profile
viewpoint
If you are wondering where the data of this site comes from, please visit https://api.github.com/users/mirsadm/events. GitMemory does not store any data, but only uses NGINX to cache data for a period of time. The idea behind GitMemory is simply to give users a better reading experience.
Mirsad Makalic mirsadm Edinburgh

mirsadm/motioncam 62

Motion Cam is a camera application for Android that replaces the entire camera pipeline. It consumes RAW images and uses computational photography to combine multiple images to reduce noise.

mirsadm/Halide 0

a language for fast, portable data-parallel computation

issue commentmirsadm/motioncam

"Enable dual exposure controls" causes crash at startup

Cheers will take a look

genpfault

comment created time in 10 hours

issue closedmirsadm/motioncam

Latest version causes purple color in overexposed areas

Hi, I have some raw bayer data which i average using libMotionCam binary. It was working fine but today i built the latest version from master branch (commit hash 83a52b7984d58ac8dc895ce02e42f63e31a50975) and running it on same data, i noticed that all overexposed areas are being displayed as purple. Here is a preview generated from such raw data: frame_8

Now if i average it by feeding it as a zip (containing all required metadata) to libMotionCam binary on my mac, this is what i get: Archive zip

i can deduce that the preview functions are working fine, it seems that the post processing is broken and some clipping is occurring in those bright regions. Note that same zip which contains multiple frames of this shot with same settings produces good output with earlier versions of libMotionCam binary. I tried with several images and all have this purple color in overexposed regions.

Here is the post processing settings i used:

"postProcessingSettings": {
        "blacks": 0.032499998807907104,
        "contrast": 0.5,
        "flipped": false,
        "noiseSigma": 0,
        "chromaEps": 32,
        "sceneLuminance": 0,
        "spatialDenoiseAggressiveness": 1,
        "temperature": 3678,
        "tonemapVariance": 0.25,
        "whitePoint": 1,
        "sharpen1": 1.2999999523162842,
        "shadows": 4.0658516883850098,
        "blueSaturation": 1,
        "jpegQuality": 98,
        "greenSaturation": 1,
        "saturation": 1,
        "exposure": 0,
        "gamma": 2.2000000476837158,
        "tint": 24,
        "sharpen0": 3.4000000953674316
    }

Is this a bug or does the latest version requires shooting dark exposures and needs HDR fusion to fix it?

closed time in 3 days

f0enix

issue commentmirsadm/motioncam

Latest version causes purple color in overexposed areas

That's been fixed in the latest commit. Thanks for reporting that.

f0enix

comment created time in 3 days

push eventmirsadm/motioncam

Mirsad Makalic

commit sha f8be3eb56888f26dbb396be4d346bf42078013d6

- Improved colour denoising - Hot pixel suppression

view details

push time in 3 days

push eventmirsadm/motioncam

Mirsad Makalic

commit sha a93460ae6f3a383fef909297b3709163418662c1

- Revert hot pixel suppression

view details

push time in 4 days

issue commentmirsadm/motioncam

Latest version causes purple color in overexposed areas

Cheers, I've downloaded it. Will look at it tonight if I can. Feel free to delete it from there.

f0enix

comment created time in 7 days

issue commentmirsadm/motioncam

Latest version causes purple color in overexposed areas

are you able to provide the zip file for me? it certinaly looks like a clipping bug

f0enix

comment created time in 7 days

issue closedmirsadm/motioncam

HDR Image Quality

Hi @mirsadm,

HDR image output in some scenes contain a big white area have more noise. I think it appear due to you use only single under-exposure frame without denoise and a-part of white area come from under-exposure frame.

closed time in 7 days

make-stronger

created tagmirsadm/motioncam

tag4.0.0

Motion Cam is a camera application for Android that replaces the entire camera pipeline. It consumes RAW images and uses computational photography to combine multiple images to reduce noise.

created time in 9 days

release mirsadm/motioncam

4.0.0

released time in 9 days

push eventmirsadm/motioncam

Mirsad Makalic

commit sha 83a52b7984d58ac8dc895ce02e42f63e31a50975

- Camera switching - Processing improvements

view details

push time in 9 days

push eventmirsadm/motioncam

Mirsad Makalic

commit sha 07f836bf4336d10e859b7fc2bd7b9d31f2cb87c1

- Denoising improvements - Post processing improvements - Blue noise dithering

view details

push time in 18 days

issue closedmirsadm/motioncam

Compare with Hdr-plus.

