108 and only 4:4:4 (ie., you'll need to subsample yourself
109 afterwards). It is possible to split the output into multiple
110 textures, though, if you want a luma/chroma split (NV12-like)
113 - Support top-left origin for output. This is only really useful
114 if you are rendering directly into some memory area with top-left
115 origin; most users will get the expected behavior by using
116 bottom-left as before.
118 - Rework uniform handling for less OpenGL overhead. Note that
119 this means your effects now need to register uniforms in the
120 C++ code instead of declaring them in the .frag file.
122 - Make the PaddingEffect border subpixel-aware, and also support
123 an arbitrary (subpixel) border offset. This means that you can
124 compose PaddingEffect with an integral left/top offset
125 (use the new IntegralPaddingEffect for potentially more speed)
126 and ResampleEffect with a subpixel left/top offset (and zoom to
127 compensate) to get Lanczos interpolation for the pixels and a
128 simple bilinear interpolation for the border itself.
130 - Fix a bug that could cause very bad filter weights in
133 - Various performance improvements, in particular with regards to
134 CPU usage in ResampleEffect.
137 Movit 1.1.3, March 29th, 2015
139 - Fix accuracy issues in ResampleEffect, particularly when
140 zooming. Note that this might cost a few percent performance.
141 Reported by Christophe Thommeret.
143 - Make the number of BlurEffect taps configurable.
144 Based on patch by Christophe Thommeret.
146 - Multiple other bugfixes, some courtesy Dan Dennedy and
147 Christophe Thommeret. In particular, thread-safety fixes
148 related to locales and chain finalization.
151 Movit 1.1.2, August 12th, 2014
153 - Performance bugfix: Fix texture freelist behavior so that it's
154 LRU instead of the exact opposite. Patch by Christophe Thommeret.
156 - Performance bugfix: Correct the number of blur taps read in the
157 blur filter (it was reading about twice as many as it should).
158 Also found by Christophe Thommeret.
161 Movit 1.1.1, April 12th, 2014
163 - Fix an issue that could cause assertion failure in ResourcePool
164 when FBOs are reused, especially with NVIDIA's drivers.
167 Movit 1.1, April 10th, 2014
169 - NOTE: Movit now uses libepoxy rather than GLEW. This is needed because
170 of core context and GLES3 support (see below). Movit 1.1 should be
171 generally API-compatible with 1.0.3 (with the exception of the next
172 point), but is not ABI-compatible, so there has been a soname bump
175 - NOTE: The rules for using an EffectChain or ResourcePool in multiple
176 OpenGL contexts or threads have changed somewhat; see the comments in
179 - Movit will now work in OpenGL core contexts, and on GLES 3.0 or newer.
180 (GLES2 devices are not supported.) This also holds for all unit tests,
181 provided that you use SDL2 and not SDL1 (because SDL2 is needed to set
182 up such contexts). Note that the included demo application still requires
183 a classic OpenGL context.
185 - ResampleEffect can now do sub-pixel translate and/or zoom.
187 - LumaMixEffect now has an “invert” flag, as a convenience to e.g. change
188 a left-to-right wipe into a right-to-left one.
190 - Significant reduction in driver overhead, especially on NVIDIA drivers.
192 - Various smaller bugfixes and performance improvements.
195 Movit 1.0.3, March 16th, 2014
197 - Yet more build system tweaks mainly related to distribution packaging.
200 Movit 1.0.2, March 16th, 2014
202 - Make a few tweaks to “make install”, to make distributions' lives easier.
205 Movit 1.0.1, March 16th, 2014
207 - Fix so that shared libraries are built.
210 Movit 1.0, March 16th, 2014