Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
    Andrey Penechko
    @MrSmith33
    What is coordinate system used for quaternions? right-handed or left-handed?
    Andrey Penechko
    @MrSmith33
    Looks like it is right-handed system. I am making camera that uses quaternions. I use that quaternion to generate rotation matrix and to rotate target, right and up camera vectors. What happens, is when I try to rotate target vector by quternion used for matrix gen it causes wrong rotation. The problem is in the order of quaternion multiply rotationQuatVert * rotationQuatHor gives correct matrix, while for vector rotation rotationQuatHor * rotationQuatVert is needed to produce correct rotation.
    Andrey Penechko
    @MrSmith33
    Figured it out, correct way was to inverse calculated matrix:
    rotationQuat = rotationQuatHor * rotationQuatVert;
    Matrix4f rotation = rotationQuat.toMatrix4x4.inverse;
    camera.target = rotationQuat.rotate(vec3(0,0,-1));
    tskarthikeyann
    @tskarthikeyann
    does anyone have sample code using dlib.a static library
    Timur Gafarov
    @gecko0307
    I'm afraid dlib doesn't have static library configuration for DUB. I'm not a DUB user, so can't add it.
    Andrey Penechko
    @MrSmith33
    @tskarthikeyann, if you use dub, it will automatically build .a/.lib file. If you need one for some purpose - look at .dub folder generated in dlib's folder.
    Basile Burg
    @BBasile
    just a word to inform that dlib can be build in Coedit via the metad meta-repository. The project file allows to build the static library, the library manager allows to integrate it in the IDE, for example a runnable module can use dlib and be executed on the fly (without project file, just a single click).
    Tristan Hume
    @trishume
    Neat project! I love languages with huge standard libraries and this kind of acts as a nice extension to Phobos.
    Andrey Penechko
    @MrSmith33
    What do you plan for compression part, gecko0307?
    Timur Gafarov
    @gecko0307
    You mean dlib.coding? I have no plans for it now. I'm working on dlib.audio and recently dove into progressive jpeg.
    tAndrii
    @shishkander
    Hi! I've been looking through the codebase but couldn't find many unittests - just 1 in fact. https://github.com/gecko0307/dlib/search?utf8=%E2%9C%93&q=test
    are there no tests thus, or am I missing them somewhere?
    Timur Gafarov
    @gecko0307
    There are still not much of them, unfortunately. Never have enough time to write unittests. You are free to add some ;)
    tAndrii
    @shishkander
    so how do you test the library?
    Nafees Hassan
    @Nafees_10_twitter
    Hi! How can I rotate a SuperImage by x degrees?
    anyone online here?
    Timur Gafarov
    @gecko0307
    There's no rotating function for images yet, but I can add it if necessary.
    Timur Gafarov
    @gecko0307
    I've added dlib.image.transform module with rotateAroundCenter function
    Roman Chistokhodov
    @FreeSlave
    DynamicArray has weird appendLeft (slice version). It appends elements of slice in reverse order. Is it intended?
    Roman Chistokhodov
    @FreeSlave
    Does deleting array of structs call their destructors?
    semsimsom
    @semsimsom

    hi,
    I have windows 10 64 bit and i try and try to install dlib for python with this command:
    python setup.py install
    but i get error below:
    running install
    running bdist_egg
    running build
    Detected Python architecture: 32bit
    Detected platform: win32
    Removing build directory C:\Users\Avatar\Desktop\dlib-master./tools/python/build
    Configuring cmake ...
    -- Building for: Visual Studio 12 2013
    -- The C compiler identification is MSVC 18.0.21005.1
    -- The CXX compiler identification is MSVC 18.0.21005.1
    -- Check for working C compiler: C:/Program Files (x86)/Microsoft Visual Studio 12.0/VC/bin/cl.exe
    -- Check for working C compiler: C:/Program Files (x86)/Microsoft Visual Studio 12.0/VC/bin/cl.exe -- works
    -- Detecting C compiler ABI info
    -- Detecting C compiler ABI info - done
    -- Check for working CXX compiler: C:/Program Files (x86)/Microsoft Visual Studio 12.0/VC/bin/cl.exe
    -- Check for working CXX compiler: C:/Program Files (x86)/Microsoft Visual Studio 12.0/VC/bin/cl.exe -- works
    -- Detecting CXX compiler ABI info
    -- Detecting CXX compiler ABI info - done
    -- Detecting CXX compile features
    -- Detecting CXX compile features - done
    -- Could NOT find Boost
    -- Found PythonLibs: C:/Python27/libs/python27.lib (found suitable version "2.7.9+", minimum required is "2.6")
    error: cmake configuration failed!

    i install cmake GUI and dont know haow can i set it.
    and i build from source with cmake command like in readme file:
    mkdir build; cd build; cmake .. ; cmake --build .

    at the end i say that, i want use dlib on python .
    does anyone can help me???

