Dual quaternion by chargerKong · Pull Request #19026 · opencv/opencv

chargerKong · 2020-12-06T13:22:26Z

Pull Request Readiness Checklist

See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request

I agree to contribute to the project under Apache 2 License.
To the best of my knowledge, the proposed patch is not based on a code under GPL or other license that is incompatible with OpenCV
The PR is proposed to proper branch
There is reference to original bug report and related work
[] There is accuracy test, performance test and test data in opencv_extra repository, if applicable
Patch to opencv_extra has the same branch name.
The feature is well documented and sample code can be built with the project CMake

build_image:Custom=centos:7
buildworker:Custom=linux-1,linux-4,linux-6

basic operations, functions(exp,log,norm,inv), to/from mat, sclerp.

modules/core/include/opencv2/core/dualquaternion.hpp

modules/core/include/opencv2/core/dualquaternion.inl.hpp

chargerKong · 2021-01-03T16:26:48Z

Happy new year! @savuor
I’m sorry for not updating the code last month because I went to the headquarters to participate in the oral defense.
I'll update this next week

change algorithm of norm, inv, getTranslation, createFromPitch, normalize； change type translation to Vec3; comment improve;

savuor · 2021-01-16T00:06:26Z

modules/core/include/opencv2/core/dualquaternion.hpp

+                  std::vector<Vec<_Tp, 3>> &out_normals,
+                  const std::vector<DualQuat<_Tp>> &dualquat,
+                  const std::vector<std::vector<_Tp>> &weights,
+                  const std::vector<std::vector<int>> &joints_id,


It's better to use InputArray and OutputArray for vertices, normals, weights and ids instead of vectors. Internally the type and dimensionality should be checked.

savuor · 2021-01-16T00:39:13Z

modules/core/include/opencv2/core/dualquaternion.inl.hpp

+         std::vector<Vec<T, 3>> &out_vert, std::vector<Vec<T, 3>> &out_normals,
+         const std::vector<DualQuat<T>> &_dualquat, const std::vector<std::vector<T>> &weights,
+         const std::vector<std::vector<int>> &joint_id, QuatAssumeType assumeUnit)
+{


I think this function should be moved to some .cpp file. Here's motivation why:
Typical use case for this function is to transform point clouds and vertices which are counted in thousands. This may involve massive parallel processing using parallel_for_, SIMD or OpenCL.
Having such a code in a header makes no sense; this would slow down the compilation. Having this in .inl.hpp has even less sense since this is not a good function to be inlined.

OK, I've temporarily removed this function, but have no idea where to put it

You can create a separate .cpp file, like dqs.cpp or something

Is it possible to separate declarations and definitions of template functions? I did some consulting and I saw that most of the template class functions that can be run were written together. Is there a way to separate them?

modules/core/include/opencv2/core/dualquaternion.inl.hpp

savuor · 2021-01-16T00:50:52Z

modules/core/include/opencv2/core/dualquaternion.inl.hpp

+    }
+    for (size_t i = 0; i < in_vert.size(); ++i)
+    {
+        DualQuat<T> dq_blend;


Here gdqblend can be reused with QUAT_ASSUME_UNIT argument

add func(obj) operations;

add assumeUnit in getRotation; remove dqs; change std::vector to InputArray

savuor

Most of the code looks good to me.
The only thing to change is to use InputArray instead of vector<DualQuat<T>> or Vec<DualQuat<T>, n>. This makes sense not only for gdqblend but for dqs as well. A type should be checked along with size, it's OK to allow floats and doubles only in such functions.
Doing this would resolve the problem with definition of template function in .cpp file since there'll be no templates.
P.S. Sorry for not answering for such a long time.

savuor · 2021-01-29T14:52:28Z

modules/core/include/opencv2/core/dualquaternion.hpp

+     * @note the type of weights' element should be the same as the date type of dual quaternion inside the dualquat.
+     */
+    template <int cn>
+    static DualQuat<_Tp> gdqblend(const Vec<DualQuat<_Tp>, cn> &dualquat, InputArray weights,


I think dualQuat should be an InputArray too, as well as weights, with type/size checks

OK. I‘ll complete it next week

I found it will raise an error when Vec<DualQuat<double>, n> converts into InputArray, and Matx will also happen like this

error: ‘type’ is not a member of ‘cv::DataType<cv::DualQuat<double> >’

Maybe to fix this by introducing a separate alias for this function with Vec<DualQuat<float/double>> as an argument type? Like, there'll be a function gdqblend(InputArray dualQuat,...) that requires dualQuat to have 8 rows/channels of float/double. And 2 functions gdqblend(const Vec<DualQuat<float/double>, 8> &dualquat,... that call the InputArray version.

Do you mean to call function gdqblend(const Vec<DualQuat<float/double>, n> ... ) in function gdqblend(InputArray dualQuat,...) ?

The opposite: the basic function is the one with InputArray, it's called by the function with const Vec<DualQuat<float/double>, n>.

…dualquat

savuor · 2021-02-16T17:18:26Z

@chargerKong @alalek Looks like #19538 fixes the problems in this build. Can we merge the one to another?

alalek · 2021-02-16T21:22:47Z

There is no automatic merge process between patches.
Multiple "merge" commits in mentioned PR makes review of changed parts hard.

