1 Movit 1.6.2, March 18th, 2018
6 Movit 1.6.1, January 31st, 2018
8 - Loosen up some restrictions on strong one-to-one-effects. In particular,
9 this fixes a bug with MixEffect and compute shaders that hit some
13 Movit 1.6.0, January 24th, 2018
15 - Support for effects that work as compute shaders. Compute shaders are
16 generally slower than fragment shaders for the same algorithm,
17 but allow some forms of communication between shader invocations
18 and have more flexible output, which can enable more efficient algorithms.
19 See effect.h for more details. Note that the fastest rendering API on
20 EffectChain is now to a texture if possible, not to an FBO. This will
21 only matter if the last effect is a compute shader.
23 - Movit now includes a compute shader implementation of DeinterlaceEffect,
24 which is automatically used instead of the fragment shader implementation
25 if your GPU and OpenGL driver supports it (in practice, this means on
26 all platforms except on macOS). The compute shader version is typically
27 20–80% faster than the fragment shader version, depending on your GPU
30 A compute shader implementation of ResampleEffect was written but
31 ultimately failed to be faster, and so is not included.
33 - Support for microbenchmarks of effects through the Google microbenchmarking
34 framework (optional). Currently, DeinterlaceEffect and ResampleEffect has
35 benchmarks; enable them by running the unit test with --benchmark (also try
38 - Effects can now explicitly request _not_ to have mipmaps, which means they
39 can do so without needing to request bounce and fiddling with the sampler
40 state. Note that this is an API change for effects.
42 - Movit now requires C++11, both to build and to #include the header files.
43 Support for SDL1 has been dropped; unit tests and the demo program now need
46 - Various smaller bugfixes and optimizations.
49 Movit 1.5.3, August 10th, 2017
54 Movit 1.5.2, July 5th, 2017
56 - The texture generation for ResampleEffect (when the size is changed)
59 - Work around an NVIDIA driver threading bug by aggressively caching VAOs.
62 Movit 1.5.1, May 29th, 2017
64 - Allow YCbCrInput to change input format after finalize.
66 - Some minor YCbCrInput bugfixes.
69 Movit 1.5.0, March 21st, 2017
71 - Support interleaved Y'CbCr input (4:4:4 in a single texture).
73 - Support 10-bit and 12-bit Y'CbCr, both for input and output. For planar,
74 these are supported packed in 16-bit ints; for interleaved, 10:10:10:2 is
75 supported. (Efficient conversion to and from v210, ie. 10-bit 4:2:2,
76 is possible using compute shaders, but Movit does not include support
77 for them at the current point.) Note that this now means the num_levels
78 flag in YCbCrFormat actually matters, although 0 will be interpreted
79 as 256 (8-bit) for the benefit of older applications.
81 - Limited support for having multiple Y'CbCr outputs from a chain.
83 - Allow changing the Y'CbCr output coefficients runtime, ie., after finalize.
85 - Fix an issue where the last pass would have been rendered with the sRGB
86 flag set, which confused Qt applications running in certain NVIDIA
90 Movit 1.4.0, November 5th, 2016
92 - Allow setting the intermediate format for chains, instead of hardcoding
93 it at 16-bit RGBA; advanced users can use this to e.g. ask for 8-bit
94 sRGB intermediates, reducing the amount of memory bandwidth needed
95 at the cost of reduced precision. Whether this is a good tradeoff or not
96 depends on the exact chain and your requirements.
98 - Fix an issue where a (cached) shader program could be used from multiple
99 threads at a time, causing the uniforms to contain unpredictable values.
101 - Make the error printed on check_error() slightly friendlier: Include the
102 enum if possible, and print it to stderr instead of stdout.
105 Movit 1.3.2, February 23rd, 2016
107 - Fix an issue with initialization in certain locales. Patch from
108 Jean-Baptiste Mardelle.
111 Movit 1.3.1, February 15th, 2016
113 - Fix an issue where certain effect chains (particularly involving
114 out-of-tree effects that return only a constant color) could cause
115 texture coordinates not to be set properly. (The new code probably
116 also has slightly lower OpenGL driver overhead.) Reported by
117 Christophe Thommeret.
120 Movit 1.3.0, January 31st, 2016
122 - Movit now requires GLSL 1.30 (so a driver from 2008 or later);
123 before, it claimed to support 1.10, but actually used 1.30 features.
124 Note that some OpenGL drivers, in particular on OS X, only supports
125 GLSL 1.30 (actually, 1.50) if you have an OpenGL core context.
127 - Add a deinterlacer, based on YADIF.
129 - Allow parallel output to RGBA and Y'CbCr textures at the same time.
131 - Make FlatInput and YCbCrInput support taking in external OpenGL
132 textures. Also allow them to change width/height after instantiation.
