Recently there was this tech demo from Game Awards showing the Unreal Engine 5 running a Matrix theme Demo and I was blown away watching that! It was insanely impressive. So I had to wonder about my favorite game and Its graphics... Sure you play Odyssey and the game is still astonishing but yet somehow looks inferior to SC its direct competitor that is yet to release. So I was wondering; just how great the difference between UE4 to UE5 is; which begs the question... Is it possible for Frontier's own Cobra engine be updated to have massive improvements in terms of animations, lighting and graphics at some point like this off the shelf engine can? It makes me wonder if the most wanted feature "ship interiors" and interiors in general are happening at a slow rate due to maybe the engine being maybe "outdated" by comparison with the current ones today. Would be nice to know if Cobra engine still have room to evolve.
Elite Dangerous Cobra Engine Updatable?
- Thread starter CMDR Yami Sukehiro
- Start date
Yes.
No.
No.
Ask Frontier.
We probably won't ever know.
Generally speaking, everything in programming can evolve (this includes coding everything back from scratch), it just depends on how much time, pain and blood it'd require.
Often it's not worth it.
Also you can't compare a tech demo to a real game.
Obviously UE5 has better graphical fidelity than the Cobra Engine though. It's brand new and costs a fortune after all.
Obviously UE5 has better graphical fidelity than the Cobra Engine though. It's brand new and costs a fortune after all.
So you think after 33 years it's now not possible to change? I look forward to your TED talk on suddenly unchangeable software
The game engine FDev uses for it's games is proprietary, there are no licencing restrictions, if a new game requires a feature no version of the engine currently uses they can add it, or even design an entirely new version that will still be referred to externally (as far as marketing & customers are concerned) as 'the Cobra Engine'.
The upside is that freedom of control, the downside is there is no pool of experienced Cobra Engine Developers in the job market, any new dev will have to learn whatever quirks the in-house developed dev tools have, and if an experienced FDev developer leaves the company it is a more significant blow to the company than if they were using one of the more widely licenced out engines. Those years of experience with the Cobra engine are of limited appeal on the dev's CV too, compared to one with experience with a more widely used engine.
But what it is capable of is unrestricted except by the time to develop the new feature & integrate it.
The upside is that freedom of control, the downside is there is no pool of experienced Cobra Engine Developers in the job market, any new dev will have to learn whatever quirks the in-house developed dev tools have, and if an experienced FDev developer leaves the company it is a more significant blow to the company than if they were using one of the more widely licenced out engines. Those years of experience with the Cobra engine are of limited appeal on the dev's CV too, compared to one with experience with a more widely used engine.
But what it is capable of is unrestricted except by the time to develop the new feature & integrate it.
1. Cobra is always evolving, each update just that. Frontend game client updates you see, backend engine is being updated to support the frontend and infrastructure.
2. 'Realism' for lighthing & graphics is a very specific thing, but they evolved just that from Horizon to Odyssey
Finally, be skeptical of 'showcases' as they are slices that just show one tiny aspect. It does not need to provide 'a full game experience' and will always be a superior showcase.
When you add all the complexity of a 'full game', AI, events and the world - then you may have to make sacrifices in certain CPU & GPU intensive tasks to have room 'for the rest of the game'.
2. 'Realism' for lighthing & graphics is a very specific thing, but they evolved just that from Horizon to Odyssey
Finally, be skeptical of 'showcases' as they are slices that just show one tiny aspect. It does not need to provide 'a full game experience' and will always be a superior showcase.
When you add all the complexity of a 'full game', AI, events and the world - then you may have to make sacrifices in certain CPU & GPU intensive tasks to have room 'for the rest of the game'.
From my point of view, cobra engine is like a dinosaur that would be better left to die, it only causes problems and more problems.
By the way, I just watched an 8 years old tech demo of UE4 and it still looks better than most current games. If something looks good / realistic or not doesn't just depend on the capabilities of the engine, it also depends on the time artists spend on the scenes to add details and stuff.
Sadly no other engine would be capable of running Elite.
I saw Frontier employee comments about various work related stuff on some website and complains about Cobra engine were rather numerous. According to them it's outdated, unfriendly mess to use, making even simple tasks rather difficult if you're not experienced programmer, especially that many people who knew it best moved on.
A common trope in software dev is that any existing code is an outdated, unfriendly mess, and needs rewriting. If the management are daft enough to let this happen then when the next new people turn up in a year's time they then complain that the code is an outdated, unfriendly mess, and needs rewriting. And so it goes.
