| Age | Commit message (Collapse) | Author |
|---|
|
On my setup, since commit fb8ad13d ("libcamera: software_isp: Move exposure+gain
to an algorithm module"), at start camera output stays very dark for dozen
of seconds, and then later slowly gets to normal. This is because existing
sensor exposure+gain settings are not used at start. We save initial
values in frameContext but in the agc algorithm we use IPA context.
Fix the problem by using in frameContext sensor values, since we already
use those in blc algorithm and change exposure type to int32_t to
unnecessary castings.
Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
Reviewed-by: Milan Zamazal <mzamazal@redhat.com>
Tested-by: Robert Mader <robert.mader@collabora.com>
Tested-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
The names used by the IPA interface and the names used for buffer
completions handlers in libcamera clash in the use of the term "buffer".
For example video device buffer completion handler is called
"bufferReady" and the IPA event to ask the IPA to compute parameters are
called "fillParamsBuffers". This makes it hard to recognize which
function handles video device completion signals and which ones handle
the IPA interface events.
Rationalize the naming scheme in the IPA interface function and events
and the signal handlers in the pipelines, according to the
following table. Remove the name "buffer" from the IPA interface events
and events handler and reserve it for the buffer completion handlers.
Rename the IPA interface events and function to use the 'params' and
'stats' names as well.
IPA Interface:
- fillParamsBuffer -> computeParams [FUNCTION]
- processStatsBuffer -> processStats [FUNCTION]
- paramFilled -> paramsComputed [EVENT]
Pipeline handler:
- bufferReady -> videoBufferReady [BUFFER HANDLER]
- paramReady -> paramBufferReady [BUFFER HANDLER]
- statReady -> statBufferReady [BUFFER HANDLER]
- paramFilled -> paramsComputed [IPA EVENT HANDLER]
Cosmetic change only, no functional changes intended.
Signed-off-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Reviewed-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
This patch allows obtaining a black level from a tuning file in addition
to the camera sensor helper. If both of them define a black level, the
one from the tuning file takes precedence.
The use cases are:
- A user wants to use a different black level, for whatever reason.
- There is a sensor without known gains but with a known black level.
Because a camera sensor helper cannot be defined without specifying
gains, the only way to specify the black level is using the tuning
file. Software ISP uses its fallback gain handling in such a case.
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
The black level in software ISP is unconditionally guessed from the
obtained frames. CameraSensorHelper optionally provides the black level
from camera specifications now. Let's use the value if available.
If the black level is not available from the given CameraSensorHelper
instance, it's still determined on the fly.
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Tested-by: Robert Mader <robert.mader@collabora.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
Like the hardware pipelines do. Not clearing frameContexts otherwise can
trigger asserts like "Frame context for ... has been overwritten by ..."
when switching between cameras using the swISP, e.g. on phones.
Clearing the configuration and active state will become more important
with upcoming changes such as getting the black level from the camera
helper.
Fixes: 04d171e6b299 ("libcamera: software_isp: Call Algorithm::queueRequest")
Signed-off-by: Robert Mader <robert.mader@collabora.com>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Milan Zamazal <mzamazal@redhat.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
This is the last step to fully convert software ISP to Algorithm-based
processing.
The newly introduced frameContext.sensor parameters are set, and the
updated code moved, before calling Algorithm::process() to have the
values up-to-date in stats processing.
Resolves software ISP TODO #10.
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
Use the standard libcamera mechanism to report the "current" controls
rather than delaying updates by counting from the last update.
A problem is that with software ISP we cannot be sure about the sensor
delay. The original implementation simply skips exposure updates for 2
frames, which should be enough in most cases. After this change, we
assume the delay being exactly 2 frames, which may or may not be correct
and may work with outdated values if the delay is shorter.
According to Kieran, the wrong parts are also wrong on the
IPU3/RKISP1/Mali pipelines and only RPi have this correct. We need to
fix this, by correctly specifying the gains in the libipa camera sensor
helpers. The sooner the better because this change could introduce a
risk of increasing oscillations.
