Unity Roadmap for 2023 - GDC Talk

Video is up on youtube! (No idea why it is unlisted)

What are your thoughts?

 

looks like 90% stuff that has already been shown at length over last year or so - which to me is a good sign as maybe it means focus is going to be on making current complete and WIP developments better instead of adding yet more projects to the mix. I could be wrong though and history points to me being wrong on this one

 

"Your feedback sets our priorities" is the title of the first major slide.

The guy that started this thread, and the project he was working on when fired, is a prime example of why this claim is highly dubious and causes viewers to wonder about the veracity of the rest of the video.

 

I am curious how the studio feedback is prioritized now though. Having been directly on the receiving end of feedback from studios when I was at Unity; they will always ask for what directly impacts their project & needs; and this doesn't always correlate with the wider needs of the company and engine. Of course; there is alot of overlaps between studio requests; like "Please stop multiplying shader variants automatically" or "UGUI performance needs improving". But Unity's goal is to make the stock price go up and that usually means other focuses than these more technical specific details.

Plus some feedback (such as "Stop wasting resources on DOTS" or "Remove all 3D and go 2D only" or "Only support consoles, mobile is not worth it anymore") is either not feasible or something Unity would ever do for very obvious reasons.

 

I was referring to the oldest of requests of Unity, by its entire user base... that Unity learn the pain of using Unity by using Unity, that that's what you were doing when fired, and that was canned, at an incomplete stage, when last 20% that is the last 80% of work still lay in front of Unity for them to learn about 90% of the pain its users suffer the most.

 

I think the idea was not to make UNity learn the pain, but to improve the engine. While those two goals sometimes align, they're not the same.

P.S. This reminds me of cenobites.

 

Oh yes, that too.

Basically every game studio that speaks to Unity regularly asks for Unity to build a game with Unity, release it and maintain it. Curious how that feedback is processed now.

 

I tend to think that it had become a considered essential, by most users, because they felt their outcrying about the pains of finishing in Unity and need for improvements in the engine and editor were being under appreciated because Unity hadn't experienced that pain for themselves. And still hasn't, despite getting to about the 80% completion point in the project slated/created for them to go through to completion via that next 80%... and the subsequent general feeling of disappointment that swept through the Unity user base when they learnt the entire team had been sacked - and the project cancelled - something everyone wanted to see them experience and learn from was gone to the winds.

 

One thing I did notice from the GDC presentation is all the content/features is from releases within a roughly 6 months window. Nothing saying "we want to add x but its probably years and years away". Feels alot more realistic and current.

 

Sadly, at the current trajectory of things like The New Input System and UI Builder, URP etc, that doesn't seem to indicate they're any closer than moonshots.

 

Mainly just ongoing support/improvements?! Some of this stuff has been in development for soooo long.
So better they focus on finishing what is on the plate now before starting the next course.

UI Toolkit still not fully runtime ready - will people use it for runtime or stick to UGUI?

ECS/Dots etc. No idea on what pickup this will get - would prefer better C# optimisation across the board, not Unity-specific ways to write faster code. But if easy to learn OK.

URP/HDRP divide. Not sure where one ends and the other starts, but I am going to guess that unless you want highest levels of realism (real-lighting values etc.), use URP in future, if you have dropped built-in...

 

What does native C++ multithreading for WebGL mean? How do we write that code in the context of a typical Unity project that uses C# scripts?

 

I think it went like this: with IL2CPP the project is first converted to C++, then from C++ to WebAssembly.

 

If the roadmap seems pretty underwhelming in both terms of features and bugfixes, please notice that the actual name of the presentation is "The Unity roadmap for games"...

 

Right, but I had the impression this is something different than multithreaded C# code compiled to C++. At least this link indicates that "Multithreading C# code is not yet available", though perhaps the docs are not updated.

 

Code Monkey also has a nice summary:

The approach of data oriented programming is by no means a Unity-specific concept. It's dictated by the way our Hardware is that this is what leads to better performance.

The ongoing migration to MS's C# environment has also been mentioned and that should improve C# in general too.

Good question. I would have assumed and am hoping they mean multithreading via the job system. That C++ is mentioned explicitly is confusing though.

 

2022 roadmap was the same. I can't tell if they're focusing on current projects or don't want to publicly commit to anything new since historically it hasn't gone well for them. If they don't announce anything that still needs years of development, no one will be disappointed.

 

And still no DEDICATED global illumination tools, haven't heard a peep about tools from their weta acquisition to be used by US, and nothing groundbreaking.

Honestly, I just finished watching Unreal 5.2 and their State of Play and as a LONG TIME Unity user, I have to ask what the hell are you guys doing? Metahuman Animator looks CRAZY and you don't have an answer to any of this!! Most of your user base CAN'T use Ziva, And unless you own a 3D printer to make HMCs, you can't use the facial capture tools.

