The next step change will have to be some form of advanced kinematic rigging that creates advanced motion granularity akin to the graphical fidelity of nanite graphical tech.
UE5 is a step in that direction, but there is still lots of work to be had.
What will really be a game changer is AI that can generate the bulk structure having been trained on our best of the best hand built models and we can iterate on the emotional details.
I’m working on the “bulk structure” part, training models to generate random game worlds that roughly adhere to the rules, look and feel like everyone favorites right now (while avoiding copyright issues).
After that my goal is empowering consumers directly to nudge the styles in their preferred direction.
I’m mostly motivated by the MBA-ifying of everything. My goal now is to just have AI produce new content for me, even if such a thing puts game developers out of a job. I’m starting to experiment with cartoons as well. Optimizing for myself like we do.
I still want to comment because your comment doesn't do that demo justice.
It's the first time I expierienced that awkward feeling looking into reality while watching the video of that city.
We are Eye opening close to cross over uncanny valley after reaching it that soon.
There is zero chest wall movement. If you have a friend who works in medicine — especially emergency med or intensive care - send them this link and ask them how natural this looks. Bet most of them will immediately notice the same issue.
Unity isn't exactly perfect either, there's just confusion about which version to start out on but the one that has the most amount of tutorials and userbase seems to be the answer.
Absolutely correct that UE C++ is daunting. You just have way too much responsibility and you absolutely need experience with C++. It also takes more developers who are harder to find compared to Unity devs.
Unreal Engine really isn't it for indie or small studios. It just takes so much longer to make something on it, and you almost certainly end up working with C++ to fix performance issues, debugging, etc.
For large studios especially film studios using it to create 3D environments? It's perfect and those are UE's target market since they are guaranteed to have revenue income that can pay UE since it works on percentage of revenue generated and small studios, indie devs, the risk is far greater.
This symmetric financial incentives mean the indie, small studios are always sidelined as they don't pay the bills. That's where Unity really shines.
Is there any good engine made for programmers? A kind of "bootstrap" of video games?
After seeing how far unreal had come and downloading it to test it out I finished up the consulting project I was on I started looking into making VR games.
Initially I messed around with some other projects too but for maybe the last 5 or 6 months I've been working on pretty much just this.
It was a lot to learn but once you understand the core concepts it does not seem that different to me that normal software development.
I'm currently learning how to use c++ instead of blueprints. Each thing I learn makes the engine seem easier and like any other piece of software. There is some cruft but less than you might expect.
I watched a bunch of courses and videos but looking back I think the best place to start would have been the docs. For unreal and blender. They took me a while to work through and I had to stop and Google terms constantly but they were really helpful for understanding some of the basic mental models and terminology in the space.
Please correct me if I'm wrong, but isn't this basically a software-based implementation of NVIDIA's hardware-based RTX?
This documentation page covers some of the differences between Lumen's software and hardware raytracing modes: https://docs.unrealengine.com/5.0/en-US/lumen-technical-deta...
And if you want a ridiculous amount of technical detail about how Lumen works, check out this talk: https://www.youtube.com/watch?v=QdV_e-U7_pQ
That was never the case, OptiX was never intended for use in games. VK_NV_ray_tracing and DXR 1.0 were pretty much there since launch for the former and quite soon afterwards for the other.
Unreal is really starting to actually be the best of the best
"Lumen implements efficient Software Ray Tracing, allowing for global illumination and reflections to run on a wide range of video cards, while supporting Hardware Ray Tracing for high-end visuals."
And the Lumen / Unity demo was on a PS5, which is using (relatively low-end) AMD hardware.
So it isn't particularly logical to think you'd need Nvidia-specific ray-tracing for the feature.
UE4 could target WebGL but for UE5 I'm just finding a few hobby/fanboy forum posts not anything official from Unreal.
WebGL 2.0 is stuck in 2011 hardware, something like Unreal is overkill.
No wonder they decided to use pixel streaming with server side rendering instead.
We're working on a WebGPU backend right now for both UE4 and UE5, and have already upgraded UE4 to support WebGL 2.0 from 4.24-4.27, as support for HTML5 was removed back in the 4.23 release.
Another major innovation is we've imported a more improved compression format (Basis) as part of our pipeline, we've also created an asynchronous set fetching that only grabs the required data that a user needs to see at any given moment, streaming the rest of the assets in the background as necessary. This dramatically reduces load times, which was one of the biggest complaints with Unreal on the web previously.
Website for more info - https://theimmersiveweb.com/
Our Discord: https://discord.gg/cFJV6Yu
The biggest limitation is that Nanite currently only supports rigid, non-deformable meshes, so it's mainly intended for scenery rather than characters.
https://www.youtube.com/watch?v=eviSykqSUUw
Some highlights:
Seam staggering between levels for smooth swaps
Current hardware rasterizers chug on small polys so they made a GPU software rasterizer (!)