This patch also prepares moving exposure+gain to an algorithm module
where the original delay mechanism would be a (possibly unnecessary)
complication.
Resolves software ISP TODO #11 + #12.
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
After black level handling has been moved to an algorithm module, white
balance and the construction of color tables can be moved to algorithm
modules too.
This time, the moved code is split between stats processing and
parameter construction methods. It is also split to two algorithm
modules:
- White balance computation.
- Gamma table computation and color lookup tables construction. While
this applies the color gains computed by the white balance algorithm,
it is not directly related to white balance. And we may want to
modify the color lookup tables in future according to other parameters
than just gamma and white balance gains.
Gamma table computation and color lookup tables construction could be
split to separate algorithms too. But there is no big need for that now
so they are kept together for simplicity.
This is the only part of the software ISP algorithms that sets the
parameters so emitting setIspParams can be moved to prepare() method.
A more natural representation of the gains (and black level) would be
floating point numbers. This is not done here in order to minimize
changes in code movements. It will be addressed in a followup patch.
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Acked-by: Umang Jain <umang.jain@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
The black level determination, already present as a separate class, can
be moved to the prepared Algorithm processing structure. It is the
first of the current software ISP algorithms that is called in the stats
processing sequence, which means it is also the first one that should be
moved to the new structure. Stats processing starts with calling
Algorithm-based processing so the call order of the algorithms is
retained.
Movement of this algorithm is relatively straightforward because all it
does is processing stats.
The comment about getting black level from the tuning files is dropped.
The black level will be taken from CameraSensorHelper if available and
that will be implemented in one of the followup patches.
Black level is now recomputed on each stats processing. In a future
patch, after DelayedControls are used, this will be changed to recompute
the black level only after exposure/gain changes.
The black level depends on the sensor used, the computed value can be
reused in a followup capture sessions with the same sensor. Thus it is
sufficient to (re)set the initial value in BlackLevel::init.
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
This patch adds Algorithm::process call for the defined algorithms.
This is preparation only since there are currently no Algorithm based
algorithms defined.
As software ISP currently doesn't produce any metadata, a dummy and
unused metadata instance is created to satisfy Algorithm::process API.
This should be changed in future.
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
This patch adds Algorithm::prepare call for the defined algorithms.
This is preparation only since there are currently no Algorithm based
algorithms defined.
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
This patch adds Algorithm::queueRequest call for the defined algorithms.
As there are currently no control knobs in software ISP nor the
corresponding queueRequest call chain, the patch also introduces the
queueRequest methods and calls from the pipeline to the IPA.
This is preparation only since there are currently no Algorithm based
algorithms defined and no current software ISP algorithms support
control knobs.
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
This patch adds Algorithm::configure call for the defined algorithms.
This is preparation only since there are currently no Algorithm based
algorithms defined.
A part of this change is passing IPAConfigInfo instead of ControlInfoMap
to configure() calls as this is what Algorithm::configure expects.
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
We are ready to introduce algorithms now. First, let's create
algorithms. The algorithms are not called yet, calls to them will be
added in followup patches.
The maximum number of contexts is set to the same value as in hardware
pipelines.
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
This patch adds frame and bufferId arguments to stats related calls.
Although the parameters are currently unused, because frame ids are not
tracked and used and the stats buffer is passed around directly rather
than being referred by its id, they bring the internal APIs closer to
their counterparts in hardware pipelines.
It serves as a preparation for followup patches that will introduce:
- Frame number tracking in order to switch to DelayedControls
(software ISP TODO #11 + #12).
- A ring buffer for stats in order to improve passing the stats
(software ISP TODO #2).
Frame and buffer ids are unrelated for the given purposes but since they
are passed together at the same places, the change is implemented as a
single patch rather than two, basically the same, patches.
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
The Module class is a base class for all IPA modules.
