Former ‘Starfield’ Dev Says Bethesda Should Switch To Unreal Engine 5

[PortalShifter]

Should Bethesda switch to Unreal Engine?

Lol of course they should. Anyone making serious games in 2025 is using Unreal. Even Bethesda's own staff know this...

https://www.forbes.com/sites/paultassi/2024/12/27/former-starfield-dev-says-bethesda-should-switch-to-unreal-engine-5/

"At least one former Bethesda dev agrees, Nate Purkeypile, a Lead Artist who worked on Starfield, Fallout and Skyrim, and has now set up his own studio. In a recent interview with KIWI TALKZ, he explained why he thinks it’s time for Bethesda to move over to Unreal Engine 5 after all this time:

“I think it’s pretty telling that there are some studios with similar tech that have made that leap lately like CD Projekt Red…It’s not to say that it wouldn’t be a ton of work…sure, that’s completely true, it’s not like you’re just going to be able to start making the game, but when you have this large toolbase and large knowledgebase people could pull from, it’s easier to hire people and ramp up.”

“I’m of the opinion…that it’s easier to stay relevant and not always try to play catch-up. Like a lot of what was done on Starfield was trying to get all these rendering systems and animation systems up to snuff and it makes it hard to even actually make the game. The way I describe it to people is that you’re on a train, but also laying the train tracks at the same time…but it’s also on fire the entire time.”"

[momopovich]

Publicado originalmente por Varanus:

Publicado originalmente por momopovich:

https://store.steampowered.com/app/1807810/The_Axis_Unseen/

With a staggering all times players peak of 86

https://steamdb.info/app/1807810/charts/

I put in on my wishlist, as a looked interesting. Friend of mine bought it for me, for christmas. The lighting and art style are very off-putting. As is the music. Also the game gives you very little idea what you're suppose to be doing.

Judging by the screenshots it seems that the main thing you're supposed to be doing is shooting enemies with a bow. It's hard to tell if you can actually unsheath the bow.

I suspect the thought process for this game was "Skyrim is very succesful, so I will pick the most popular gameplay loop in Skyrim (archer style), market the game as "done by a Skyrim dev" and earn passive income".

Congrats on being one of the hundred(s ??) players who own the game. Looking forward to your analysis if you manage to get enough motivation for a more than 1 hour long session.

[/dev/random]

Something tells me it won't help Bethesda.

[momopovich]

Publicado originalmente por /dev/random:

Something tells me it won't help Bethesda.

I doubt Bethesda needs help from Nate Purkeypile. Quite the other way around.

[momopovich]

Interesting fact : Purkeypile tells negative reviewers that they play wrong :

https://steamcommunity.com/profiles/76561198373875348/recommended/1807810/#developer_response

[Varanus]

Publicado originalmente por momopovich:

Judging by the screenshots it seems that the main thing you're supposed to be doing is shooting enemies with a bow. It's hard to tell if you can actually unsheath the bow.

I suspect the thought process for this game was "Skyrim is very succesful, so I will pick the most popular gameplay loop in Skyrim (archer style), market the game as "done by a Skyrim dev" and earn passive income".

Congrats on being one of the hundred(s ??) players who own the game. Looking forward to your analysis if you manage to get enough motivation for a more than 1 hour long session.

Well after almost an hour of playing it, I dont really have anything good to say about it. Shot some stuff. Stabbed some stuff. Neither felt really good. Intro was horrible. I have to play it more, to unlock the different arrow types and magic, to really get a full feel of the mechanics. But what I've played so far doesnt really make one eager to do that.

[Low Standards]

This would have been a lively discussion if everybody wasn't blocking anymore. Sad to see this at this moment.

[Ihateeverybody]

I have the ability to Block with my mind, When the internet finally burns down I will be in a better position than the people who sold their souls to the Machines.

Sorry, never mind that. Everything is going to be fine. The power will never go off. So don't you worry about transferring the sum total of human knowledge, civilization and culture to it.

It was fun while it lasted.

[Defective Dopamine Pez Dispenser]

To expand a little on my earlier post since I'm caffeinated now...

Engines aren't magic or monolithic in their implementation or use, and a lot of this will of course vary by developer, game, and system, but my experience with UE5 games thus far is a combination of:

• Enjoyable games, mechanically and in terms of design (Robocop, SH2, BMW, S.T.A.L.K.E.R. 2 - all games I've enjoyed playing)
  
• Visually impressive, to be sure. However ...
  
• Incredibly distracting temporal ghosting - quite literally the worst I've ever seen, and seems to be an issue across games thus far irrespective of settings (FXAA, TAA, TSR, DLAA, DLSS, XeSS, FSR, on/off, doesn't matter. Can be reduced but not eliminated, especially anywhere with poor temporal coherence such as movement between light and dark areas, and especially small moving objects such as leaves or trash blowing in the wind, and it. Is. Maddening. Frankly. Totally undermines the visual beauty of everything else in these games for me.)
  