Hi @mirsadm,

Did you make a comparison with hdr-plus (timothybrooks implemented) on the results of merged raw image?

closed time in 24 days

bxomark

issue commentmirsadm/motioncam

Compare with Hdr-plus.

No I have not. That is a reimplementation of the original HDR+ paper from Google. This project doesn't really have too much in common with that. Feel free to compare the result, it would be an interesting experiment.

bxomark

comment created time in 24 days

push eventmirsadm/Halide

dpalermo

commit sha ebae3cd7ce0f52ea82a3c93eaf794df9d12304f0

Changes for building with Hexagon SDK 4.3.0.0 & android-ndk-r19c (#6072) * Changes for building with Hexagon SDK 4.2.0.2 & android-ndk-r19c * Drop libsim_qurt_vtcm.a (now part of libsim_qurt.a) * Fix for clang-format-lint * Update to use SDK 4.3.0.0 / HEXAGON_Tools 8.4.11 * Updated binaries & README.md * trigger buildbots * Updated binaries after merge of master * Update SDK comment for >sm8350 Co-authored-by: Steven Johnson <srj@google.com>

view details

Evan Lee

commit sha 292a35adb86621d7eb92f96f470390d5da4cbeca

Added fixes to issues regarding using HALIDE_DEBUG_MATCHED_RULES (#6088) * added operator<< for IsMaxValue, IsMinValue, and moved build_replacement(after) to be called before debug matched rules Co-authored-by: Steven Johnson <srj@google.com>

view details

Volodymyr Kysenko

commit sha 4fda2c6a2ccbdf67e505862335cab17be2136978

Handle negative shifts in CodeGen_C (#6087) * Handle negative shifts in CodeGen_C * trigger buildbots * Emit code directly if shift was casted to signed int Co-authored-by: Steven Johnson <srj@google.com>

view details

Steven Johnson

commit sha a590c1771866954b8f81b9529aada60499ae3120

[hannk] Remove unused Op::clone() methods (#6092) We don't call these anymore, so remove them and the related TensorMap code.

view details

Steven Johnson

commit sha 66ff71f2e920e26fcb46284142294589f9d06d2d

[hannk] Cleanup: move SmallVector, Tensor to their own source files (#6091) * Move SmallVector, Tensor to their own files * cleanup

view details

Steven Johnson

commit sha 27ae113842582b07c7cf0dfcd8fd03186dbfbba8

[hannk] More Hygiene (#6093) * [hannk] More Hygiene - TensorStorage takes a more sensible set of args for ctor - Tensors don't need to be movable or copyable - Since we are now using C++17, we can use std::make_unique instead of make_op * Restore make_op * clang-format * Remove unnecessary TensorStorage methods

view details

Volodymyr Kysenko

commit sha d81f5c3e9cf1dc513a75ff0a6aa263cea25bd437

Provide bounds of rvars for all functions in the fused group (#6078) * Provide bounds of rvars for all functions in the fused group * Just use constant * Comments + rename variable

view details

Steven Johnson

commit sha 5aeb8db54f9beaf1c1d761d4ce696ee29a9a1b3e

[hannk] Don't mark Tensors as input or output (#6094) * Refactor transforms.cpp, no functional change * Use Op::is_input(), Op::is_output * Update configure_cmake.sh