    Thanks a lot....

    Eugene Wissner
    @belka-ew
    can it be that you mean wrong dlib?
    Dlib here is Dlib for D language
    semsimsom
    @semsimsom
    I'm so sorry about it...
    Eugene Wissner
    @belka-ew
    no problem )
    semsimsom
    @semsimsom
    but i appreciate if anybody can help me, thanks.
    Timur Gafarov
    @gecko0307
    @MyLittleRobo Indeed appendLeft works incorrectly. I'll fix it. Deleting array of structs doesn't call destructors, nor does it for classes. You should delete each element manually.
    Timur Gafarov
    @gecko0307
    @semsimsom Sorry, can't help you. This chat has nothing to do with dlib for C++ or Python, this is a totally different project.
    Eugene Wissner
    @belka-ew
    Тимур, пока я исправлял ошибки для Windows я решил попробовать имплементировать заодно Microsoft's IO Completion Port (event loop для Винды) и сейчас у меня даже стало что-то вырисовываться.
    Я пока что ничего не комитю, потому что не уверен в структуре и постоянно что-то меняется. Поэтому, когда уже будет более менее полная картинка, тогда закомитю. Там придут event Loop для Windows, BSD сокеты и Windows Overlapped сокеты. Достаточно много нового.
    Но еще надо немного поработать, так что мой тикет пока побудет открытым, но я по-тихоньку приближаюсь к концу
    Andrey Penechko
    @MrSmith33
    How do you guys serialize vectors, so that they have no duplicated members? If I use getMembers trait for (de)serialization, then I get all the possible members of union, while I want only arrayof. One solution will be to add some UDA to not serialize other members
    Patrick Barker
    @grillz
    Hey does anyone know how to modify the top coordinate of a dlib rectangle?
    yeshiyuan
    @yeshiyuan00
    hello everyonr !
    Mathieu Roux
    @Begah
    Hello everyone : quick question, i moved from using Gl3n to dlib.math. I find a design choice rather puzzling. In math, (explained here for example : https://mathinsight.org/matrix_vector_multiplication), colum-vector x matrix multiplication is defined as : matrix vector, but in dlib it is defined as vector matrix. This is normal for a row vector but not a colum vector ( as stated in the comments in matrix.d line 365).
    Timur Gafarov
    @gecko0307
    This is very old design issue which I decided not to fix, because that would break too much existing code.
    Actually I think its okay since dlib has only one type of vector-matrix multiplication, contrary to GLSL for example.
    Timur Gafarov
    @gecko0307
    I also still find 'vector x matrix' notation more convenient for affine transformations, though understand that it's not common.

    Expressions are just easier to read:

    Vector3f v = (myPosition * myMatrix1) * myMatrix2;

    versus

    Vector3f v = myMatrix2 * (myMatrix1 * myPosition);

    In first case matrices are written in order in which they are applied, and the common 'inverse' approach looks a bit unintuitive to me.
    Mathieu Roux
    @Begah
    Matrix multiplication is associative, therefore :
    'Vector3f v = myMatrix2 (myMatrix1 myPosition)' is equivalent to 'Vector3f v = myMatrix2 myMatrix1 myPosition'
    Anyway, i will comply to this design decision :smile: and thanks for the info.
    Mathieu Roux
    @Begah
    Another design choice i found weird ( caused my application to act weirdly without crashing ). So in gl3n or glsl, we can do :
    vec3 v = vec3(1) and it is equivalent to vec3 v = vec3(1, 1, 1)
    In dlib, this doesn't work, vec3 v = vec3(1) gives a such a vector : ( 1, nan, nan ).
    I implement in vector.d a "fix" for this, wondering if it can be put into the normal release since it does have the potential to break working code.