134 - Various compatibility and performance fixes. In particular,
135 fp16 conversion on older (non-Haswell) CPUs is much faster, due to
136 new conversion code by Fabian Giesen.
139 Movit 1.2.0, September 24th, 2015
141 - Movit now ships a version.h with a #define MOVIT_VERSION that
142 increases on every API change, including in git. (The number
143 itself carries no semantic meaning beyond this.) Movit 1.2.0
144 is not API or ABI compatible with 1.1.x, so there has been a
145 soname bump to libmovit.so.3.
147 - More flexible Y'CbCr input; in particular, a special input
148 for 4:2:2 interleaved data (UYVY), and support for semi-planar
149 input (Cb and Cr in the same texture, like in NV12). Note that
150 you should now also set the new “num_levels” member of
151 YCbYCbCrFormat to 256 (signifying 8-bit input), although
152 it is not used yet; this is a stepping stone towards supporting
153 10- and 12-bit Y'CbCr.
155 - Basic support for Y'CbCr _output_. Currently only 8-bit,
156 and only 4:4:4 (ie., you'll need to subsample yourself
157 afterwards). It is possible to split the output into multiple
158 textures, though, if you want a luma/chroma split (NV12-like)
161 - Support top-left origin for output. This is only really useful
162 if you are rendering directly into some memory area with top-left
163 origin; most users will get the expected behavior by using
164 bottom-left as before.
166 - Rework uniform handling for less OpenGL overhead. Note that
167 this means your effects now need to register uniforms in the
168 C++ code instead of declaring them in the .frag file.
170 - Make the PaddingEffect border subpixel-aware, and also support
171 an arbitrary (subpixel) border offset. This means that you can
172 compose PaddingEffect with an integral left/top offset
173 (use the new IntegralPaddingEffect for potentially more speed)
174 and ResampleEffect with a subpixel left/top offset (and zoom to
175 compensate) to get Lanczos interpolation for the pixels and a
176 simple bilinear interpolation for the border itself.
178 - Fix a bug that could cause very bad filter weights in
181 - Various performance improvements, in particular with regards to
182 CPU usage in ResampleEffect.
185 Movit 1.1.3, March 29th, 2015
187 - Fix accuracy issues in ResampleEffect, particularly when
188 zooming. Note that this might cost a few percent performance.
189 Reported by Christophe Thommeret.
191 - Make the number of BlurEffect taps configurable.
192 Based on patch by Christophe Thommeret.
194 - Multiple other bugfixes, some courtesy Dan Dennedy and
195 Christophe Thommeret. In particular, thread-safety fixes
196 related to locales and chain finalization.
199 Movit 1.1.2, August 12th, 2014
201 - Performance bugfix: Fix texture freelist behavior so that it's
202 LRU instead of the exact opposite. Patch by Christophe Thommeret.
204 - Performance bugfix: Correct the number of blur taps read in the
205 blur filter (it was reading about twice as many as it should).
206 Also found by Christophe Thommeret.
209 Movit 1.1.1, April 12th, 2014
211 - Fix an issue that could cause assertion failure in ResourcePool
212 when FBOs are reused, especially with NVIDIA's drivers.
215 Movit 1.1, April 10th, 2014
217 - NOTE: Movit now uses libepoxy rather than GLEW. This is needed because
218 of core context and GLES3 support (see below). Movit 1.1 should be
219 generally API-compatible with 1.0.3 (with the exception of the next
220 point), but is not ABI-compatible, so there has been a soname bump
223 - NOTE: The rules for using an EffectChain or ResourcePool in multiple
224 OpenGL contexts or threads have changed somewhat; see the comments in
227 - Movit will now work in OpenGL core contexts, and on GLES 3.0 or newer.
228 (GLES2 devices are not supported.) This also holds for all unit tests,
229 provided that you use SDL2 and not SDL1 (because SDL2 is needed to set
230 up such contexts). Note that the included demo application still requires
231 a classic OpenGL context.
233 - ResampleEffect can now do sub-pixel translate and/or zoom.
235 - LumaMixEffect now has an “invert” flag, as a convenience to e.g. change
236 a left-to-right wipe into a right-to-left one.
238 - Significant reduction in driver overhead, especially on NVIDIA drivers.
240 - Various smaller bugfixes and performance improvements.
243 Movit 1.0.3, March 16th, 2014
245 - Yet more build system tweaks mainly related to distribution packaging.
248 Movit 1.0.2, March 16th, 2014
250 - Make a few tweaks to “make install”, to make distributions' lives easier.
253 Movit 1.0.1, March 16th, 2014
255 - Fix so that shared libraries are built.
258 Movit 1.0, March 16th, 2014