I'm trying to get color for my color picker, but I get a float value, that differ from value I stored. For example I set 0.5, but I take 0.498039 (this is the real numbers). I don't build any FBO and read color from GL_BACK directly:
glReadBuffer(GL_BACK);
glReadPixels(x, y, 1, 1, GL_RGB, GL_FLOAT, &color);
How can I save precision of floating point value? Is it possible to change GL_FLOAT to another which would save precision? Is it possible to get in &color numbers greater than 1.0?
The precision is limited by the precision of the frambuffer (back buffer). This precision cannot be set individually and is (most likely) limited to 8 bit per channel. Actually the Default framebuffer is generated once, when the OpenGL window and OpenGL Context is generated.
Thus it makes no sense to read the read the buffer to 32bit float target, because the source buffer just has 8 bits.
Anyway it is possible to render the scene to a named Framebuffer Object where the color plane of the attached Renderbuffer has a floating point format (e.g. GL_RGBA32F). See LearnOpenGL - Framebuffers.