view details

Dillon Sharlet

commit sha 45f31f7c8389c2ffffb724c62d7c0fbaa6dab099

Fix is_monotonic issue (#6081) (#6083) * Fix #6081 * Slightly less bizarre implementation of select visitor. Co-authored-by: Steven Johnson <srj@google.com>

view details

Steven Johnson

commit sha b94a5266b2cfb2cc312a873647d43f903b2c165d

[hannk] Replace Tensor::set_external_host with set_external_buffer (#6100) * Replace Tensor::set_external_host with set_external_buffer * Also remove stale comment

view details

Steven Johnson

commit sha d82fec4428944c46a40756c7c0bc6afd21b8a4c4

[hannk] Fix various build glitches for Bazel/Blaze (#6098) - Make small_vector.h standalone-compilable - Move Tensor::replace_all_consumers_with() to a local function near PadForOps to dodge a circular include dep between Tensor and Model

view details

Steven Johnson

commit sha 93292a28b1596b2fc5bd85dd947060f8db1c31c9

[hannk] Add --keep_going flag to ModelRunner (#6101) This allows you to run a compare operation against a bunch of graphs without exiting at the first one that is out-of-spec for comparison. (Useful when you want to verify that no *new* differences are introduced by a change.)

view details

Steven Johnson

commit sha f285f0866cd53a5582618fefb9039ea6c149f309

[hannk] Minor cleanups (#6103) * [hannk] Minor cleanups * Restore get_tensor

view details

Alexander Root

commit sha 2da7ca5c2350aedf2e97151d54c1dc7b46e630c5

Call simplify and remove_likelies for find_constant_bounds (#6099)

view details

Dillon Sharlet

commit sha a987222608c84756b305fec23d28d8e36d264f70

Remove likelies and promises before trying to check for monotonicity. (#6105)

view details

Dillon Sharlet

commit sha 7791e844cb76bdf5a2a11d862d67b6c30040d350

Remove floats from extern_producer (#6109) * Don't rely on floats/trig unnecessarily * Use different period

view details

Steven Johnson

commit sha 3f3dd702a0be2a511525dbbabf5859298f010557

Disable Ubuntu Packaging Action (Issue #6111) (#6112) * Disable Ubuntu Packaging Action (Issue #6111)

view details

Steven Johnson

commit sha 0da1354ba40c473c492182a70741289168fa906a

Avoid pathological cases in halide_benchmark() (#6110) In the variant that tries to compute a good samples/iters value based on min_time, there's a pathological case if the environment's timer is relatively coarse, and the op being profiled is relatively fast; in this case, you can end up with timings that are very close to zero (or *literally* zero), and our attempt to calculate the number of iterations can explode into the billions, making the benchmark appear to hang (as it may take an absurd length of time to run). To fix this, add a maximum value for iters_per_sample, and smarten the calculation for when the measured time is tiny.

view details

Alex Reinking

commit sha 28165672d981c2f2c182674002e7d8915dc18295

Enable ubuntu packaging (#6113) * Revert "Disable Ubuntu Packaging Action (Issue #6111) (#6112)" This reverts commit 3f3dd702 * Explicitly update to avoid out of date package lists. Fixes #6111

view details

Kai Wolf

commit sha bfd9ceade12b7be14b594c2a39061b79fb46b920

Update LoopNest.cpp (#6086) Remove obsolete assert for output accessing other outputs

view details

push time in 24 days

push eventmirsadm/motioncam

Mirsad Makalic

commit sha 6aa3531782c1799e96e3daf67af0f07134aaee79

- Dynamically choose underexposed settings for HDR - Add local contrast enhancement - Smoother preview in burst mode