(Not saying the code is not an outdated, unfriendly mess, and needs rewriting, but really - that is 90% of code
Even when you look at the comments and find you wrote it yourself )
Last edited:
Glassdoor by any chance?
That really needs to be on the forum bingo card.
That really needs to be on the forum bingo card.
I wonder if anyone ever writes 'the code here is lovely, easy to change and a joy to work with. Sadly the team were monsters, so I left.'
ED's primary flaw isn't in the graphics quality (which remains passable for the style it's in), it's the depth of implementation. Mile wide, inch deep, with many things.
Visually though, the vertex count you can push through a modern card has increased significantly, so an asset overhaul is in order. I don't think the underlying APIs have changed that much (DirectX) so the engine itself would probably stand more or less unchanged*.
Bear in mind when making SC/ED comparisons - ED attempts (not always successfully) to maintain a stable (ish) build and ensure the follks who already own the game can still play but wants to paywall certain aspects. SC says 'everyone gets everything, but balls to stability and old hardware, let's go bleeding edge, with EVERYTHING, ALL THE TIME'. They've got similar 'end targets' in mind (Braben and Roberts' visions), but they have (well I'd say had, up to Odyssey's release) differing implementation strategies.
This means ED has a lot more legacy assets to maintain (and shows where a lot of SC's constant reiteration time goes - how many versions of the Constellation have been redone now?). Overhauling them to accomodate say, ship interiors, without completely breaking the existing asset for those that should not be able to see said interiors can be a lot of work, especially if you're using a datagram to generate a local asset. (This ship, this skin, this/these colour/s, etc, etc). How do you standardise that so multiple client versions (paywall levels) can get the data they need (and ideally not the data they don't)?
Anywho, for the OP's question. Yeah, course you could slap a new renderer in it. But there's more to the game than the front end.
*: It looks like COBRA currently generates occulsion volumes badly (if it's resolving from visual mesh representations this could be why) and/or requires manual adjustment. These are used for culling, and have certainly been or are a factor in some of the horrid performance related and graphical bugs we've seen. Fair disclaimer: I focussed on graphics coding in my degree, but that was over 20 years ago and I didn't spend too long in coding afterward. I'm rusty.
Visually though, the vertex count you can push through a modern card has increased significantly, so an asset overhaul is in order. I don't think the underlying APIs have changed that much (DirectX) so the engine itself would probably stand more or less unchanged*.
Bear in mind when making SC/ED comparisons - ED attempts (not always successfully) to maintain a stable (ish) build and ensure the follks who already own the game can still play but wants to paywall certain aspects. SC says 'everyone gets everything, but balls to stability and old hardware, let's go bleeding edge, with EVERYTHING, ALL THE TIME'. They've got similar 'end targets' in mind (Braben and Roberts' visions), but they have (well I'd say had, up to Odyssey's release) differing implementation strategies.
This means ED has a lot more legacy assets to maintain (and shows where a lot of SC's constant reiteration time goes - how many versions of the Constellation have been redone now?). Overhauling them to accomodate say, ship interiors, without completely breaking the existing asset for those that should not be able to see said interiors can be a lot of work, especially if you're using a datagram to generate a local asset. (This ship, this skin, this/these colour/s, etc, etc). How do you standardise that so multiple client versions (paywall levels) can get the data they need (and ideally not the data they don't)?
Anywho, for the OP's question. Yeah, course you could slap a new renderer in it. But there's more to the game than the front end.
*: It looks like COBRA currently generates occulsion volumes badly (if it's resolving from visual mesh representations this could be why) and/or requires manual adjustment. These are used for culling, and have certainly been or are a factor in some of the horrid performance related and graphical bugs we've seen. Fair disclaimer: I focussed on graphics coding in my degree, but that was over 20 years ago and I didn't spend too long in coding afterward. I'm rusty.
Uh what didn't you get?
I guess the occulsion volume bit?
Think of an occlusion volume (aka occlusion mesh) as what the game thinks of as 'solid'. It hides stuff that's behind it. This is usually a simpler model than what's drawn to make the math faster. That model is either generated automatically or built by hand and lined up (in the game world) to what's drawn. A generated mesh may end up with odd holes in it that stop it working properly leading to things getting culled that shouldn't be or things that should be culled getting drawn.
I guess the occulsion volume bit?
Think of an occlusion volume (aka occlusion mesh) as what the game thinks of as 'solid'. It hides stuff that's behind it. This is usually a simpler model than what's drawn to make the math faster. That model is either generated automatically or built by hand and lined up (in the game world) to what's drawn. A generated mesh may end up with odd holes in it that stop it working properly leading to things getting culled that shouldn't be or things that should be culled getting drawn.