Dual quaternion * create dual quaternion; basic operations, functions(exp,log,norm,inv), to/from mat, sclerp. * add dqb, dqs, gdqb, to/from affine3; change algorithm of norm, inv, getTranslation, createFromPitch, normalize； change type translation to Vec3; comment improve; * try fix warning: unreferenced local function * change exp calculation; add func(obj) operations; * Change the algorithm of log function; add assumeUnit in getRotation; remove dqs; change std::vector to InputArray * fix warning: doxygen and Vec<double, 0> * fix warning: doxygen and Vec<double, 0> * add inputarray param for gdqb * change int to size_t * win cl warning fix * replace size_t by int at using Mat.at() function * replace double by float * interpolation fix * replace (i, 0) to (i) * core(quat): exclude ABI, test_dualquaternion=>test_quaternion.cpp Co-authored-by: arsaratovtsev <arsaratovtsev@intel.com> Co-authored-by: Alexander Alekhin <alexander.a.alekhin@gmail.com>

chargerKong force-pushed the dualquat branch 2 times, most recently from 84cf64b to cd8ef0c Compare December 6, 2020 14:26

create dual quaternion;

7fd2080

basic operations, functions(exp,log,norm,inv), to/from mat, sclerp.

chargerKong force-pushed the dualquat branch from cd8ef0c to 7fd2080 Compare December 6, 2020 15:12

asmorkalov requested a review from vpisarev December 7, 2020 08:54

asmorkalov added category: core feature labels Dec 7, 2020

savuor reviewed Dec 9, 2020

View reviewed changes

modules/core/include/opencv2/core/dualquaternion.inl.hpp Show resolved Hide resolved

savuor reviewed Dec 10, 2020

View reviewed changes

modules/core/include/opencv2/core/dualquaternion.inl.hpp Outdated Show resolved Hide resolved

savuor reviewed Dec 10, 2020

View reviewed changes

modules/core/include/opencv2/core/dualquaternion.inl.hpp Outdated Show resolved Hide resolved

savuor reviewed Dec 10, 2020

View reviewed changes

modules/core/include/opencv2/core/dualquaternion.inl.hpp Outdated Show resolved Hide resolved

savuor reviewed Dec 10, 2020

View reviewed changes

modules/core/include/opencv2/core/dualquaternion.inl.hpp Outdated Show resolved Hide resolved

chargerKong commented Dec 10, 2020

View reviewed changes

modules/core/include/opencv2/core/dualquaternion.inl.hpp Show resolved Hide resolved

chargerKong commented Dec 10, 2020

View reviewed changes

modules/core/include/opencv2/core/dualquaternion.inl.hpp Show resolved Hide resolved

chargerKong commented Dec 10, 2020

View reviewed changes

modules/core/include/opencv2/core/dualquaternion.inl.hpp Outdated Show resolved Hide resolved

chargerKong force-pushed the dualquat branch from 0eee60e to d561a1c Compare January 9, 2021 07:47

add dqb, dqs, gdqb, to/from affine3;

0039634

change algorithm of norm, inv, getTranslation, createFromPitch, normalize； change type translation to Vec3; comment improve;

chargerKong force-pushed the dualquat branch from d561a1c to 0039634 Compare January 9, 2021 08:28

Merge remote-tracking branch 'opencv/master' into dualquat

4e9a19e

chargerKong force-pushed the dualquat branch from 6d61a50 to 4e9a19e Compare January 13, 2021 07:48

try fix warning: unreferenced local function

c5460a2

savuor reviewed Jan 16, 2021

View reviewed changes

modules/core/include/opencv2/core/dualquaternion.inl.hpp Outdated Show resolved Hide resolved

savuor reviewed Jan 16, 2021

View reviewed changes

modules/core/include/opencv2/core/dualquaternion.inl.hpp Outdated Show resolved Hide resolved

savuor reviewed Jan 16, 2021

View reviewed changes

modules/core/include/opencv2/core/dualquaternion.inl.hpp Outdated Show resolved Hide resolved

savuor reviewed Jan 16, 2021

View reviewed changes

change exp calculation;

6045698

add func(obj) operations;

chargerKong and others added 3 commits January 19, 2021 17:14

Change the algorithm of log function;

3548ff7

add assumeUnit in getRotation; remove dqs; change std::vector to InputArray

fix warning: doxygen and Vec<double, 0>

5245791

fix warning: doxygen and Vec<double, 0>

32834d9

chargerKong force-pushed the dualquat branch from 5245791 to 32834d9 Compare January 20, 2021 08:22

chargerKong changed the title ~~[WIP]Dual quaternion~~ Dual quaternion Jan 21, 2021

savuor requested changes Jan 29, 2021

View reviewed changes

chargerKong and others added 9 commits February 15, 2021 12:52

add inputarray param for gdqb

d6022de

Merge branch 'dq_t' into dualquat

adbb09a

change int to size_t

62ab588

Merge branch 'master' of https://github.com/opencv/opencv into dualquat

b8f379b

win cl warning fix

d6a38c5

replace size_t by int at using Mat.at() function

480f58e

replace double by float

86986b5

interpolation fix

b6ffc64

Merge branch 'dualquat' of https://github.com/DumDereDum/opencv into …

d7eb160

…dualquat

replace (i, 0) to (i)

dc31a28

savuor approved these changes Feb 17, 2021

View reviewed changes

core(quat): exclude ABI, test_dualquaternion=>test_quaternion.cpp

63d901c

alalek assigned savuor Feb 17, 2021

alalek merged commit 0aca3fb into opencv:master Feb 17, 2021

alalek mentioned this pull request Feb 17, 2021

Dualquat minor fixes #19538

Closed

6 tasks

alalek mentioned this pull request Apr 9, 2021

(5.x) Merge 4.x #19885

Merged

Uh oh!

Conversation

chargerKong commented Dec 6, 2020 • edited by alalek Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Pull Request Readiness Checklist

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

chargerKong commented Jan 3, 2021

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

savuor left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

savuor commented Feb 16, 2021

Uh oh!

alalek commented Feb 16, 2021

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

chargerKong commented Dec 6, 2020 •

edited by alalek

Loading

savuor left a comment •

edited

Loading