Qt 5.5 QOpenGLTexture copying data issue - c++

Question

I'm working with Qt 5.5 OpenGL wrapper classes. Specifically trying to get QOpenGLTexture working. Here I am creating a 1x1 2D white texture for masking purposes. This works:
void Renderer::initTextures()
{
QImage white(1, 1, QImage::Format_RGBA8888);
white.fill(Qt::white);
m_whiteTexture.reset(new QOpenGLTexture(QOpenGLTexture::Target2D));
m_whiteTexture->setSize(1, 1);
m_whiteTexture->setData(white);
//m_whiteTexture->allocateStorage(QOpenGLTexture::RGBA, QOpenGLTexture::UInt32);
//m_whiteTexture->setData(QOpenGLTexture::RGBA, QOpenGLTexture::UInt8, white.bits());
// Print any errors
QList<QOpenGLDebugMessage> messages = m_logger->loggedMessages();
if (messages.size())
{
qDebug() << "Start of texture errors";
foreach (const QOpenGLDebugMessage &message, messages)
qDebug() << message;
qDebug() << "End of texture errors";
}
}
However I am now trying to do two things:
Use allocate + setData sequence as separate commands (the commented out lines), e.g.
m_whiteTexture->allocateStorage(QOpenGLTexture::RGBA, QOpenGLTexture::UInt32);
m_whiteTexture->setData(QOpenGLTexture::RGBA, QOpenGLTexture::UInt8, white.bits());
for the purpose of more complicated rendering later where I just update part of the data and not reallocate. Related to this is (2) where I want to move to Target2DArray and push/pop textures in this array.
Create a Target2DArray texture and populate layers using QImages. Eventually I will be pushing/popping textures up to some max size available on the hardware.
Regarding (1), I get these errors from QOpenGLDebugMessage logger:
Start of texture errors
QOpenGLDebugMessage("APISource", 1280, "Error has been generated. GL error GL_INVALID_ENUM in TextureImage2DEXT: (ID: 2663136273) non-integer <format> 0 has been provided.", "HighSeverity", "ErrorType")
QOpenGLDebugMessage("APISource", 1280, "Error has been generated. GL error GL_INVALID_ENUM in TextureImage2DEXT: (ID: 1978056088) Generic error", "HighSeverity", "ErrorType")
QOpenGLDebugMessage("APISource", 1281, "Error has been generated. GL error GL_INVALID_VALUE in TextureImage2DEXT: (ID: 1978056088) Generic error", "HighSeverity", "ErrorType")
QOpenGLDebugMessage("APISource", 1281, "Error has been generated. GL error GL_INVALID_VALUE in TextureSubImage2DEXT: (ID: 1163869712) Generic error", "HighSeverity", "ErrorType")
End of texture errors
My mask works with the original code, but I can't get it to work in either (1) and (2) scenarios. For (2) I change the target to Target2DArray, change the size to include depth of 1, adjust my shaders to use vec3 texture coordinates and sampler3D for sampling, etc. I can post a more complete (2) example if that helps. I also don't understand these error codes and obviously difficult to debug on the GPU if that is what is going wrong. I've tried all sorts of PixelType and PixelFormat combinations.
Thanks!

Answer 1

This question is very old, but I just came across a similar problem myself. For me the solution was to call setFormat before
m_whiteTexture->setFormat(QOpenGLTexture::RGBA8_UNorm);

Answer 2

As I found out here: https://www.khronos.org/opengl/wiki/Common_Mistakes#Creating_a_complete_texture
The issues with the original code, is that the texture is not complete.
As mentioned by #flaiver, using QOpenGLTexture::RGBA8_UNorm works, but only because Qt uses different kind of storage for this texture (effectively it uses glTexStorage2D, and that is even better, as per OpenGL documentation), which is not the case for QOpenGLTexture::RGBA.
To make the texture work, even if you do require specifically QOpenGLTexture::RGBA (or some other formats, e.g. QOpenGLTexture::AlphaFormat), you need either set texture data for each mipmap level (which you don't really need for your case), or disable using mipmaps:
// the default is `QOpenGLTexture::NearestMipMapLinear`/`GL_NEAREST_MIPMAP_LINEAR`,
// but it doesn't work, if you set data only for level 0
// alternatively use QOpenGLTexture::Nearest if that suits your needs better
m_whiteTexture->setMagnificationFilter(QOpenGLTexture::Linear);
m_whiteTexture->setMinificationFilter(QOpenGLTexture::Linear);
// // optionally a good practice is to explicitly set the Wrap Mode:
// m_whiteTexture->setWrapMode(QOpenGLTexture::ClampToEdge);
right after you allocate the storage for texture data.