Graph Tool Foundation

So it seems like the path forward for those of us working with the Graph API or GTF package is to go ahead and use the Experimental Graph API and wait for the integrated release of GTF within Unity, then port API changes at that time?

 

Yes, definitely! The earliest that we could possibly see GTF is 2023.1 (since 2022.2 is already in beta and it isn't in there). And it will not be backwards compatible since it will be built-in to the engine. So using the GTF package means using a package that will not get any bug fixes or updates.

 

This really puts me between a rock and a hard place. I want to develop Editor Tools with a node graph, but it is not really feasible for me to rip the Graph Tool Foundation code somewhere from GitHub. So all I can do is use the GraphView API and hope, that it is easy to migrate, once GTF gets its full release? I am also wondering if it is really a good idea to make this part of the Core instead of a package.

 

I agree, the state of this and the path forward is very unclear.

https://forum.unity.com/threads/gra...disconnect-are-internal.1315425/#post-8321775

 

Public API is not their priority. Looks like it'll take several years still to get to that point.

 

I have been uncertain about what node editor API should use for over a year, since I'd like it to be my definitive choice. Unfortunately this can't be the case for all the reasons listed in this thread, but if I can give an advice to someone who is starting out is just going with GraphView for now (really, go for it).
I'm trying it following the tutorial from Mert Kirimgeri on Youtube, but converting it to my needs (not a dialogue system).
At the state of things imho it's very important to distinct the logic layer and the editor (GraphView) layer, where the editor layer should just have the task of offering a UI for your system, while all the logic should be somewhere else: this has the advantages of not committing to GraphView or whatever node editor api you want to use, and that your system will also work at game runtime.

 

So even GraphView is marked experimental, not sure that Unity's own internal teams have a solid foundation for their graphs!
There are github node systems, anyone used any?

 

@andyz I've only tried xNode https://github.com/Siccity/xNode but it is great.

 

Well this aged like fine milk...

 

Nothing is wrong, those lumps are supposed to be in there and you're supposed to like it, move along!

 

There's a new blog post concerning Graph based tools in Unity: https://blog.unity.com/technology/games-focus-empowering-teams-for-cross-functional-collaboration

So two to three years, I guess. And since public API is not a priority, probably longer than that.

 

The thing is -- you don't "battle test" an eventually public API internally.

I get it that people complain about their stupid/poor/misinformed/misunderstood design decisions, and you don't want that feedback muddying up making the early-but-poorly-designed-product better, but you've got to let it evolve organically AND out in the open. That's the whole purpose of something like github.

In fact -- Unity is destined (and a great fit) for open-source development. Community collaboration. Transparency. Unity and Blender are two peas in the same pod, except one is starting to get rotten because it has cut itself off from the pod (and therefore the plant and roots where it originated), and the other is flourishing and is tasting sweeter and sweeter each day because it embraces its origin and uses that to guide its growth. Can you guess which pea is which??

@Unity

Either you get your act together and hire better and more practiced tool and product designers (or at least _listen_ to the good ones you actually have) and stop with the corporate posturing of your "tech" (i.e. by touting and showcasing "oP3n W0rlD 0pTim1zAshUn!" on a tiny well-funded game like Alba that wouldn't even be considered _close_ to "open world" by anyone besides yourselves, while saying very little about "optimization" techniques in detail anyway, since this all comes off as completely insincere and outright tone-deaf to the state of the community and to the state of the product itself) -- OR you will have a failed technology platform, languishing in the waste of a rotten company, with a(n) (allegedly) rotten CEO. Unity -- You are very much on track for the latter. I suggest some soul-searching.


I think this company needs to start behaving a lot more like the Blender Foundation, and a lot less like Google.


This is especially true, considering all of the goodwill this company's founders have earned over all of these years -- Unity could definitely prosper with a model like Blender's.

Better (and more public) transparency into its deep internal development is a great start.
And that can start with major foundational tools like GTF.


Also, as an aside -- please stop relying on the Asset Store.
Lots of the 'tools' there are half-assed S***. And the ones that aren't, have authors who still struggle with basic UI / UX implementation for features Unity keeps changing, pulling the rug out from underneath them each and every time, and still taking a huge 30% chunk from them for basically nothing.
If Unity was more transparent with their process, people would be more understanding, and as a result, once the tool developers have access to Unity's code changes, for example, we could see what direction Unity is heading in on our own, as well as how far along they are, and we could also keep up and plan our own tools (and input/feedback) better -- and some of us might even be able (and willing) to greatly contribute to Unity's development, given the right level of knowledge (meaning you wouldn't suffer from brain-drain anymore because nobody knows the specifics of the underlying technology anymore after your most skilled programmers bail on you and leave your S***ty ass). I personally know a number of people who would sacrifice their own personal time and energy to improve Unity -- if they had the support of Unity's internal team and developers on their side to help them sort out the things they don't fully understand (and are poorly documented) about the engine and its design -- as long as they have a chance at seeing their changes potentially be accepted to the engine itself (or at least critiqued and reviewed, asking for more or less changes, with feedback on why or why not these changes are accepted -- or not -- into the engine, and what needs to be improved). This is exactly how Blender does its open source these days. If you haven't looked into it, I suggest you do so, @Unity.

