| 3 anecdotes: - For Talos Principle 2, Croteam switched from their own engine to UE5. The description was "It would be like attempting to sprint and catch up with a train that is already far down the track and accelerating even faster.". From a user's perspective, observe the graphics of Serious Sam Siberian Mayhem and Talos Principle 2: Same company, released at a similar time. Talos Principle (UE5) looks dramatically better than SS. (In-house engine) - Jon Blow rolls his own engines (and lang), and releases games very slowly - Expedition 33 recently released as a phenomenal game that leaned heavily on features in UE5 (Face/body, graphics, map/terrain gen etc.) They focused on the game itself, and let the engine do the heavy lifting... to a superb result. From my own anecdotes in graphics programming: Producing a simple engine is easy. Producing something that's photorealistic etc is massively more difficult. Let along all the other things an engine provides for you. Modern games have so many complexities the engine abstracts out; we don't need to roll a new engine for each game or studio, each trying to have optimized netcode, human characters, photorealistic lighting, GUI map editors + terrain gen etc. I use my own graphics engine for my scientific programs, but it has much simpler requirements than a game engine. |
Very different world from 20-30 years ago where the initial "game design" was happening in paint programs or even graph paper notebooks.