Except your point still isn't really that good because you, like several other people in the thread act like game developers have to check specific boxes, and are required by the laws of gaming to allocate as much resource as equally possibly to all these different places.
In other words, you think it's a REQUIREMENT that Bethesda games should have amazing mindblowing graphics, because it's a post-apoc game that makes millions.
Have you ever considered that MAYBE, they don't care about the graphics as much, because they know most of the fans don't either? Yes, it would be nice if Fallout had great graphics, but it didn't the last two times, and it doesn't now.
They added a full, dynamic extremely in-depth crafting and building system, where almost every single physical asset in the game can be picked up, rearranged, or given a purpose in gameplay. They've completely retooled the way combat and stats work. This game has 100s and 100s of NPCs, all with stats and items and behavior patterns that need to be accounted for. They literally spent over 2 years of doing the protagonist recording lines. This game has over 111,000 voiced lines, to ensure that the game delivers the same level of freedom of choice that Fallout fans have come to expect. More than Fallout 3 and Skyrim combined. They have gone above and beyond to ensure that this is an exciting, full game with a huge world.
None of those are good excuses. Beautiful open world games have already been achieved, and there's a lot of space between this game and something like The Witcher 3.
You should not view those things Cloyster mentioned as 'excuses' so much as 'design priorities'.
Fallout is keeping track of the movements of a world with a staggering amount of individual moving parts including
- each and every physical 3D physics item that each street, field, building, etc. in the game holds
- hundreds of NPCs with unique faces and schedules, covered in yet more physical items via their layered clothing/armor/weapons
- the individual rubble, homes/buildings, and other elements that can be torn down for supplies
and the game is rendering those things individually - every piece of loot, every chair or shelf or fridge and everything therein is its own 'thing'
You do realize there are performance tradeoffs right? Look up draw calls in relation to performance.
A Bethesda game has a massive amount of draw calls because almost every object is dynamic (meaning it is not baked into the environment thus saving draw calls).
Every single mesh that can be moved is a draw call and if they don't use texture atlasing then you can say that every single mesh is 2 draw calls (mesh and material). But wait then you have to add shadows which, depending on the method can add at least 1 more draw call.
So lets say they use texture atlasing for all the small objects. That still means that there are at least 2 draw calls per movable object. Remember that just about everything in the game can be moved (everything that isn't is likely batched into one big mesh to save on draw calls when possible).
Walk into a house in FO:NV and you might find 50 objects you can mess with. Thats at bare minimum 100 but more than likely closer to 200 draw calls, then you have to take into account the draw calls the house itself and all the non movable objects that are not batched. The UI itself is probable 3 or more draw calls.
Beyond that you also have the character and weapon (likely 10 draw calls) and any enemies which would be 4-10 draw calls each. Add into that the draw calls from outside the house which depending on LOD distance can balloon out into 2000+ easy if there is grass and trees.
Culling can help with some of the draw calls but it has its own performance cost.
Now with all of this we still haven't gotten to post processing, Textures sizes, AI, scripting, physics (Beth games have dynamic physics which eats the CPU like nothing), and the lighting engine.
Are there multiple lights in the house casting multiple shadows? Whelp if there are then you can go ahead and double your shadow draw calls.
Do you not see how a Beth style game balloons out into a performance nightmare?
How do other games look so good (Witcher 3)?
Well they use a hell of a lot of static assets that can be batched (combined in both mesh and texture) to vastly limit draw calls which allows them to have better "graphics". Ever notice how almost all loot in the witcher games is found in chests? Well, that is another way to limit draw calls, keeping dynamic objects off the map (lower draw calls, lower physics budget, ect).
People need to educate themselves on how games are made before they start frothing at the mouth over OMG bad graphics. Not every game has the same base, or even the same goals.
so the reason an open world game like Witcher 3 looks so good, relatively speaking (besides a solid and cohesive art style and a much more conducive type of setting to work with), is because its loot is 2D art with text descriptions, and not physical objects being tracked in the open world, and because when you walk into a house in Witcher 3, generally the individual scene decorations are going to be static grouped decorations as a single asset. For what a game like this is doing with its underlying systems, and the unique way it enables the devs to decorate scenes and contextualize every part of the physical world, I think the way the game looks is well within reason.
There is absolutely no reason they couldn't develop a more modern engine with the same capabilities. CDPR did it for RedEngine3, Epic did it in UE4, Frostbite is constantly improving, as is CryEngine.
This engine is seriously crufty. It is absolutely true that their entire object system is vital for modding and their renderer has been patched up for more modern graphics effects, but even their PBR pipeline is providing a seriously unimpressive showing.
RedEngine 3 is an iteration on the engine CDPR has been using and improving since The Witcher 1, and not to dismiss the tremendous work they put in toward achieving their open world, but the game doesn't look or play all that different from Witcher 2 outside of the open world mechanics and some combat rejiggering, so I'm not sure why you posit that as a huge game changing engine revision. And the other 3 engines you mentioned are game engines designed foremost for licensing or intercompany development, engines that particularly hinge on their ability to produce competitive and industry leading visuals.
I think that as long as this game runs considerably better and more consistently than previous iterations while looking the way it does (considering the things I mentioned in the above post), it will represent a reasonable advancement for the engine.