The enthusiast showed a realistic city of Solityud from Skyrim on the Unreal Engine 5 engine



Leo Torres shared a new video showing the reconstruction of the capital The Elder Scrolls V: Skyrim, Soliad, in the Unreal Engine 5 engine. Detailing the city is sixteen times higher than in the original Skyrim, which gives us an idea of ​​how a classic RPG remake from Bethesda in UE5 may look like.

But the most interesting here is that the artist created Soliyud in a more realistic and accurate scale from the point of view of history. Compared to the original version, Soliad now looks like a real place. Moreover, in this video there is a cinematic mode and the passage mode.

If we talk about technical details, then the video was created on Unreal Engine 5.3.1 on Nvidia GeForce 3060ti. At the same time, the artist used Nanite for all models and foliage, as well as lumen for global lighting and reflections. Thus, the level of detail is exceptional.

What I liked to just wander in this game. There have been no adventures in the games for a long time.

Yes, yes, at least install right now and shoot again)

But I would update the computer – I could roam in the Starfield 🙂

I updated the computer, and still more Skyrim I want than in Starfield
*The latter played, does not add. A week later abandoned

Yes, the same thing. Skyrim I have hundreds of hours, Starfield – 70 played, built, climbed and abandoned. Not clinging. Waiting for mods and dls.

Starfield tedious cloak. What worked at Skyrim at the same level does not work at Starfield in terms of adventure, because you need to apply much more forces and fantasies.

"What gives us an idea of ​​how a classic RPG RPG remake from Bethesda in UE5 may look like."
We already see how nectogen from the gazebo may look, hear. Buy 4090!
The enthusiast is well done, but after the breeze in the form of Starfield hope for an adequate continuation of TES no longer. Everything was degrand.

Yes, with a chic count, but with wretched wooden animations that they carefully store from the time of Oblivion .

And with optimization that came out of the chat. About the unique open gameplay is better to be silent or taldy "This is a typical game from the gazebo"

Killed optimization for the sake of new -fangled chips of graphony, I can somehow understand . And you can at least set any average quality settings, we are not on boxes nailed with nails.

but for decades to insert at all of their releases wretched, wooden animations . which you admire from the first and until the last second of the game – bloom.

So, how should Igor look at 23

Why then looks like this?

I guess I will not live until those days when the schoolchildren will fake on such a schedule)))

It looks of course interesting, but UE5 has one nuance – for a good picture, he needs top -end iron (what game on UE5 do not look at, all the same thing, and I don’t need the hands of the developers here – on UE4 and large companies and indie developers made games, and such that it was not happening now)

Unreal Engine 5 is great for architectural presentations, there is nothing new. However, the engine sharply slows down, when trying to add a large number of moving objects. And this is also not news. In the 6th version, they should eliminate the problem, then perhaps in the cities someone will appear.

In which version 6?Another 8 years to wait ?

any engine will slow down. However, I just think it is possible and the anreal is to immerse shorter the man who will be particles with the peaks of animations completely without the participation of the processor. Only tracks and central tochi will be. And then the engine will pull these particles through the processor and spell the processor NSP instead when approaching the camera.
or partially you have to transfer this case to matrix evictions of a video card is a matter of implementation.
I would probably if I had a cool programmer, I would have static in the NPS memory and objects for them would be some kind of radius where they can appear and quasi of a random algorithm for where they can actually appear on the basis of a generation of generation that is not moving in time so that It was like in Cyberbag 2077 turned away, and there the other gopniks were drowned out and passers -by disappeared. But there are still some statistical curves of the situation that the type of mining on the GPU was tied with certain points of the addresses of the type of type a person goes to the store home and to work and for this the necessary wave Fourier transformation. And for cities it is still not bad that they have some kind of simplified economic model and perhaps they will need more stationary NPS for optimization.
I have such a game of such a game for 2 years.

Will only slow down what the resources lack. The idea of ​​adding objects determined by the direction of the camera did not justify itself. Because the weak systems did not even have time to draw grass, but it is not necessary powerful. The idea to create a simplified grid with the addition of plugins along the way is also far from perfection. CP. will focus on priority details forgetting about the environment. All have pros and cons of. You need a grid repeating all bends and details, but for this it needs to be divided into more segments. 5 promised to implement this, but it was not yet possible. Now there is already static technology, NPS and details. Such will be simply part of the interior until you interact with them. When interacting, the object becomes invisible, and its place is occupied by NPS, or subject. The Creation Engine 2 engine was capable of this, but for some reason this opportunity was not implemented. Nevertheless, enthusiasts (will not poke their fingers 🙂 already work in this direction. The project is poorly supported, so another 10 years, if there is no miracle.