• Significant frame time issues
  
• That odd deltatime issue DF noticed in SH2 that is apparently an issue devs are aware of in the engine and struggling to fully eliminate
  
• Lengthy shader warmup processes, and some stability issues associated with that as well
  
• Shader compilation stutter despite said shader warmup the first time through games. (As opposed to, say, Veilguard, which uses Frostbite, and has a shader compilation that while annoyingly long depending on your CPU upon first launch, avoids shader comp stutter entirely essentially once in-game as a result.)
  
• Significant performance impacts when even modest numbers of NPCs relative to other games initially spawn or complex pathing has to be handled.
  
• None of the physics-enabled, fully manipulable and interactive, object permanence the Creation Engine is known for.

Does this mean a TES game with the features we commonly associate with them can't be made in UE5? Of course not. With enough time, effort, and proprietary implementation and branching, anything is possible. As usual. But I honestly have a tough time imagining them being able to do so anytime soon, given what it would require for them to migrate their whole tech stack, and given that they've said this iteration of Creation Engine was a substantial time investment and basically tailor made for the next TES game.

Which makes perfect sense. As we've discussed before on this forum, many of the issues people have with Starfield are owing to its scale, something that simply won't apply to TES 6 in all likelihood. The next TES, presumably at least, will be back to just a single province of Tamriel, on a single contiguous map, like every TES since Morrowind. And given what I see in Starfield it seems like it's more than up to that task, when not having to deal with hundreds of Skyrim-sized maps spread across planets, multiplied over 1,000 planets.

Heck, we even see the ability to seamlessly enter and exit large cities like New Atlantis, instead of having to go through a "city gate" ala Skyrim. We haven't had that in a TES game since Morrowind, lest people forget.

So yeah, I don't really see much compelling need for them to shift to a new engine for the next TES. After that? Who knows. Maybe. Maybe by then UE6 or 7 or whatever will have all of these capabilities trivially baked in too in a completely interchangeable way with what the tools they already have do, and the aforementioned issues (to the extent they're even engine specific at all, which I don't know) will be sorted. Or maybe I/O will be so wide and fast and we'll have so much compute and render by then it won't matter anyway. 🤷🏻‍♂️

In the mean time, all I know is, while UE5 thus far has shown me some fun, beautiful games, it's certainly no magic bullet. You're not somehow going to magically get a better game just because of an engine migration.

And I'm not entirely sure we've yet seen anyone do with it what I would like to see in a TES. Take Avowed. Avowed is a game I'm hotly anticipating and hoping turns out great, and probably the closest comparison to something remotely like a TES we've seen to date in UE5, and... structurally and in other ways, it's still not really comparable.

I see no evidence of anything like Radiant AI or other dynamic NPC scripting/scheduling. I see no evidence of the ability to pick up and place physics-enabled objects (as many as we can within memory essentially, as we can in Creation Engine.) Heck, it isn't even clear to me that it's a single open world. And they've said it's more comparable in structure and openness to The Outer Worlds ("open zone" as they put it) than to Skyrim.

That would probably leave S.T.A.L.K.E.R. 2 (for now) as the largest scale open world game I've seen created in EU5 thus far. And I love that game, but it's obviously by design a very different experience. It's also the game with the most complex NPC behaviors I've seen thus far in EU5 and so far they consist of patrolling and some contextual behaviors in response to weather and the like.

Which is more than sufficient for what I wanted from that game, mind you. I was never hung up on it needing to replicate SoC's AI system fully. But it's not exactly something that inspires confidence that something like Oblivion's or Skyrim's Radiant AI could be done in EU5 in a large open world RPG. Especially given that, again, when you approach areas in HoC where even modest NPC groups spawn in, there can be significant performance hits.

I'm sure it could be done given enough time and effort... I just haven't seen it done yet, is all I'm saying. Meanwhile the Creation Engine is right there and ready to be used with that capability built in already, and specifically intended to handle it (even if Starfield doesn't leverage it much if at all, owing to its scale and different design vs a TES game.)

Engines can be made to do anything with enough time and effort. The question is whether it's best fit for task in this particular instance when something else that already is, already exists.

Edited for clarity.

[Rizilliant]

Publicado originalmente por PortalShifter:

Should Bethesda switch to Unreal Engine?

Lol of course they should. Anyone making serious games in 2025 is using Unreal. Even Bethesda's own staff know this...

https://www.forbes.com/sites/paultassi/2024/12/27/former-starfield-dev-says-bethesda-should-switch-to-unreal-engine-5/

"At least one former Bethesda dev agrees, Nate Purkeypile, a Lead Artist who worked on Starfield, Fallout and Skyrim, and has now set up his own studio. In a recent interview with KIWI TALKZ, he explained why he thinks it’s time for Bethesda to move over to Unreal Engine 5 after all this time:

“I think it’s pretty telling that there are some studios with similar tech that have made that leap lately like CD Projekt Red…It’s not to say that it wouldn’t be a ton of work…sure, that’s completely true, it’s not like you’re just going to be able to start making the game, but when you have this large toolbase and large knowledgebase people could pull from, it’s easier to hire people and ramp up.”

