the UI on OS X.
The problem is due to a large variable on the stack, and pthreads have a smaller
stack size than the main thread by default. On Linux the pthread stack size seems
to be 2MB, OS X 512KB and Windows 256KB - 512KB.
- reading bmp images was failing (needed to increase the size of the header to 64 bytes)
- the dnd image was being incorrectly checked (was always returning true even when none was used).
Added some special case for two-component channels name.
Maybe magic could be simplified to just use last char of
channel name as an id, but extra paranoid check never hurts.
There were an issues with data structures defined in headers
and being used by both C and C++ on systems with stdbool
unavailable.
This happened because bool in this case will be defined as
unsigned int, which is 4 bytes. But C++'s bool is only 1
byte and this lead to alignment issues.
Now bool is always 1 byte, also made sure there's no situation
like bool foo = BitField & BitFlag, which could give overflow
issues. Use (BitField & BitFlag) != 0 instead.
Fixes#35553: Compositor broken (Backdrop & Preview)
2 bugs here
- missing NULL check in IMB_colormanagement_display_settings_from_ctx()
- deadlock in draw_image_paint_helpers()
Simple solution is to not draw paint helpers for render/viewer images.
consistent with projection painting.
Also did some refactoring of this code, moving the brush image creation code
out of brush.c and making it consistent with image updating code.
Animplayer treated .tif extensions as movie files, so that didn't work.
Added another hardcoded check for it, like for png tga exr jpg etc.
Why FFmpeg thinks .tif is a movie... that's for another day :)
Quicktime QTKit did same btw.
* 2D image painting support for masking to limit the max contribution of a stroke
to a pixel, to get it working compatible with projection painting. Not strictly
a bugfix, but the inconsistency here was annoying.
* Fix python errors in Texture Mask panel in image editor, was missing overlay
options.
* Clamp paint mask to 0..1 in case some texture exceeds it, this could give black
pixels due to integer overflow.
previous fix for 2D image painting were done, and also:
* For brushes that do masking (keeping the max contribution to a pixel limited),
the alpha from textures is now part of the mask. In many cases the logic
worked out the same and where it didn't it used to cause artifacts.
* Color interpolation for smear and soften tools now happens in premultipied
space, to avoid bleeding of RGB colors from zero alpha areas.
* Fix a few cases where byte <=> float conversion was not doing the proper
straight <=> premul conversion.
* Replace some float division by constants with multiplications, is faster.
Note: float texture painting seems to have some issues updating the OpenGL
texture, but issue was already there before this commit.
old issue, the formulas here were never quite right, should all work ok now
with byte and float images.
Some differences:
* Colors with zero alpha from the background will never have an influence, so
you don't get alpha fringes when painting over such areas. This does give
hard edges when looking at the RGB channels alone, but there's no way to
avoid that and fringes at the same time, same behavior as other painting apps.
* Add/Subtract/Multiply/Lighten/Darken now leave the alpha channel unchanged
and work only the RGB channels, again same behavior as many other apps.
* Erase/Add alpha now compensates for premultiplied float images to keep the
straight RGB colors the same.
Next: fix projection painting.
The problem was that vertex colors only have 8 bits of precision, and integer
division always rounds down, so after some color blending iterations everything
gets darker. Instead use integer division that behaves like round() instead of
floor() for blending operations.
The latest ffmpeg versions include a workaround to deal with a certain
pecularity in Canon DSLR footage: instead of decoding pictures with the
proper resolution of 1920x1080 they decode it with 1920x1088 and add a
black bar at the bottom.
Needless to say, that this screws up things in a lot of areas within blender
(proxy indices, mask animations etc.)
Since all blender versions besides Linux x86 32bit seem still to include
older ffmpeg versions which still contain this bug, this patch adds
a workaround for older versions until we have all versions on all platforms
up to date.
See also: http://git.libav.org/?p=libav.git;a=commit;h=30f515091c323da59c0f1b533703dedca2f4b95d
Allocate float buffer outside of image buffer,
so work-in-progress color space conversion doesn't
interfere with other parts of blender.
Covers most of cases -- since image buffer wouldn't
have partially-update float buffer all the rest
areas would be happy.
However, if there're places which updates float
buffer from byte buffer, it's still possible
some WIP color space conversion is displayed on
the screen.
But what a heck someone will do such a crappy
conversion anyway!
- GLSL shader wasn't aware of alpha predivide option,
always assuming alpha is straight. Gave wrong results
when displaying transparent float buffers.
- GLSL display wasn't aware of float buffers with number
of channels different from 4, crashing when trying to
display image with different number of channels.
This required a bit larger changes, namely now it's
possible to pass format (GL_RGB, GL_RGBAm GL_LUMINANCE)
to glaDrawPixelsTex, This also implied adding format to
glaDrawPixelsAuto and modifying all places where this
functions are called.
Now GLSL will handle both 3 and 4 channels buffers,
single channel images are handled by CPU.
- Replaced hack for render result displaying with a bit
different hack.
Namely CPU conversion will happen only during render,
once render is done GLSL would be used for displaying
render result on a screen.
This is so because of the way renderer updates parts
of the image -- it happens without respect to active
render layer in image user. This is harmless because
only display buffer is modifying, but this is tricky
because we don't have original buffer opened during
rendering.
One more related fix here was about when rendering
multiple layers, wrong image would be displaying when
rendering is done. Added a signal to invalidate
display buffer once rendering is done (only happens
when using multiple layers). This solves issue with
wrong buffer stuck on the display when using regular
CPU display space transform and if GLSL is available
it'll make image displayed with a GLSL shader.
- As an additional change, byte buffers now also uses
GLSL display transform.
So now only dutehr and RGB curves are stoppers for
using GLSL for all kind of display transforms.
Uses GLSL for drawing image in Image Editor space.
This requires change in image_buffer_rect_update, so
original float buffer is being updated as well. This
is unlikely be something bad, but will keep an eye
on this change.
Also no byte buffer allocation happens there, this
is so because byte buffer used for display only
and in case of GLSL display such allocation and
partial update is just waste of time.
Also switched OpenGL render from using CPU color
space linearization to GLSL color space transform.
Makes OpenGL rendering pretty much faster (but
still slower than in 2.60).
internal changes:
- Added functions to setup GLSL shader for color
space conversion in colormanagement.c. Currently
conversion form a colorspace defined by a role to
linear space is implemented. Easy to extend to
other cases.
- Added helper functions to glutil.c which does
smarter image buffer draw (calling all needed OCIO
stuff, editors now could draw image buffer with a
single function call -- all the checks are done in
glutil.c).
- Also added helper function for buffer linearization
from a given role to glutil.c. Everyone now able to
linearize buffer with a single call.
This function will do nothing is GLSL routines fails
or not supported.
And one last this: this function uses offscreen
drawing, could potentially give issues on some
cards, also will keep an eye on this.
