I’m talking about an 3D engine, not a simple browser-like 2D engine.
So am I. My current project is three-dimensional.
Yes, I have never tried, why should I?
You have never tried, so your only view on the matter comes from other people, who also have probably never tried. In other words, hearsay. I have tried, and have direct experience in the area.
My experience shows, that the making the engine takes a very small amount of time compared to the making of the main game. Especially when the game is made alongside the engine, instead of separately.
But if you do it on a commercial basis, time is money and your game should need extra-ordinary functions the usual engines cannot provide. Don’t code stuff players don’t appreciate.
“Time is money” is neoliberalism and is thus an ideology, not fact. My ideology is different.
My choice of a custom engine very much brings with it observable results: my software works on old computers and I thus do not coerce anyone to spend hundreds of dollars on useless hardware. My software does not consume gigabytes of RAM to draw one triangle on the screen.
And I, as a player, do appreciate some effort put into a project, something the use of a custom engine very much shows. This alone proves that such an audience exists. They’re just quiet on average.
I am unlikely to play a game made with Unity or Unreal, and I am definitely not playing a game made with any AI, because it is by implication zero-effort and worthless, and is thus undeserving of my money. Despite not being aware of such an audience, you’ve pulled the number five out of your ass, thanks to your own neoliberal bias.
A custom engine does not mean 100% from scratch, anyway. I don’t know where this misconception comes from. I still use ODE as the physics engine, GLFW as the window library and SoLoud as the audio engine.