Sooo… this ended up being a huge wall of text, I tried cutting it up into neat paragraphs.
No LLMs were subjected to the cruel & unusual punishment of writing this.
Assets interoperability:
I don’t see the metaverse/shared-assets stuff being anymore relevant now than metaverse/nft stuff was 10 years ago.
I suspect that AAA/AA games will never want to use UE6’s future interoperability. It’s too long a shot for narrow-minded executives.. And even if they were not, the fantasy of having your group of friends easily migrate would be a zero-sum game, since you’d make one UE6 game lose players to have another UE6 game win players.
However, it is possible that indies may end up using the feature more than the big studios, by creating small ecosystems made up of opportunistic pairings done after the fact, much like we see more and more devs creating unexpected Steam bundles between games.
AI Integration:
Fancy word completion black boxes devices (so called "AI"s) could not possibly create meaningful interactive art.. Much like the aviation vs airships of old, LLMs (being the airship in this metaphor) are a dead end of diminishing returns and increasing costs. Meanwhile, circular investor money won’t be able to keep subsidizing token price forever. LLMs will likely become unafforable, or consirably dumber.. Probably both, unless maybe governments recklessly subidize them to prevent the economy from collapsing.
The results of the UE5.8 demo were very mid, but predictable, since the model cannot do spatial reasoning. If there was any “wow” factor to the demo, it wasn’t the result of the MSP server itself, but rather pre-existing features like PCG shining once again (as they should).
But there is hope for UE6 since you people seem to now focus on integration with less visual systems (like state trees, etc) that will probably benefit from having “AI” integration as the form of an assistant, rather than a cheap (and blind!) level designer.
Concessions:
I am forced to recognize that there is at least a clear vision and the required investments to realize that vision, so I’m willing to let Epic cook before judging the aforementionned new features.
I understand the appeal of doing away with old systems in order to ultimately change things for the better. And I appreciate the seriousness of Epic’s commitment by dedicating something like 3 years (18 months to early access + 12 to 18 months to release) of focused work to realize the vision that is UE6.
The Lore VCS is an excellent idea (including its open-sourcing), and coupled with sufficient focus on engine multithreading, these two “low level” additions could very well be significant enough on their own to warrant a version number change.
(Even though marketing would have a tough time selling it, just like you seem to be having trouble talking about IRIS because it does not easily translate into screenshots)
Deprecation of Blueprints (visual scripting):
That is all for “what is bieng added”… Let’s talk about “what is being removed”.
Never could have I guessed how tone-deaf Epic would be here. I understand that engineers that don’t rely on visual scripting, that perhaps don’t even do their UI in anything other than Slate (like myself btw), would accurately value the Blueprints Visual Scripting system.
However, I cannot believe that, after all this dogfooding with Fortnite, that there is any lack of feedback on how visual scripting has its use & place inside a game dev team.
I cannot believe that with all these UE evangelists everywhere, there was nobody to tell you how critical the mere existence of a Visual Scripting system is to attract hobbyists and small-sized teams of varying degrees of experience.
I cannot believe that after using visual scripting in many aspects of the engine (materials, animations, actual AI) with great success, that Epic’s conclusion would be to.. stop doing that.
Deprecation of the Actors system
As I said earlier, sometimes it is necessary to do away with old roadblocks in order to meaningfully improve things. I believe that most people will ultimately reach the same conclusion with regards to the deprecations of actors.
However, if the UE5 lessons are anything to go by, it is likely that this entire Scene Graph thing will take many UE6 versions to actually be production ready. If that’s the case, then a more reassuring, and a more honest approach would have been to proclaim the deprecation of actors for the end of UE6. Just like Nanite used to be a performance hogger with many missing features, and Lumen used to be a noisy, temporally reprojected mess, in UE 5.0, but have since then improved considerably.
Unfinished business
The lack of a 5.9 version (so far) is.. troubling. There are so many features still experimental yet completely part of Epic’s official communication.
If you look at Unity, there is ample empirical evidence demonstrating how this kind of ADHD approach to feature development can become extremely toxic to game dev’s confidence in your product. I urge you to reconsider the lack of a UE 5.9 version.
To whoever reads this wall of text: I salute you.