I especially suggest looking at the technology management structure in the Blender Foundation and see how they manage their resources and technology internally too. They are about as close to a tried and true game-engine development company as one might ever hope to get -- and their product is "free" also. I'm not suggesting making Unity "free", but Unity is a _platform_, not just a product. Blender is still a product. But to better develop Unity's product (which also doubles as its _platform_), it can look at how Blender operates to see how ruthless (or not) it should be. For example, BF even cut out its own "game engine" portion of their (free) product to focus its limited resources on the part that mattered most to them -- the ability to better create 3D art. Now their product is developing at breakneck speed. Unity is actually doing the OPPOSITE. Being able to focus on a core vision is the kind of ruthlessness Unity needs -- not merging with an AD company to make a quick buck or two for an aging and increasingly unpopular CEO. :/

Just my two-cents.


So please -- I ask not just for better transparency on these kinds of projects, but for transparency in the company as a whole. You no longer have any real competitors, but if it makes you feel better, keep your ad-related S*** locked-up tight. I don't care. -- I only care about the game-engine portion of Unity. That's what it's selling to developers and movie-makers and other industries alike. Unity currently has an identity-problem. So that's why, if you're looking to find someone to emulate -- I suggest emulating the Blender Foundation.

Since Unity's CEO doesn't quite have a strong enough vision for Unity (because he clearly can't understand why _anyone_ would want to make a game and not want to spy on people so that one can selfishly quantify how and why someone enjoys a game they've made), it is clear he has never been a child who has drawn a picture he's proud of, and desires to show it to someone else (such as his mother or father), to see how well (or poorly) he really drew it to decide whether or not he was good at art, or whether he needed to improve his ability further. This limits the CEO's ability to see beyond his own nose to understand the long-term benefits of true openness, honesty, and ultimately -- transparency -- even in a capitalistic world. Maybe he has something to hide? But I guess, for some people, enough money makes up for the giant hole in a person's soul, and as such, maybe we should ALL be making as much of it as we can -- even at the cost of bankrupting our fellow man.

At least it's you and not me, right? :/

Anyway, I digress.

I think at the very least, Unity (as a company) should start considering where it came from, and why something as simple as a Graph Tools Foundation(al) API should matter more to Unity than just a bullet-point on a checklist to keep users from running away in droves. But I'm not sticking around because of a bullet-point on a checklist. I'm sticking around because I see the true potential in Unity as a creative medium -- and as a platform that doubles as a powerful springboard to all other major parts of the science and technology landscape. Unity can save the world.

Or not.

Unity being reduced to being an AD PLATFORM is just insulting to me personally. :/

 

Their focus is on their own tools, which is fine. But it seems they want to roll out GTF for all existing Unity's graph based tools in one go, perhaps as a big marketing push. Either it's going to be great in 3 years or a complete dumpster fire developed in yet another closed off silo. I'd much prefer a more gradual release cycle.

It also risks falling victim to the same fate GraphView API went through where a lot of its API is closed off, internal only and only exposes stuff needed for ShaderGraph, which is not really that great for a general purpose graph API.

Also, this is yet another Unity system that has a 5+ year development timeline. GTF was announced more than two years ago. And by their own words, they expect 2-3 years more, which judging by past performance means it's likely to be more than 3 years.

 

I really do wonder what's taking so long there, I've written way more node editors than I'm willing to admit. Obviously it never took that long, but it had (as far as I can tell) the same featureset (the features of GTF now).
It's just drawing a node graph, blackboard, inspector and context menu. They also added data handling (which should be implemented as optional feature if you ask me). But how can it take more than one year to create such a tool? Aren't there devs working on this full time?

And... it's already there? GTF exists and kinda works, it's not very flexible, but ... are they planning to rewrite it from scratch multiple times?

 

GTF is based on UI Toolkit, which has been in development for 7 years and still is nowhere close to finished. So we have GTF team depending on UI Toolkit team which might or might not be in the same country and time zone.

Then we have ShaderGraph, VFX Graph and UnityVS teams which are all separate and all need different things from GTF. Again, these teams might or might not be in the same country/time zone.

