Issue was caused by delayed or missing image user frame
number update, which lead to image loading failure in
cases node is updating from image signal callback.
Solved in a way that file from image datablock is used
for sockets detection instead of loading image for
current frame.
PyNodes opens up the node system in Blender to scripters and adds a number of UI-level improvements.
=== Dynamic node type registration ===
Node types can now be added at runtime, using the RNA registration mechanism from python. This enables addons such as render engines to create a complete user interface with nodes.
Examples of how such nodes can be defined can be found in my personal wiki docs atm [1] and as a script template in release/scripts/templates_py/custom_nodes.py [2].
=== Node group improvements ===
Each node editor now has a tree history of edited node groups, which allows opening and editing nested node groups. The node editor also supports pinning now, so that different spaces can be used to edit different node groups simultaneously. For more ramblings and rationale see (really old) blog post on code.blender.org [3].
The interface of node groups has been overhauled. Sockets of a node group are no longer displayed in columns on either side, but instead special input/output nodes are used to mirror group sockets inside a node tree. This solves the problem of long node lines in groups and allows more adaptable node layout. Internal sockets can be exposed from a group by either connecting to the extension sockets in input/output nodes (shown as empty circle) or by adding sockets from the node property bar in the "Interface" panel. Further details such as the socket name can also be changed there.
[1] http://wiki.blender.org/index.php/User:Phonybone/Python_Nodes
[2] http://projects.blender.org/scm/viewvc.php/trunk/blender/release/scripts/templates_py/custom_nodes.py?view=markup&root=bf-blender
[3] http://code.blender.org/index.php/2012/01/improving-node-group-interface-editing/
Before this change only old flag "Premultiply" was used to
detect alpha mode, which is not enough actually.
Now the logic here is:
- If "Premultiply" was enabled it is likely float image with
straight alpha, which shall be premultiplied before usage.
In this case image/sequence Alpha Mode is set to Straight.
- Otherwise use default alpha mode for image format based on
an extension. This could fail in some cases like TIFF, but
this wasn't handled fully correct in older blender anyway.
Initial discovered issue was that EXR images saved in older
Blender versions were set to Straight alpha mode, which is
obviously a straight way to lots of headache.
The use alpha option moved from the texture datablock to the image, and now it
will duplicate the image datablock in case you have one texture using alpha and
the other not.
Issue was caused by couple of circumstances:
- Normal Map node requires tesselated faces to compute tangent space
- All temporary meshes needed for Cycles export were adding to G.main
- Undo pushes would temporary set meshes tessfaces to NULL
- Moving node will cause undo push and tree re-evaluate fr preview
All this leads to threading conflict between preview render and undo
system.
Solved it in way that all temporary meshes are adding to that exact
Main which was passed to Cycles via BlendData. This required couple
of mechanic changes like adding extra parameter to *_add() functions
and adding some *_ex() functions to make it possible RNA adds objects
to Main passed to new() RNA function.
This was tricky to pass Main to RNA function and IMO that's not so
nice to pass main to function, so ended up with such decision:
- Object.to_mesh() will add temp mesh to G.main
- Added Main.meshes.new_from_object() which does the same as to_mesh,
but adds temporary mesh to specified Main.
So now all temporary meshes needed for preview render would be added
to preview_main which does not conflict with undo pushes.
Viewport render shall not be an issue because object sync happens from
main thread in this case.
It could be some issues with final render, but that's not so much
likely to happen, so shall be fine.
Thanks to Brecht for review!
Issue was caused by FILE multilayer ImBuf sharing buffers with render
result, and SEQUENCE multilayer ImBufs duplicating buffers. Which is
nice by it's own. But, changing image source wouldn't remove any loaded
image buffers, meaning if you've got loaded FILE multilayers they'll
likely became invalid.
That behavior of handling multilayers on changing source was done as
a fix for #24976, which is now actually not needed (removing check
for multilayer doesn't change behavior at all).