You guys show demos while they look amazing, there was also a demo made by ordinary users that didn't get much play. If you REALLY want Unity to do well, why don't you build a demo from out of the box tools to shiwv what can be feasibly done by designers like me, who are running one man shows. Also, a lot of your samples are not artist friendly... I saw your eye shader for Enemies and while it's looks beautiful, it's unnecessarily complicated.

And lastly, HAVE SOMETHING FINISHED to actually show... Don't have all of these ideas and none of them are done

 

Im assuming you missed out on the Ai trailer . I honestly can't recall the last time I witnessed that bold claims with zero to show - seemed desperate tbh. Like that their stocks just flatlined through it though.

 

Typical Unity trend chasing, I have 0 expectations.

 

I saw that trailer... Straight clout chasing behind Unreal 5. Unity needs to get back to what made it special and stand out

 

It is slowly turning into one of THOSE threads again, isn't it...

 

Hey if interesting tools pop out of it, let them chase trends xP
There's a small thread on the core of it and it shows that it's at least not a relabeling or anything but stuff specifically for Unity: https://forum.unity.com/threads/unity-and-bariumai-text-to-pbr-material-generator.1392916/

 

No, but the keynote didn't bring much fire

 

Has anyone ever seen DragonCoder and Code Monkey in the same room?

 

Unity's tools that actually exist are barely finished in a lot of cases so I don't have much hope for their AI initiative.

 

I actually hate this. The roadmap site and various team members are usually talking about what is currently going on, supported by the occasional blog posts, but now we don't have the visibility of the estimated trajectory Unity is doing. I miss it, it was great to know what Unity thinks about the engine's and the editor's future. Obviously people whining about it killed this as usual. Just like the access to experimental packages.

 

The most annoying thing in this roadmap was the material variants part! It has been advertised for soo many times! If i remember correctly it was repeatedly showcased in literally all the 2022 events!

 

I see, so we can expect some speedup of internal systems. Definitely nice!
Would be really cool if the job system would be compatible with that too though.

 

Does this mean OnAudioFilterRead will finally be brought to WebGL?

 

I love unity but this GDC for me was kind of "meh... at least they are geting out of preview some things" not bad, but ok...

What is HMCs?

Yes is a bit complicated but in my humble opinion, is the mos realistic eye shader in a game engine right now, the character from Enemies is the most realistic i've seen on a game engine, better than anything from metahuman at least (in my opinion)

Agree but i think 2022 LTS is going to be a solid version, that being said, 2022 LTS feels like that was the version that should have come in 2020

 

Helmet-mounted camera. Second link is an unrelated company but it shows off some designs for one.

https://blog.unity.com/industry/enemies-evolution-of-digital-humans-with-ziva
https://www.javelin-tech.com/blog/2016/03/motion-capture-camera-design/

 

An HMC is not that hard to get. You can make a basic HMC rig for like 20 bucks. If you are capturing facial expressions, you are already expending much more than that. It's certainly cheaper than the stereo camera needed by Metahuman. Plus a head mounted camera can really improve performance, as it allows actors to move around a lot more.

I don't mean this in opposition to your general point. I also wished there were more exciting things by the end of the video.

Although, in retrospect, now that I'm thinking about it, maybe it's a good thing we got this kind of announcements. I feel Unity got harder to use when they kept announcing new revolutionary thing after new revolutionary thing some years ago. I think my experience working with Unity is improved a lot more by these many small enhancements and features, than by huge paradigm shifting developments. But then again, I don't do AAA type stuff.

EDIT
Sorry, I should make a correction. I'm thinking my basic HMC price may be too low. Some friends of mine made a nice enough mocap helmet some time ago, and they spent like 20 bucks, but they already had a bicycle helmet and some stuff to fasten the camera. Still, it's something relatively cheap. Plus there are places like fab labs in many cities around the world where you can rent a 3d printer if you need it. So the spirit of this still stands.

 

think Unity already has amazing graphics, sort - of obvious in the 2019 prototype, think was called ' Heretic ', and also trailers of the engine, the talk says stuff like 1. easier to use, 2. faster loading, 3. faster code, and overall those are the things really looking for, if get half the stuff in the talk over 2023 think it'll be a vastly better engine, it'll appeal to many more also customers, and as for me it's precisely the content, features, or updates looking for . . .

the four things that really have my interest, not sure if a few of them already have been added . . .