Someone also has to port Animator to GTF, which could be yet another team. So it's a multidirectional communication between 5-7 separate teams that are geographically scattered and everyone needs something different from GTF. Hashing it all out in a company of Unity's scale takes forever. And iteration is slow as well due to internal processes focusing on stability, backwards compatibility and planning ahead half a year if not longer. It's just the nature of the beast this size, unfortunately.

GTF package is cancelled, it hasn't been updated in more than a year. As far as I'm aware, it can't produce working builds. And they're integrating it as a core engine tech, i.e. not a package. It will change in major ways throughout the following years to accommodate existing Unity graph based tooling and public API is not their priority according to recent Q&A here in the forums.

 

Animator won't be ported to GTF btw

 

Got a source on that? I believe the intention is to eventually have the editor be fully UI Toolkit based. Animator would fall under that unless they have a full replacement in the works.

 

This is correct. They do plan to fully move the editor to UIToolkit. And 2(?) years ago they did a AMA type of thing after Unite and I specifically asked if the plan was to move the Animator to GTF, and the answer was yes. (You can stock my comment history if you really want to find the source)

Now from my understanding, it is not in the immediate plans. Only ShaderGraph, VFXGraph, and VisualScripting are currently being worked on for GTF.

 

Sounds like _someone_ is terrible at running / organizing a technology company.

I guess they just need to mandate Zoom meetings throughout the day to cover for all possible time-zones -- all across the company -- and never get anything done _that_ way instead. :/

 

They're just unable to commit what is required, probably due to internal reasons.

Eventually they're migrating to

Graph Tool Foundation Package

, so the

GraphView

will no longer exist. Until then, use the only functional system which is

GraphView

. In a few years you can transition to GTF but don't hold your breath and don't expect any suggestions on the

GraphView

API to be considered.

Now : GTF not publicly maintained.
Now : GraphView developed internally, porting to become GTF eventually.

Future : Graph Tool Foundation Package
Future : GraphView deprecated.

 

"Internal reasons" is much too vague of a "reason" for most people to risk placing their business's future on top of Unity.

I respect your opinion though.

So, while I agree with your hard-as-nails practical assertion of "just do what you gotta do", I've gotta say I'm tired of hearing that being the go-to "solution" for Unity's current incompetence as a company.

I think it's not a good idea to defend Unity anymore. They really need to wake up and look at the writing on the wall -- at least from a business perspective.

I'm not all "gloom-and-doom" though.

For a while, when DOTS was announced, followed by Animation Rigging, Unity seemed to be going in a great direction. But it always seems like they run into practical "design barriers" that prevent them from properly delivering their technology. Part of me wonders if they're actually _planning_ to deliver technology anymore -- seems like they tout a project, throw two or three guys at it, and cancel it within a year. Anyone heard anything about Kinematica lately? Lol -- That's dead. Visual Scripting? -- Seems pretty dead too. Just like any previous direction they might have hinted at in the past. "Democratizing Game Development!" -- Anyone remember that slogan? -- It's also dead.

Throw some "exciting to markets" stuff on their timeline, extend the life of their company, and fail to deliver on every single thing. That's Unity's MO these days. And probably the CEO's chance to pad his pockets before he is either fired or bails on his own.
They just don't want gamedevs to leave their engine in droves, so they keep stringing us along with "just wait until 2-3 years from now! -- I promise! -- It'll be great!!!"


Unity these days just reminds me of that alcoholic husband who _swears_ he's going to quit drinking -- he promises, over and over, that he'll _definitely_ stop beating his wife (us gamedevs) and _for sure_ go check himself into rehab next week! -- But next week comes and goes, and then the week after, and then months, then years and years -- and he's _still_ unwilling to admit he just doesn't want to stop drinking. And his wife (us) keep hanging onto his word every single time. But only because we don't want to start over -- we don't want change!! -- but it's so clear that he's never going to stop making promises he never plans to keep. So the question becomes -- Is this how we want our life to be?


I mean, c'mon -- Unity outright lies to us developers.

Calling Alba an "Open World" game was the straw that broke the camel's back for me.

But then they proceeded to provide "tips" to NOT simply iteratively develop your game -- which is supposed to be Unity's bread and butter -- the whole _purpose_ of Unity's "easy-to-use" tools and approach to game development!


Lol

I'm just tired of the bullsh*t. You can defend them if you want -- but I'm calling bullsh*t on that particular defense going forward. Just saying.

If you seriously want the engine to improve, and really stand by the company -- you should invest in them changing.

-- By NOT investing in them at all until they do.

And that includes trying to keep them morally afloat when they refuse to listen to you.


I don't know what kind of capitalistic nonsense you've been fed your whole life, but this isn't cool. And taking it up the rear-end -- and then asking for more -- isn't how real people do things.

 

I never said I support the state of affairs or their approach to their engine. I don't even know what Alba is.