Unreal Engine is constantly updated, now Unreal Engine 5 is already relevant.3. It was also with Unreal Engine 4. So. which is possible, much earlier.

The idea of ​​adding objects determined by the direction of the camera did not justify itself. Because the weak systems did not even have time to draw grass, but it is not necessary powerful.

If we are not talking about, then UE4-5 calculates the trace of the sphere and at the same time he calculated the objects at the same time, and the basis of the size of the screen relative to the screen. I faced the problem of transparent walls for a short time on UE4 The problem of the missing objects was solved by an increase in the sphere (in particular, Men occupied the possibility of spiting trees, they turned to the camera on the GPU and the shadow turned from Sonets for one approach without a processor and 4 landfills and along the Krum The screen was the problem that these things disappeared) I also experimented with the normal 3D of my opinion 1.5 million trees in the frame or how many at a 30 km range of rendering everything worked fine with shades of 10 km (all the backs with shadows are simply limiting the lifting of the shadow itself to the depth of the frame) without delayed shadows from UE5. To the heap, I still have on all the boats or only on the neighbors there was a vertical displacement like the wind. The problems were there mainly with large grids because the FPS fell on a large dense grid almost to zero and there were problems with the performance of shadows on large grids the way I decided to go about the simple path and simply filled the world with the geometry on which Dollen had 3D fog and clouds With shadows. And in principle, this is all optimized.
As for the grass, then, as it were, you need to maximize the transparent materials simply. With trees, not everything is so simply because it is still needed intermediate boats with large textures and with Tava I think solved. I also tried to test how much you can draw landfills. AMD FH6100 could draw approximately 60 million landfills in a frame with shadows before the FPS descend below 25. Modern computers are 4 times more powerful in processors and probably there can be optimized there in terms of shaders due. The only thing I think I can have a problem because they should be worse than potentially worse should cope with a large number of small operations where there will be animation of grass, but they seem to have asynchronous shaders and the discharge of calculations for large numbers. Although Anrial has 4-5 old core of shaders, it is not the fact that it deeply uses DH12 ultimat. So productivity in the GPU does not steal much to me much. 30 FPS on the people’s 3060 in 1080r on I think in an extreme case in 720r.
But the performance of the logic itself in the processor and the logic of navigation of physics and ITP on blueprints is at least very strongly scared. The inventory slowed down. And what does I do not quite understand. here we need the power of I9 14900K 6GHz.
and with water actually hellish torment. Very complicated. and from the point of view of the shadeers garbage. There is simply the waves of Gershld or a couple there like that plus mixing on the height map in the world space is needed some kind of script that says the object in the water or not because it is not so important anyway not to get used to the Dali and it is not important there is a meter above it or it Below the rotation function is very complicated.

At 5m, Arile saw that they returned the tesselia of the landscape through Nanits. I think nothing more is needed. It was also good such a gider to make a mixture of nanits somehow write down, generate a CC collision grid and for the tesselia so that the legs do not fall out and that the objects do not go through the lights in the snow.

So does not seem to be in terms of optimization with something like this, but you need to conduct a lot of research. And in the year I moved to the year 4, then I realized that I arrived but I have not been engaged in normal graphics recently.

The engine can.
but from the point of view of game design – why?
Big cities in such a game would be tiring. they were made toy so that everything was not so tedious.
One metropolitan city with dozens and centuries of thousands of inhabitants would be like the whole Skyrim size and is generally inconclusive, by and large – the distances will kill the fan.
After all, the player needs a diverse world with a bunch of all kinds of landscapes and points of interest, and not one worked out anthill. And this means that the scale should grow for the entire game world. but traveling through realistic geography is not so fun.

So the city can be made large, and only part of it can be visited, and the scale will not be running around 30 minutes through the streets.

Yeah, tell me the gates of Baldur from 3 act. Tiring them to run on them

With large cities of medieval type there is such a problem. I somehow closed and tested the city a little more than a kilometer all exhausted and rocked me to the same time while I came through fortification barriers to the palace. Perhaps because there was no texture. But the big worlds like the norms of 10-30 km to run away with a horse of 10-20 km per hour on a relaxation is not a problem.

The most flawed way to implement in games
In some cases, you can understand closed houses in games, but in Skyrim is such a game design that you see the door you can go and see what you can mush.
So like Dragon Age. Small locations and 3 measures are simple.