The original code used to specify the probabilities in log space and
scaled for the RaspberryPi hardware with 192 AWB measurement points.
This is reasonable as the whole algorithm makes use of unitless numbers
to prefer some colour temperatures based on a lux level. These numbers
are then hand tuned with the specific device in mind.
This has two shortcomings:
1. The linear interpolation of PWLs in log space is mathematically
incorrect. The outcome might still be ok, as both spaces (log and
linear) are monotonic, but it is still not "right".
2. Having unitless numbers gets more error prone when we try to
harmonize the behavior over multiple platforms.
Change the algorithm to interpret the numbers as being in linear space.
This makes the interpolation mathematically correct at the expense of a
few log operations.
To account for that change, update the numbers in the tuning example
file with the linear counterparts scaled to one AWB zone measurement.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
Now that the lux module is available, add it to the rkisp1 tuner.
While at it, sort the imports correctly.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
For the lux algorithm, reference values get calculated based on a tuning
image taken at a known lux level. The reference data contains the mean Y
of the image, lux level, exposure time, gain and aperture. This module
calculates these values for insertion into the tuning file.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
To support the bayesian AWB algorithm in libtuning, the necessary data
needs to be collected and written to the tuning file.
Extend libtuning to calculate and output that additional data.
Prior probabilities and AwbModes are manually specified and not
calculated in the tuning process. Add sample values from the RaspberryPi
tuning files to the example config file.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
The rkisp1 tuner used a static module to insert the AWB algorithm into
the tuning file. Replace that with the new AWB module.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
This AWB module uses the awb function from Raspberry Pi to calculate the
needed white balance gains per colour temperature. It stores these gains
in the tuning file.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
The color temperature member of the image object was named "col" in the
past. Now it is named "color" (which is still not very expressive).
There are still a few unspotted accesses to the col member. Fix them to
access the color member.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Replace the Cam object with a list parameter to be able to call the
ctt_awb.awb() function from libtuning code.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
To be able to use the awb function copied from the Raspberry Pi tuning
scripts in the libtuning code, we need to remove the Cam object. Use the
logging framework to replace all accesses to Cam.log.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Tuning fails for raw images that don't have the channels ordered in
RGGB. In 19dc8c28f6 ("utils: tuning: libtuning: Implement the core of
libtuning") the channels of the image were reordered to RGGB
unconditionally in _read_image_dng(). That change was not applied to the
ctt_awb code, so that the channels were reordered twice. Fix by removing
the double ordering.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Tested-by: Isaac Scott <isaac.scott@ideasonboard.com>
Reviewed-by: Isaac Scott <isaac.scott@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
In d0478c41f4 ("libcamera: Rename "shutter speed" to "exposure time"")
the tuning file entry "shutter" was renamed to "exposure-time". As the
tuning files use camel cased key names, change "exposure-time" to
"exposureTime" for consistency. It doesn't break our users setups as
there are no tuning files using that entry in the wild (at least
officially).
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
The terms "shutter" and "shutter speed" are used through libcamera to
mean "exposure time". This is confusing, both due to "speed" being used
as "time" while it should be the inverse (i.e. a maximum speed should
correspond to the minimum time), and due to "shutter speed" and
"exposure time" being used in different places with the same meaning.
To improve clarity of the code base and the documentation, use "exposure
time" consistently to replace "shutter speed".
This rename highlighted another vocabulary issue in libcamera. The
ExposureModeHelper::splitExposure() function used to document that it
splits "exposure time into shutter time and gain". It has been reworded
to "split exposure into exposure time and gain". That is not entirely
satisfactory, as "exposure" has a defined meaning in photography (see
https://en.wikipedia.org/wiki/Exposure_(photography)) that is not
expressed as a duration. This issue if left to be addressed separately.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Naushir Patuck <naush@raspberrypi.com>
Acked-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Change the first parameter of Tuner.add() to accept either a list of
modules or a single module. This allows more compact code and is in sync
with Tuner.set_output_order().
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
The instances of the static modules need to be passed to
tuner.set_output_order() instead of the class. This is the reason for
the intermediate variables. To keep the code tidy, apply the same
pattern to the other modules.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
With the addition of gamma out correction the relative luminance target
was set too low. As brightness is a bit subjective it is difficult to
come up with the "correct" value. With 0.5 the patch 22 on the macbeth
chart (neutral grey, 18% reflectance) ended up a bit below 50% grey,
which seems reasonable.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
In cases where the calibration image contains super dark areas, or when
an invalid blacklevel was supplied, the grid might get close to zero or
negative. This would have bad effects on the 1/grid later. So clamp the
values to a small positive number.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
The current LSC algorithm for the rkisp1 just forwards the LSC tables to
the hardware, so absolute factors are needed and not ratios compared to
green. Therefore every channel needs to be calculated independently.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Add awb, blc, cproc, filter, and gamma to the tuning file. These don't
need any configuration.
At the moment there are no inter-module dependencies in the tuning
process. We can therefore safely sort them alphabetically. As soon as
the first dependency gets introduced (most likely lsc -> ccm) we will
see how to solve that.