1. AAAA // PS5 - quality graphics, or super - movie fidelity
2. faster load - time, less waiting in the editor
3. export to web - pages
4. better coding, etc etc

those are the things they're talking about, and overall think Unity has everything you could ever want, however it could still be a better engine, also the various packages could use updates, aka it's going to be a busy year, also sort - of looking forward to a few surprise features // ' industry - leaps ' over the year . . . .

( checked it, Unity also has support for WebGL builds, much easier for a jam, or sharing a vertical slice, or more prototype, engine has a ton of good stuff, overall think the roadmap only makes a ' great ' engine better, however should be aware of bugs // quick-n-dirty features . . . . )

 

I guess I don't understand why it was worded that way then. Or featured on the roadmap at all. I see "multithreading" and "WebGL" and am immediately excited, but of course it's not what I or anyone else actually means when they think of that.

Should say I'm grateful that progress is still being made towards improving the web stuff. I'm sure it wasn't a simple process. Do you know of any performance benchmarks or expected improvements that us mortal, non-C++ plugin-writing developers can expect as a result of this update? And do you know if C# multithreading (including via the job system) is likely to ever be supported?

 

Unity currently suites my needs, but the company looks like it's on cruise control.

I miss the energy of when you guys were gonna kick ass with DOTS and reinvent visual scripting, and revamp each and every component to be better and faster.

WTF is even going on? Was the performance revolution of DOTS a death knell? All forward progress seems culled.

 

Oh, yeah. The first time I designed my own HMC was in 2018 using a skateboard helmet and additive manufacturing. It was too heavy and now I'm designing one that is ultralight and durable. I even had stereo cameras (it was actually two action cameras that had to be synced in post). And there is a program I'm looking to test called Avatary, which allows you to capture your face.

Unity while is still great, is getting to a point where it's getting away from being artist friendly.

 

The only problem with these demos are again using tools that most of us don't have readily available (like 4D captures), and the features they are using doesn't happen out of the box. If Unity wants to show how easy it is to make a trailer or create a working game prototype, SHOW US using the tools that are already there.

 

Ignore Unity's demos. When Unity creates them they will often create a tool to make it easier to create them but then that tool never ships so even if you can create the environment, the characters, etc you can't recreate that demo yourself.

 

This. And even working at the fidelity of the Enemies demo, you'll need access to Ziva Trainer and about $9,000 to solve your character. Also it's a Maya plugin

 

You can, if you also recreate their tool and development process. Not sure if that is worth the effort, though.

 

And that in ITSELF is a lot of work. If you're an experienced 3d artist and have a small photogrammetry rig or a handheld scanner, you can try. The most manual part of the process is building the control rig.

If you have a nonhuman character, you'll have to sculpt not only the creature, but ALL of its facial expressions in ZBrush

 

Yeah agreed! We are just recently getting the probe volume light baking toolset used in book of the dead in 2018!

 

See, now you know why I prefer low poly visuals.

 

I'll admit, I've been wanting to see more non photorealistic rendering, but your can't escape the work required

 

Not surprising. Everything is going too fast during production so everyone forgets after a day or two. No one has enough time to write a list of "what problems we've had with unity".

The presentation contains only a few selected features from a few selected categories. For example, 2D got soft shadows in 2023.1, and HDRP will get something similar soon:
https://twitter.com/robcupisz/status/1505877478526574594

There was no information about any of that. I suppose someone didn't know about these, or unity checked their target group and realized studios still use 2021 and 2020 LTS. Every new feature in 2023 is a far future for them, especially If it's somewhat unstable or experimental.

Honestly, I'm happy they don't overpromote barely working, experimental features.

 

agreed (kind of)
Look at this project, very cool
https://forum.unity.com/threads/dust-chamber-story-driven-antisurvival.1135852/

I really wanted to see more realistic graphics, i'm a bit tired of low poly, stylized, pixel art unity games -.-" (i'm making a low poly game tho xD) and the assumption that "unity can not do proper 3d" "unity is just for 2d"

 

I was working on a fighting game with Unity, but I need to get better hardware to realise my vision and using HDRP for it as well. The problem with full 3d games is it requires a lot of work and time, and most devs look for the fastest way to produce content. Also, most indies don't have the resources to do it at AAA quality.

Need to build a character on the quality of Louise from Enemies without 4D data? You'll have to go into ZBrush and sculpt a character of incredible detail and do that many times for the characters in your game.

Need high quality facial animation? You better know how to do facial performance capture since most of us can't use Ziva Trainer.

You also need to ensure your pipeline is SOLID!!

For my game development projects, except for programming, I am neck deep in the art pipeline, including my own rigging. Hell, even my skin shaders and eyes are on par with some of the best games out visually.

To correct my earlier statement, it costs $1800 to convert your character into a Ziva Puppet, according to a video I saw about it. Still out of reach for a lot of us