view details

Mirsad Makalic

commit sha 56a038c19007c765aaeb38cf2221752e762403e6

- Processing improvements

view details

push time in a month

push eventmirsadm/motioncam

Mirsad Makalic

commit sha 103adaf8c515488563e015744a6a028d4d1564d8

- Colour processing improvements

view details

push time in a month

push eventmirsadm/motioncam

Mirsad Makalic

commit sha c56f2fa727430b274c1e84ed14069a104add8fec

- Colour processing improvements

view details

push time in 2 months

push eventmirsadm/motioncam

Mirsad Makalic

commit sha daafd8a01175aa09207fcd4c212bdb1a4c5b96e5

- Refinements

view details

push time in 2 months

push eventmirsadm/motioncam

Mirsad Makalic

commit sha 16bce4e2e52317c1b1a590448a097427b1296f86

- Refinements

view details

push time in 2 months

push eventmirsadm/Halide

Dillon Sharlet

commit sha 33308d9f3441be7f3861d87d358663b27ba43ccf

Add pmaddubsw support (#5997) * Add pmaddubsw support * Move pmaddubsw checks to ssse3 * These patterns rae a bit finnicky

view details

Andrew Adams

commit sha 6bb87cf58454626a725f59dff1f9615350ea0838

Stop interleaving stores from generating too-large vectors (#5996) * Stop interleaving stores from generating too-large vectors * Remove integer constant * Use mul_would_overflow helper instead

view details

Alex Reinking

commit sha b2947e9c54e045cf1be3656fed33e4921922275e

Fix Windows apps (#5999) * Place DLLs on Windows by copying. * Disable Hannk on Windows by default

view details

Alexander Root

commit sha c61b8cb0c3b385a72ecf5f9434f29958cd4cadde

Guard against overflow in constant folding for EQ rewrite rules (Fixes #5998) (#6002) * guard against constant folding overflow in EQ rewrite rules * trigger buildbots Co-authored-by: Steven Johnson <srj@google.com>

view details

Alex Reinking

commit sha 568d18cd71d503c017a2e217b020dfda8047c398

Fix dependencies in FFT app.

view details

Alex Reinking

commit sha 5cdbcb0087a3016de154cf62e2a1a69bcece36a1

Re-work packaging to support complex formats (like DEB)

view details

Alex Reinking

commit sha 02a8f878af3cf306f730b087a6def9fb7f155861

Remove same-directory shared/static mixing

view details

Alex Reinking

commit sha f0db1c6f6f0f40126c07594738dba1a8cec68ea4

Split Halide CMake helpers into separate package.

view details

Alex Reinking

commit sha bf3258529fdc01847ab146687970d40989c79167

Move packaging support files into common directory.

view details

Alex Reinking

commit sha a81b86b0010fc58a9fa4ced080dd979a70b7112f

Move packaging scripts from tools/ to packaging/<type>/

view details

Alex Reinking

commit sha 3f3ce62c7a5b0cace8fd1d1bf9c163b011c40ce3

Use presets in tgz/package.sh

view details

Alex Reinking

commit sha 75a079a853e98a7d1d7e2e685fb1242089a41460

Use presets in zip/package.bat

view details

Steven Johnson

commit sha e0ac07b3fee14b57329bc5c7f294b9dd6ef53881

Upgrade hannk TFLite version to 2.5.0 (#6009) https://github.com/tensorflow/tensorflow/releases/tag/v2.5.0

view details

sksarda

commit sha 69eed6ea0ad97d7cb800b68d5c9ea6bb3cb53809

Add -fpic option to debug version on non-windows .ll file (#6000) Else, it eliminates reference to global offset table for symbols to be resolved from remote library causing runtime crash with -debug option. Co-authored-by: Suyog Sarda <ssarda@codeaurora.org>

view details

Volodymyr Kysenko

commit sha b829d123e6326de9a5669bafec8b5c092ae5f476

Support Shuffle::extract_element from list of scalars in CodeGen_C (#6007)

view details

Volodymyr Kysenko

commit sha 878c3ecb47fee46cbc47052b5d51e6f903fec9c2

Fix CodeGen_C::print_scalarized_expr (#6006) * Fix CodeGen_C::print_scalarized_expr * CppVector/NativeVector object doesn't have .replace() anymore. * Initialize vector with zero to avoid warning. * Actually, can't assign to CppVector (only to NativeVector), so do ::broadcast instead * Leave it uninitialized