Just to be sure added check to RNA, so signal wouldn't be fired if
source wasn't actually changed form a menu.
It was caused by image threading safe commit and it was noticeable
only on really multi-core CPU (like dual-socket Xeon stations), was
not visible on core i7 machine.
The reason of slowdown was spinlock around image buffer referencing,
which lead to lots of cores waiting for single core and using image
buffer after it was referenced was not so much longer than doing
reference itself.
The most clear solution here seemed to be introducing Image Pool
which will contain list of loaded and referenced image buffers, so
all threads could skip lock if the pool is used for reading only.
Lock only needed in cases when buffer for requested image user is
missing in the pool. This lock will happen only once per image so
overall amount of locks is much less that it was before.
To operate with pool:
- BKE_image_pool_new() creates new pool
- BKE_image_pool_free() destroys pool and dereferences all image
buffers which were loaded to it
- BKE_image_pool_acquire_ibuf() returns image buffer for given
image and user. Pool could be NULL and in this case fallback to
BKE_image_acquire_ibuf will happen.
This helps to avoid lots to if(poll) checks in image sampling
code.
- BKE_image_pool_release_ibuf releases image buffer. In fact, it
will only do something if pool is NULL, in all other case it'll
equal to DoNothing operation.
This adds support of movie textures for Cycles rendering.
Uses the same builtin images routines as packed/generated images,
but with some extra non-rna hookups from blender_session side.
Basically, it's not so clear how to give access to video frames
via C++ RNA -- it'll require exposing ImBuf to API, doing some
threading locks and so. Ended up adding two more functions which
are actually bad level call, but don't consider it's so much bad
-- we have few bad calls already, which are actually related.
Changed a bit how builtin images names are passing to image
manager. Now it's not just an ID datablock name, but also a frame
number concatenated via '@' character, which makes itpossible to
easily know frame number to be used for movie images, without
adding extra descriptors to image manager.
Decoding of builtin name is a bit slower now, but it should be
still nothing in comparison with rendering complexity.
Also exposed image user's frame_current to python API, which
is needed to get absolute frame number of movie from node's
image user.
P.S. Generated/packed images are also using bad level call but
only does it to make things more clear here. Either all images
are using C++ RNA here or no images does. That's the most clear
for now.
This assumptions are now made:
- Internally float buffers are always linear alpha-premul colors
- Readers should worry about delivering float buffers with that
assumptions.
- There's an input image setting to say whether it's stored with
straight/premul alpha on the disk.
- Byte buffers are now assumed have straight alpha, readers should
deliver straight alpha.
Some implementation details:
- Removed scene's color unpremultiply setting, which was very
much confusing and was wrong for default settings.
Now all renderers assumes to deliver premultiplied alpha.
- IMB_buffer_byte_from_float will now linearize alpha when
converting from buffer.
- Sequencer's effects were changed to assume bytes have got
straight alpha. Most of effects will work with bytes still,
however for glow it was more tricky to avoid data loss, so
there's a commented out glow implementation which converts
byte buffer to floats first, operates on floats and returns
bytes back. It's slower and not sure if it should actually
be used -- who're using glow on alpha anyway?
- Sequencer modifiers should also be working nice with straight
bytes now.
- GLSL preview will predivide float textures to make nice shading,
shading with byte textures worked nice (GLSL was assuming straight
alpha).
- Blender Internal will set alpha=1 to the whole sky. The same
happens in Cycles and there's no way to avoid this -- sky is
neither straight nor premul and doesn't fit color pipeline well.
- Straight alpha mode for render result was also eliminated.
- Conversion to correct alpha need to be done before linearizing
float buffer.
- TIFF will now load and save files with proper alpha mode setting
in file meta data header.
- Remove Use Alpha from texture mapping and replaced with image
datablock setting.
Behaves much more predictable and clear from code point of view
and solves possible regressions when non-premultiplied images were
used as textures with ignoring alpha channel.
This commit adds a support of saving 16bit PNG files.