The output order controls the order of processing in the IPA. It is now
also in alphabetical order which happens to be no change for the modules
that existed previously. For the others, there is no need for a specific
order.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
Add a static module class, that can be used to add static data to the
tuning file. This is propably not the best solution, but allows us to
progress without writing lots of dummy classes for static cases.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
There are valid cases where a module returns None. E.g. no images were
provided for LSC calibration. We should however define proper semantics
there. Continue with a warning for now.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Make it clear that no lsc calibration was done by returning None instead
of a incomplete configuration.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Implement a minimal ccm calibration module. For now it doesn't take the
results from lsc into account and supports rkisp1 only.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
Remove the need for the Cam object, as we don't want to port
it from Raspberry Pi.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
In the Image class the variable holding the color temperature is named
color instead of col which was used by the raspberry pi scripts. Rename
accordingly.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Based on the input images, the lsc values could exceed the range allowed
by the rkisp1. As we are now clipping the values, we can simplify the
value mapping.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Print a info on every image that gets processed and a warning on every
image that gets ignored.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
Add the missing pieces and store the result inside the image object.
This solution is not very nice, and should be refactored soon. For that
we need a concept to collect temperature and/or image specific results
in a central place. For now it serves the purpose.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
At the moment this just reads the yaml file and returns it verbatim.
This needs to evolve further in the near future.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
In the tuning datasets, the files had names like
'imx335_1600l_3000k_1.dng'. That failed on the old filename parsing
function. As there is no need to dictate the order of the tags, split
the big regex into chunks and parse them one by one. This also makes
the code easier to digest.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
The old function uses PIL to save the image, which is not in the
requirements file. As we are already requiring opencv, use that to save
images instead of an additional dependency
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
In ctt_ccm.py the logging functionality of the Cam object was used. As
we don't want to port over that class, it needs to be replaced anyways.
While at it, also replace the eprint function as it doesn't add any
value over the logging framework and misses the ability for easy log
formatting.
For nice output formatting add the coloredlogs library.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
Fix imports to match new structure in the files copied from Raspberry
Pi. Add missing imports in macbeth.py. Add missing dependencies to
requirements.txt.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
Add a requirements file to ease the installation and use of the tuning
scripts. Document that in the readme. No debian packages are provided as
rawpy is not packaged as deb. So pip has to be used anyways.
Also add pyyaml which was missing in the dependencies.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Copy visualize_macbeth_chart from raspberry pi. It is copied verbatim
and does not work in this state.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Paul Elder <paul.elder@ideasonboard.com>
Copy ctt_{awb,ccm,colors,ransac} from the raspberrypi tuning scripts as
basis for the libcamera implementation. color.py was renamed to
ctt_colors.py to better express the origin.
The files were taken from commit 66479605ba ("utils: raspberrypi: ctt:
Improve the Macbeth Chart search reliability").
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Acked-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Acked-by: Paul Elder <paul.elder@ideasonboard.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Opencv fails to load the image. The added license destroys the magic
number. Fix, by moving the licence below the magic number.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Paul ELder <paul.elder@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Port commit 66479605ba ("utils: raspberrypi: ctt: Improve the Macbeth
Chart search reliability") into libtuning.
Previously the code would brighten up images in case the Macbeth Chart
is slightly dark, and also zoom in on sections of it to look for
charts occupying less of the field of view. But it would not do both
together.
This change makes the search for smaller charts also repeat that
search for the brightened up images that it made earlier, thereby
increasing the chances of success for non-optimal tuning images.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
The Image class incorrectly parses data in GBRG bayer formats as the
indices to the channels are set incorrectly - fix it.
Signed-off-by: Daniel Scally <dan.scally@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Stefan Klug <stefan.klug@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Add the skeletal AGC module to the rkisp1 tuning script. For now it just
spits out hardcoded values.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
Add a skeletal AGC module just so that we can have some AGC tuning
values that we can use to test during development of AGC in the IPAs. As
rkisp1 is the main target, we only add support for rkisp1 for now.
The parameters are mostly copied from the hardcoded values in ctt,
except for the metering modes.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-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>
Change the names of python packages in the list of dependencies to the
exact names that should be used in the pip install command.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
Add a tuning script for rkisp1 that uses libtuning. So far it only
supports LSC.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Add a tuning script for raspberrypi for alsc only, that uses libtuning.
Since there will also be a tuning script for raspberrypi that has more
modules, put the libtuning alsc module definition in a separate file so
that it can be reused later.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Add a generator to libtuning for writing tuning output to a yaml file.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Add a parser to libtuning for parsing configuration files in yaml
format.
At the moment it doesn't parse anything and simply returns an empty
config. This is fine for the time being, as the only user of it is the
rkisp1 tuning script, which only has an LSC module which doesn't consume
anything from the configuration file. When a module comes around that
requires the yaml parser, it can be implemented then.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Add a generator to libtuning for writing tuning output to a json file
formatted the same way that raspberrypi's ctt formats them.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Add a parser to libtuning for parsing configuration files that are the
same format as raspberrypi's ctt's configuration files.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Add an LSC module for RkISP1.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Add an ALSC module for Raspberry Pi.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