In addition, implement logPrefix() of the module for the softIPA.
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
The constant is used in a single place internally and doesn't belong to
DebayerParams anymore. Let's use 256 directly.
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Andrei Konovalov <andrey.konovalov.ynk@gmail.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Constructing the color mapping tables is related to stats rather than
debayering, where they are applied. Let's move the corresponding code
to stats processing.
The same applies to the auxiliary gamma table. As the gamma value is
currently fixed and used in a single place, with the temporary exception
mentioned below, there is no need to share it anywhere anymore.
It's necessary to initialize SoftwareIsp::debayerParams_ to default
values. These initial values are used for the first two frames, before
they are changed based on determined stats. To avoid sharing the gamma
value constant in artificial ways, we use 0.5 directly in the
initialization. This all is not a particularly elegant thing to do,
such a code belongs conceptually to the similar code in stats
processing, but doing better is left for larger refactoring.
This is a preliminary step towards building this functionality on top of
libipa/algorithm.h, which should follow.
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Andrei Konovalov <andrey.konovalov.ynk@gmail.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
The white balance computation didn't consider black level. This is
wrong because then the computed ratios are off when they are computed
from the whole brightness range rather than the sensor range.
This patch adjusts white balance computation for the black level. There
is no need to change white balance application in debayering as this is
already applied after black level correction.
Exposure computation already subtracts black level, no changes are
needed there.
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Andrei Konovalov <andrey.konovalov.ynk@gmail.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
The PipelineHandlerFactoryBase class has a name that is propagated to
the PipelineHandler instance it creates.
In present implementation, this name comes from the
REGISTER_PIPELINE_HANDLER registration macro. It corresponds to the
stringified name of the PipelineHandler derived class. Therefore,
PipelineHandler factories and instances names can be quite long such as
"PipelineHandlerRkISP1".
A libcamera user may have to explicitly refer to a PipelineHandler name
for configuration purpose: one usage of the name can be to define a
pipeline handlers match list and their priorities. It is desired, for
user convenience, to use a short name to designate a pipeline handler.
Reusing the short pipeline names already defined in the meson option
files is an existing and consistent way of naming pipelines.
This change adds an explicit name parameter to the
REGISTER_PIPELINE_HANDLER registration macro. That parameter is used to
define the name of a pipeline handler factory, instead of the current
pipeline handler class name.
Each pipeline registration is updated accordingly. The short name
assigned corresponds to the pipeline directory name in the source tree.
It is consistent with pipelines names used in meson.
Changing the pipeline name has an impact on the IPA modules: each module
defines a IPAModuleInfo structure. This structure has a pipelineName
member defining the pipeline handler name it shall match with.
Therefore, each internal IPA module definition has to be changed to have
its IPAModuleInfo pipelineName name updated with the short pipeline
handler name.
In addition to this pipelineName member, the IPAModuleInfo structure
also has a name member, associated to the IPA module name. Having
renamed the pipelines to a short name, the pipeline name and the IPA
module names of the IPAModuleInfo structure are the same: for in-tree
IPA, they correspond to the respective pipeline and IPA subdirectories
in the source tree. However the IPA name could be different, for
instance with a close source IPA implementation built out-of-tree. Thus,
it makes sense to keep the IPA name in that structure, as the 2
definitions may not always be redundant.
Signed-off-by: Julien Vuillaumier <julien.vuillaumier@nxp.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
[Kieran: Adjust for clang-format style fix, reformat commitmsg]
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
Source files in libcamera start by a comment block header, which
includes the file name and a one-line description of the file contents.
While the latter is useful to get a quick overview of the file contents
at a glance, the former is mostly a source of inconvenience. The name in
the comments can easily get out of sync with the file name when files
are renamed, and copy & paste during development have often lead to
incorrect names being used to start with.
Readers of the source code are expected to know which file they're
looking it. Drop the file name from the header comment block.