Actually it's a hybrid rasterizer, and gnarly features like derivatives are supported on both sides
The GPU scheduler chugged so they abused Z-test wave repacking
Streaming compressionUnity does have some nice ECS things going, but overall their tech isn't good(prime reason why is that they don't make games, they make engine).
Main branch of CryEngine is dead as engine that anyone aside from CryTek is using, but Amazon invests a lot into their fork(but it lacks vision TBH).
Godot is good for 2d, but they aren't really going for that kind of experience.
From Wikipedia: Tencent acquired a 48.4% outstanding stake, equating to 40% of total Epic, in the company in 2012.
Although I’m sure there could be a better display of the technology.
- deployment to multiple platform with ease
- easy scripting to make live services easier to iterate
- easy assets cooking, so you don't have to spend too much time to make different kind of assets for different platforms
- collaboration, so your team can work on the same project with ease
- then all the other systems audio/input/networking/terrain stuff etc
Fortnite is available on every platforms in the market (minus the web), their ability to add content and update it quickly, across all of their supported platforms at the same time is a great showcase
Just like Genshin Impact is a great showcase for Unity
So as someone who has never touched game dev, what is the learning curve like - exponential? What is "basic competence" once past "hello world"?
Thanks :-)
Extend a character class, give it a skeleton, an animation blueprint with some animation states that read the character state, hook it up to player inputs, make it do some sounds, and add a bit of flair to some actions it can take?
Can you build a basic ai behavior tree? Run some line traces to interact with colliders so your guy can interact with an object?
Shoot a gun and spawn some bullets that can damage another unit?
Etc.
The first thing I'll tell you is that the tooling for building enterprise apps is woefully inadequate out of the box. The things that come easiest to you, specifically CRUD app building, are much more difficult in a game engine. There are no native UI elements, no theming in the way we've come to expect, and the workflows are all completely different. I was used to building database connected backends, and switching to a state machine model really threw me for a loop. If you have experience building native apps, you're probably going to have a much better time than I did.
As far as learning the interface, that's not so bad. Follow some of the tutorials and you'll be able to get up and running in no time. The rabbit hole for me was understanding WHY things were doing what they were doing, as opposed to figuring out how to do it. Except in those times which I COULDN'T do something natively that I was expecting to be able to do...which happens a lot.
My recommendation is to set up the game engine of your choice, learn how to build to multiple targets, and then craft something you've already done elsewhere to see how you like it. Once you get that going, you should have a cursory understanding that will allow you to flex some creative muscle.
Good luck!
I learned the hard way and realized just how hard game development is and how much we take for granted the human cost. You really really must love games to be involved with them. You cannot do it based on ROI calculations it really requires passion and patience.
While uninstalling games to free up space to download the project, I noticed that there was no correlation between a game's disk space usage and the amount of fun it provided.
My projects use Unity and any time the camera gets 1800m from the world (0,0,0) the camera begins to shake. By 2000m the shake is too violent to use.
So you have to reset the coordinates by a hack if you want a world scale 1:1 map
Using 32-bit floating point you'll have 0.1 accuracy up to 100000 at least.
Edit - I just realized 10cm accuracy is not great (and that I'm talking to myself) so I tested with 0.001 and that's stable up to 9000.
Even as an Unreal "expert", for personal projects I still use Unity, it is just much easier to get stuff up and running and iterate.
Unreal has a steeper initial learning curve but the system itself is more pleasant and the built in components are much more helpful. I switched to unreal after a couple years of Unity and strongly advocate others to do the same.
Also, there's now an "Import Xcode GPU Debugger Plugin for Mac".
[1] https://docs.unrealengine.com/5.0/en-US/unreal-engine-5_0-re...
What's happening?
On the windows desktop my mouse cursor movers definitely at 120Hz rate on this TV.
The city from the Matrix demo is available for download. I want to try that.
Now building module 524 of 3998...
Inevitably, the Windows build is a download via a provided launcher, and the Linux build is getting the sources from Github, searching forums for the directions, and compiling for hours. Supposedly Nanite will work in Linux in the release version; it didn't in the pre-release.
Good luck of that ever happening with Unity. I was promised refund by Unity support for a totally non-responsive and outdated plugin that was on sale. They delayed and delayed until I could no longer get a refund.
Having said that I do not recommend Unreal Engine for solo indie development but Unity also feels weird with confusion around which version you need to start with, DOTs etc.
Unity seems like it is it for indie developers but its increasingly a cluttered landscape with lot of shoddy plugins
On most modding comunities they tell you to stay away from it.
Some usefull tools i found: https://github.com/panzi/u4pak This is used to unpack the game files. https://www.gildor.org/en/projects/umodel with this tool you can view some models and other assets in the game files.