view details

Andrew Adams

commit sha 50d7640010413a36c08771c516e0de64ae9bc228

Permit "safe" parallel scatters, even when they race (#4841) lets the atomic() scheduling directive also apply to simple assignments in addition to associative commutative operators, e.g. hist(f(r.x), x) = g(x) is safe to parallelize over r.x if the stores are atomic, because the RHS doesn't depend on the hist or r.x

view details

Andrew Adams

commit sha deea5ecb1ae4d7c329224162e57884a57c9aa88c

Substituting complex expressions for constrained scalar inputs makes … (#6014) * Substituting complex expressions for constrained scalar inputs makes a mess Substituting in constants and variables is probably fine. * Remove extra loop Co-authored-by: Dillon Sharlet <dsharlet@google.com>

view details

Andrew Adams

commit sha adb1e054e1d712e7156daa772b12683f6f8e3d31

Fix local laplacian upsample (#6011)

view details

Alex Reinking

commit sha ed6ecccc81aab80502cc1eaccc0da07f26550174

fix package.sh perms (#6016)

view details

push time in 2 months

push eventmirsadm/motioncam

Mirsad Makalic

commit sha 705dfdc4b50a09d165aaa565db57644ad59ef428

- Refinements

view details

push time in 2 months

issue commentmirsadm/motioncam

HDR Image Quality

Hey! Thanks for the feedback. Yes I did notice this myself. It is being too aggressive in trying to recover highlights. I am tweaking it further to avoid situations such as these. I'll create another build soon.

make-stronger

comment created time in 2 months

push eventmirsadm/motioncam

Mirsad Makalic

commit sha 92c7ba4a0ddc020056fadbcddbb1d191e014b94e

- Fix previous previous commit

view details

push time in 2 months

push eventmirsadm/motioncam

Mirsad Makalic

commit sha d478217c75e86e02f39e5076bc0edc9cc95fc1ec

- Fix previous commit

view details

push time in 2 months

push eventmirsadm/motioncam

Mirsad Makalic

commit sha 1c3d85e8d4f186695cd7c0d5632a30c3f8e5595f

- Denoise tweak

view details

push time in 2 months

issue commentmirsadm/motioncam

Lucky Features

-Use gyroscope to select best frame as default

In burst mode it will attempt to measure the sharpness of the first few images and pick the best one. I'm not sure how the gyro would help there? I did use to do the same in normal photo mode but took it out because it can lead to images being captured that were too far from the time the shutter button was pushed. I might put that back in but instead only choose from 1 or 2 images closest to the time the shutter button was pressed.

-Use gyroscope and sensor noise model to choose best iso/exposure time, automatics isn't designed for stacking

If you shoot in "night mode" it does override the iso/shutter speed. It behaves in a similar way to enhanced HDR+ in GCam. I did attempt to reimplment the auto exposure to pick more approrpiate settings but it is difficult to make it work quickly enough. The feedback loop between measuring and adjusting the settings is too long. On my phone the automatic exposure tends to pick much faster shutter speeds than the built in camera.

eszdman

comment created time in 2 months

push eventmirsadm/motioncam

Mirsad Makalic

commit sha 647a6fb6d5daab768ad5ed44dfd984c1698a0eb8

- Don't run wavelet code if spatial denoising <= 0

view details

push time in 2 months

push eventmirsadm/motioncam

Mirsad Makalic

commit sha 9314516ecb162025ac39929f41a6ca448c859df5

- Denoising performance improvements - Cleanup

view details

push time in 2 months

issue commentmirsadm/motioncam

Pixel 4a OpenCL support

Renderscript has been a bit of a disaster IMO so I'm glad they're moving away from it. Unfortunately I don't think Vulkan compute shaders are in good position at the moment either. OpenCL still seems to be the way to go even though its support is somewhat limited.

paolodepetrillo

comment created time in 2 months