The change was generated with the following script:
----------------------------------------
dirs="include/libcamera src test utils"
declare -rA patterns=(
['c']=' \* '
['cpp']=' \* '
['h']=' \* '
['py']='# '
['sh']='# '
)
for ext in ${!patterns[@]} ; do
files=$(for dir in $dirs ; do find $dir -name "*.${ext}" ; done)
pattern=${patterns[${ext}]}
for file in $files ; do
name=$(basename ${file})
sed -i "s/^\(${pattern}\)${name} - /\1/" "$file"
done
done
----------------------------------------
This misses several files that are out of sync with the comment block
header. Those will be addressed separately and manually.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
|
|
Black may not be represented as 0 pixel value for given hardware, it may
be higher. If this is not compensated then various problems may occur
such as low contrast or suboptimal exposure.
The black pixel value can be either retrieved from a tuning file for the
given hardware, or automatically on the fly. The former is the right
and correct method, while the latter can be used when a tuning file is
not available for the given hardware. Since there is currently no
support for tuning files in software ISP, the automatic, hardware
independent way, is always used. Support for tuning files should be
added in future but it will require more work than this patch.
The patch looks at the image histogram and assumes that black starts
when pixel values start occurring on the left. A certain amount of the
darkest pixels is ignored; it doesn't matter whether they represent
various kinds of noise or are real, they are better to omit in any case
to make the image looking better. It also doesn't matter whether the
darkest pixels occur around the supposed black level or are spread
between 0 and the black level, the difference is not important.
An arbitrary threshold of 2% darkest pixels is applied; there is no
magic about that value.
The patch assumes that the black values for different colors are the
same and doesn't attempt any other non-primitive enhancements. It
cannot completely replace tuning files and simplicity, while providing
visible benefit, is its goal. Anything more sophisticated is left for
future patches.
A possible cheap enhancement, if needed, could be setting exposure +
gain to minimum values temporarily, before setting the black level. In
theory, the black level should be fixed but it may not be reached in all
images. For this reason, the patch updates black level only if the
observed value is lower than the current one; it should be never
increased.
The purpose of the patch is to compensate for hardware properties.
General image contrast enhancements are out of scope of this patch.
Stats are still gathered as an uncorrected histogram, to avoid any
confusion and to represent the raw image data. Exposure must be
determined after the black level correction -- it has no influence on
the sub-black area and must be correct after applying the black level
correction. The granularity of the histogram is increased from 16 to 64
to provide a better precision (there is no theory behind either of those
numbers).
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
Define the Soft IPA main and event interfaces, add the Soft IPA
implementation.
The current src/ipa/meson.build assumes the IPA name to match the
pipeline name. For this reason "-Dipas=simple" is used for the
Soft IPA module.
Auto exposure/gain and AWB implementation by Dennis, Toon and Martti.
Auto exposure/gain targets a Mean Sample Value of 2.5 following
the MSV calculation algorithm from:
https://www.araa.asn.au/acra/acra2007/papers/paper84final.pdf
Use CameraSensorHelper to convert the analogue gain code read from the
camera sensor into real analogue gain value. In the future this makes
it possible to use faster AE/AGC algorithm. Right now the CameraSensorHelper
lets us use the full range of analogue gain values.
If there is no CameraSensorHelper for the camera sensor in use, a
warning log message is printed.
Tested-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org> # sc8280xp Lenovo x13s
Tested-by: Pavel Machek <pavel@ucw.cz>
Reviewed-by: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Andrey Konovalov <andrey.konovalov@linaro.org>
Co-developed-by: Dennis Bonke <admin@dennisbonke.com>
Signed-off-by: Dennis Bonke <admin@dennisbonke.com>
Co-developed-by: Marttico <g.martti@gmail.com>
Signed-off-by: Marttico <g.martti@gmail.com>
Co-developed-by: Toon Langendam <t.langendam@gmail.com>
Signed-off-by: Toon Langendam <t.langendam@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