“I’m of the opinion…that it’s easier to stay relevant and not always try to play catch-up. Like a lot of what was done on Starfield was trying to get all these rendering systems and animation systems up to snuff and it makes it hard to even actually make the game. The way I describe it to people is that you’re on a train, but also laying the train tracks at the same time…but it’s also on fire the entire time.”"

Theyve packed the industry with unskilled diversity hires, who dont know how to work on anything besides Unreal Engine.. Its pathetic, and will only worsen the industry... Everything is now on Unreal Engine

[WildStargazer]

Publicado originalmente por RL 101:

It would be a great idea if you want to reduce interaction and have all the objects that appear on screen stuck on shelves and walls as if it were a drawing. With Unreal Engine the whole stage is actually an empty hallway with a very nice visual appearance.

Interactable objects are nice to have, but if this is the only benefit Creation Engine has it is not much. With custom game engines devs have to do double amount of work, first they have to update engine, then create a game, and they need to teach any new dev they hire to work with their engine. Unreal Engine has it's own problems. in particular high demands to hardware, but at least you can start making game from the start and there's a large workforce available for hire that is already familiar with Unreal.

[Spades]

Its incredibly gross seeing people trash talk UE5 on here without any experience on either engine.

creation engine is OK. It does what it does OK, it is not nearly as robust as UE5 and will perform worse in pretty much every aspect.

This does not mean CE is bad, its just that UE has a large, full-time dedicated team. CE is a single studio, dividing their time between titles and engine, and most public developers/publishers HATE spending time on engines, because that is a pure cost metric in quarterly/yearly's.

Suggesting that UE5 would do modding/performance/physics, or anything else for that matter, worse, is complete BS. Plenty of studios cut corners on titles, blaming UE5 for those missing corners is ridiculous.

UE5 is FAR superior to CE, obviously. The biggest issue with a switch would be that the studio itself, and Beth community, is used to working with CE, and would have to start from near-zero.

[Wade]

It's not just the engine that is the problem, it's literally EVERYTHING.

[Spades]

Publicado originalmente por Zero McDol:

UE5 is definitely not a favorite for me.

Plus, I haven't seen much of a modding scene for Unreal Engine games, so it would be a pass for me. UE5 visuals certainly aren't enough to justify giving up the current freedom that a moddable engine gives. One of the reasons that Skyrim is still as popular is it is in 2024 is, I suspect, the fact that it has one of the largest modding scenes in gaming.

On top of which, the Unreal Engine 5 visuals aren't even that much better than current gaming engines anyways.

This is a silly comment, full of silliness.

Visuals are mostly on the developer, not the engine, apart from aspects that can be "drag and dropped" into a game title... like lighting, UE has plenty of lighting options that you can just throw at your title to make it look pretty..... detail of visuals is up to you as a developer, not the engine.

as for modding, 600+ games are made on UE..... you think none of those have a modding scene? ARK is on UE by the way.

[Spades]

People all talkin like the engine makes the game, and not the developer. An engine will just make it harder or easier to accomplish a given task. You COULD make life-like, hyper detailed, well animated NPCs in CE or UE5.... the difference being how difficult and time consuming it is in each.

[Zero McDol]

Publicado originalmente por Spades:

Publicado originalmente por Zero McDol:

UE5 is definitely not a favorite for me.

Plus, I haven't seen much of a modding scene for Unreal Engine games, so it would be a pass for me. UE5 visuals certainly aren't enough to justify giving up the current freedom that a moddable engine gives. One of the reasons that Skyrim is still as popular is it is in 2024 is, I suspect, the fact that it has one of the largest modding scenes in gaming.

On top of which, the Unreal Engine 5 visuals aren't even that much better than current gaming engines anyways.

This is a silly comment, full of silliness.

Visuals are mostly on the developer, not the engine, apart from aspects that can be "drag and dropped" into a game title... like lighting, UE has plenty of lighting options that you can just throw at your title to make it look pretty..... detail of visuals is up to you as a developer, not the engine.

as for modding, 600+ games are made on UE..... you think none of those have a modding scene? ARK is on UE by the way.

Well, it's just my opinion based on what I've seen, plus I've seen the Unreal Engine 5 get tons of praise for its visuals, so that was why I commented what I did.

I never played ARK, so I can't comment on how relevant that that game is to the modding scene, but I'll take your word for it. Beyond that, none of the Unreal Engine games that I've played offer a huge modding scene, so it's hard to me to see the modding scene that you're saying that the unreal engine has. Whereas, I've seen Skyrim's and Cyberpunk 2077 modding scene.

edit:
I'll delete my comment though, as it's not factual.