So I've decided that I'm officially friendzoning OpenGL.
I really want to like OpenGL. I really do. However, after everything we've been through, after both of us seeing each others' glaring faults come to light, it's just not the same. As hard as it is, I'll never be able to see OpenGL as more than a friend. A friend that I go to when I want to prototype something up in a jiffy, or I don't really want to have to hassle around with D3D11. I understand OpenGL, and we've been through a lot...
OpenGL, I hate to say it, but...
I have a crush on the new kid in town, Vulcan.
(Actually, I'd personally just want a graphics library that gives me the power to send data to VRAM and access it via a shader. Anything else is bonus. Allow me to write my own numbers to my buffers, allow me to full control of the resources I've allocated in VRAM. Let me incur any sort of cache misses or other nasty things. Allow me to write and interpret my own z-buffers, to write and read from RAM in the same way that you can in a multithreaded CPU program. I understand it's dangerous, but at least give me the option! No more complicated binding, and setting render targets, please, just give me the opportunity to read and write to VRAM!)
(Although, I'm probably just talking out my ass here. I know very little about GPU architecture compared to a lot of other people who roam these forums. I just get frustrated because I feel like there are a lot of barriers holding me back from doing interesting things easier than normal. I hate having to tell the graphics driver that I'm allocating a depth buffer, and that I want the RGB data to be full floating point, blah blah blah. I would just really like to be able to write and read my own data at will. However wrong I may be, this is 'MURICA goddammit! I can be ignorant and loud as much as I want!)
In other news, or rather lack thereof, I haven't really been working on my project for the last week or so. I've been trying to see if Java could ever be a decent platform to test rendering algorithms on (hint: it's not), because after my cone tracing experiment fell apart I wanted to see if I could improve my sampling patterns in a path tracer. It turns out the JVM JIT optimizer isn't quite a match for good old -O3.
*Hopefully* I'll get back to work on the non-euclidean thing. Maybe. I have some ideas as to how to increase my hype about the project, but at the moment, with midterms and everything happening, it just doesn't seem appealing right now :(
Aight, well, ramble to you guys next week.
Thanks for reading!