[bitwize]

Second this. It should be drilled into the heads of everyone who wants to get into game development: USE A MAJOR ENGINE. No exceptions. Time is money, and hand-rolling an engine is a waste of yours. Unity and Unreal have many, many more man-hours of work put into them than anything you can build, so by choosing one of them you can avoid pitfalls you WILL run into starting from scratch, on top of the labor you'll save not having to build one in the first place. On top of all this, the entire gaming workforce is oriented around these two engines, so when it comes time to collaborate, you will be able to bring people aboard who can contribute immediately.

The GPGPU stuff is critically important if you are targeting low-level programming, to which I would add AI processors (NPUs, TPUs, etc.) And bone up on your statistics and linear algebra to like, the "Ph.D in math" level. The AI rocket is about to take off, big time; you want to be on it.

[jesse__]

Counter-point : the tradeoff you accept when using a commercial engine is that they're extremely general purpose, which can be unsuitable if you're trying to do something weird.

Look at Braid if you want a good example. Using Unity or UE to build that game would almost certainly require more work than writing a simple 2d platformer engine. Furthermore, the story of how it came to be is, in fact, inexorably linked with the act of writing the engine itself.

Sometimes, writing the engine is the right choice.

[bitwize]

Braid was released in 2008 -- forever ago in game-industry time. The industry has become more consolidated around Unity and Unreal since then.

Unity has 2D platformer support built right in, and the tricky bits -- like the time mechanic -- could be written as an extension. Heck, there's probably a time rewind mechanic in the Unity Asset Store as we speak. They've so far advanced that you're just never going to keep up writing a bespoke engine because they've benefited from being integral to the industry as a whole. Furthermore, the skills you develop on your own with your little bespoke engine aren't going to translate into the industry, where everybody uses Unity or Unreal.

It's kind of like how you're never going to build a graphic design business using GIMP, Krita, or Inkscape. They're all nice tools for amateurs and dilettantes, but entire industry standards and processes have coalesced around Photoshop and Illustrator, so that's what you use. The open-source alternatives are decades behind the curve here and in some ways, will never ever catch up. Good luck getting PANTONE support, which is critical to graphic design for print, into GIMP or Inkscape, for instance.

[reesericci]

but the sprig team just wants to have fun. its exciting for them to NOT use a major engine. It feels new and exciting to work on things that are new, and not go along with the crowd. Time might be money, but we're still teenagers after all.