Alpha for such files would be premultiplied, would be corrected
with an upcoming alpha premul cleanup (it's not the only format
which will output 16bit image with premul alpha).
This codec is absolutely needed to generate DCP using OpenDCP,
before that external application to convert JP2 to J2K was used
which slowed down export a lot.
New codec is exposed to image format settings panel and called
Codec. Default one is JP2 which creates files with .jp2 extension,
new one is called J2K which creates with .j2c extension.
Other changes:
- Fixed avi jpeg warning which was treating as error here.
- Made it so extension is detecting from ImageFormatData instead
of image file type, which makes it possible to have different
extension for the same file type depending on it's settings.
IRIS format should still be changed (depending on number of
channels it'll be .bw, .rgb or .rgba extension)
- Default image format settings would be set from image buffer
when re-saving it. Makes it possible to easily open .j2c file
and save it using J2K codec (without this change it'll save as
.jp2 using JP2 codec)
now blenlib/BLI doesn't depend on any blenkern/BKE functions,
there are still some bad level includes but these are only to access G.background and the blender version define.
This commit makes BKE_image_acquire_ibuf referencing result, which means once
some area requested for image buffer, it'll be guaranteed this buffer wouldn't
be freed by image signal.
To de-reference buffer BKE_image_release_ibuf should now always be used.
To make referencing working correct we can not rely on result of
image_get_ibuf_threadsafe called outside from thread lock. This is so because
we need to guarantee getting image buffer from list of loaded buffers and it's
referencing happens atomic. Without lock here it is possible that between call
of image_get_ibuf_threadsafe and referencing the buffer IMA_SIGNAL_FREE would
be called. Image signal handling too is blocking now to prevent such a
situation.
Threads are locking by spinlock, which are faster than mutexes. There were some
slowdown reports in the past about render slowdown when using OSX on Xeon CPU.
It shouldn't happen with spin locks, but more tests on different hardware would
be really welcome. So far can not see speed regressions on own computers.
This commit also removes BKE_image_get_ibuf, because it was not so intuitive
when get_ibuf and acquire_ibuf should be used.
Thanks to Ton and Brecht for discussion/review :)
Patch by Julien Enche, thanks!
From the patch comment:
It allows Blender to load:
- 1, 8, 10, 12 and 16 bits files. For 10 and 12 bits files, packed or
filled type A/B are supported.
- RGB, Log, Luma and YCbCr colorspaces.
- Big and little endian storage.
- Multi-elements (planar) storage.
It allows Blender to save :
- 8, 10, 12 and 16 bits file. For 10 and 12 bits files, the most used
type A padding is used.
- RGB and Log colorspaces (Cineon can only be saved in Log colorspace).
For Log colorspace, the common default values are used for gamma,
reference black and reference white (respectively 1.7, 95 and 685 for
10 bits files).
- Saved DPX/Cineon files now match the viewer.
Some files won't load (mostly because I haven't seen any of them):
- Compressed files
- 32 and 64 bits files
- Image orientation information are not taken in account. Here too,
I haven't seen any file that was not top-bottom/left-right oriented.
Just makes progressive refine :)
This means the whole image would be refined gradually using as much
threads as it's set in performance settings. Having enough tiles is
required to have this option working as it's expected.
Technically it's implemented by repeatedly computing next sample for
all the tiles before switching to next sample.
This works around 7-12% slower than regular tile-based rendering, so
use this option only if you really need it.
This commit also fixes progressive update of image when Save Buffers
option is enabled.
And one more thing this commit fixes is handling display buffer with
Save Buffers option enabled. If this option is enabled image buffer
wouldn't have neither byte nor float buffer until image is fully
rendered which could backfire in missing image while rendering in
cases color management cache became full.
This issue solved by allocating byte buffer for image buffer from
tile update callback.
Patch was reviewed by Brecht. He also made some minor edits to
original version to patch. Thanks, man!
The issue was caused by SEQ_BEGIN macro modifying sequence's depth
which ruined transformation routines. Used own DFS instead which
doesn't modify sequences.
