This is bad: a hobbyist will be faced with a huge burden to bring anything 3D cross-platform to their audience. In the past, it was possible to use Qt or a nasty GLU/GLUT wrapper to write portable code.
Way back in 2006, in CS175, we implemented almost all of the core 1.5 pipeline in C++. Software OpenGL implementations may not be the fastest, but they’re more or less trivial (quaternions, trapezoid-based triangle engine, painters algorithm, z-buffering, texture mapping, bump-mapping, lighting and various shading models), and therefore accelerate-able with CPU and GPGPU SIMD ops.
Qt supports multiple backends for its own rendering, but the hobbyists will still have to rewrite their rendering code to support that new backend. For a lot of my programs, that's about equivalent to rewriting in another language.
Way back in 2006, in CS175, we implemented almost all of the core 1.5 pipeline in C++. Software OpenGL implementations may not be the fastest, but they’re more or less trivial (quaternions, trapezoid-based triangle engine, painters algorithm, z-buffering, texture mapping, bump-mapping, lighting and various shading models), and therefore accelerate-able with CPU and GPGPU SIMD ops.