Also corrected some typos in api and comments.
Color management would be applied on both of float and byte buffers on image
save in cases if file format doesn't require linear float buffer and if image
is saving as render result.
This solves both initial report issue and TODO marked in previous fix.
Also de-duplicated image buffer color managing code and gave some more
meaningful names for few functions. Also wrote documentation around this
function, so current assumptions about spaces should be clear enough.
Made regression tests by saving EXR/PNG images to all supported format and
rendering OpenGL/Normal animation, in all cases seems everything is fine,
but more tests for sure would be welcome.
this is a regression with color management, TIFF's were always being written as 8bit
however the float buffer is assumed to be linear when converting from float to 16bit pixels per channel, so support for color management might need to be added here.
We need to assign color space for generated images when it's empty.
Do this in add_ibuf_size function in cases image's color space is
empty.
So now color management is always called with correct color space.
P.S. At least it should. If there're more similar issues please
let me know :)
- Color space of byte buffer for generated images haven't been
updated properly on change
- Byte buffers weren't handling data color spaces on display transform
properly
Issue was caused by completely different way how multi-layer EXRs are loading,
they're bypassing general image buffer loading functions.
Solved by running color space transformation on render result construction
from multi-layer EXR image.
Also fixed issue with wrong display buffer computing for buffers with less
than 4 channels. Issues were:
- Display buffer is always expected to be RGBA
- OpenColorIO can not apply color space transformations on non-{RGB, RGBA}
pixels.
In fact, color management settings shouldn't be editing from this operator since
screenshot is using OpenGL which is in sRGB space, but fixed issues with Image
Format initialization and added NULL pointer checks.
Replace old color pipeline which was supporting linear/sRGB color spaces
only with OpenColorIO-based pipeline.
This introduces two configurable color spaces:
- Input color space for images and movie clips. This space is used to convert
images/movies from color space in which file is saved to Blender's linear
space (for float images, byte images are not internally converted, only input
space is stored for such images and used later).
This setting could be found in image/clip data block settings.
- Display color space which defines space in which particular display is working.
This settings could be found in scene's Color Management panel.
When render result is being displayed on the screen, apart from converting image
to display space, some additional conversions could happen.
This conversions are:
- View, which defines tone curve applying before display transformation.
These are different ways to view the image on the same display device.
For example it could be used to emulate film view on sRGB display.
- Exposure affects on image exposure before tone map is applied.
- Gamma is post-display gamma correction, could be used to match particular
display gamma.
- RGB curves are user-defined curves which are applying before display
transformation, could be used for different purposes.
All this settings by default are only applying on render result and does not
affect on other images. If some particular image needs to be affected by this
transformation, "View as Render" setting of image data block should be set to
truth. Movie clips are always affected by all display transformations.
This commit also introduces configurable color space in which sequencer is
working. This setting could be found in scene's Color Management panel and
it should be used if such stuff as grading needs to be done in color space
different from sRGB (i.e. when Film view on sRGB display is use, using VD16
space as sequencer's internal space would make grading working in space
which is close to the space using for display).
Some technical notes:
- Image buffer's float buffer is now always in linear space, even if it was
created from 16bit byte images.
- Space of byte buffer is stored in image buffer's rect_colorspace property.
- Profile of image buffer was removed since it's not longer meaningful.
- OpenGL and GLSL is supposed to always work in sRGB space. It is possible
to support other spaces, but it's quite large project which isn't so
much important.
- Legacy Color Management option disabled is emulated by using None display.
It could have some regressions, but there's no clear way to avoid them.
- If OpenColorIO is disabled on build time, it should make blender behaving
in the same way as previous release with color management enabled.
More details could be found at this page (more details would be added soon):
http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.64/Color_Management
--
Thanks to Xavier Thomas, Lukas Toene for initial work on OpenColorIO
integration and to Brecht van Lommel for some further development and code/
usecase review!