GTF is apparently a mess, just use something else if you want to, or hobble along on GraphView until they (maybe) get around to finishing it. That's basically all I said.

I don't think it's within the scope of this thread to discuss large scale policy change for Unity anyway. Maybe another thread with some concise information would be more appropriate and effective for that.

 

Fair enough.

I'm just tired of Unity being one giant turd sandwich -- Then watching people just gleefully eating its sh*t.

 

How about new animation system? Does new animation system front end also developed based on GTF?

 

As far as I would like GTF to be public released. I still very against the idea of moving it from package into internal API

Decoupling dependencies and focused more on UPM should become common practice. As they say, eat your own dog food

Are there any sample or glimpse of what it would look like?

 

The editor UI isn't a package because it is required to render the editor. I don't think it's possible to turn it into a package when they want internal tools to use it.

URP shall replace the standard render pipeline some time in the future, one of it's dependencies is ShaderGraph and one of ShaderGraphs dependencies will be GTF. So it would be a package that is installed in every Unity-Editor and can't be removed. And that's not what packages are. Packages only really make sense if you can add and remove them as required - atleast that's my understanding of them.

TL;DR: I think internal tools will require GTF, so it has to be internal too.

Note:
UIElements is actually a "built-in"-package but so is Audio, NavMesh, Physics - unlike usual packages, the code isn't on the C# side, there are just C# interfaces (static classes, components, ..) calling to C++ side. So I wouldn't call them "packages".

 

I would like ShaderGraph to be package and let it has dependency on GTF. As well as render pipeline and navmesh or even animation

All packages should be treat as first class citizen. They could be built in packages. And it has more benefit that it could be fixed or upgraded separate from unity core system. And the code are opensourced so we can debug and fix bug when there is a bug

Ideally I wish that every module of unity that don't need to be closed source should be package and could be switched for the custom version

 

Hi,
I just tried your repository (both: with Package Manager via git import and just downloading the plain zip) and it does indeed compile in Unity 2022.2.3. However, it throw an error when opening, e.g. a sample graph window:
-----------------
NullReferenceException: Object reference not set to an instance of an object
UnityEditor.GraphToolsFoundation.Overdrive.MainToolbar..ctor (UnityEditor.GraphToolsFoundation.Overdrive.CommandDispatcher commandDispatcher, UnityEditor.GraphToolsFoundation.Overdrive.GraphView graphView) (at Assets/com.unity.graphtools.foundation/Editor/GraphElements/Views/MainToolbar.cs:76)
UnityEditor.GraphToolsFoundation.Overdrive.GraphViewEditorWindow.CreateMainToolbar () (at Assets/com.unity.graphtools.foundation/Editor/GraphElements/Windows/GraphViewEditorWindow.cs:128)
UnityEditor.GraphToolsFoundation.Overdrive.GraphViewEditorWindow.OnEnable () (at Assets/com.unity.graphtools.foundation/Editor/GraphElements/Windows/GraphViewEditorWindow.cs:170)
UnityEditor.GraphToolsFoundation.Overdrive.Samples.Recipes.RecipeGraphWindow.OnEnable () (at Assets/com.unity.graphtools.foundation/Samples/Recipes/Editor/RecipeGraphWindow.cs:24)
UnityEditor.GraphToolsFoundation.Overdrive.Samples.Recipes.RecipeGraphWindow:ShowRecipeGraphWindow() (at Assets/com.unity.graphtools.foundation/Samples/Recipes/Editor/RecipeGraphWindow.cs:18)
-----------------
My guess is that the path to the MainToolbar.uxml file is somehow wrong and it can't load it so the tpl variable is NULL when calling tpl.CloneTree(this).

 

OK, seems like you just have to change the hardcoded(?) path in the AssetHelper class (AssetHelper.cs)
internal const string AssetPath = ...
to your own installation of GTF.

 

Can a window contain multiple graphViews？

 

The package itself was not updated and cannot work in 2022.2 and 2023.1 anymore without error of changed API. While the real tool was still a long way to go. Please at least fix it to work in the meantime so we can continue using it while waiting for the real ones. Did not need to add any more feature just make it so that we start using it in current version and can convert our code when it really released

 

@EricDziurzynski Hi! I have a quick question: In the current GraphView, edges don't have their own color property. Is this also the case in GTF? If I use the alpha of the color of the edge to represent the weight of the input, and there are multiple input ports connected to a single output port with different weights, only one edge can show the correct color.

 

Why did Unity abandon the GraphView API as opposed to just updating and improving it? I just started using it and it seems pretty solid. Will GraphView ever be removed from Unity, and if so, when?

 

Currently, GraphView uses too many internal APIs, making it difficult to override default behaviors. Hopefully, this situation can be improved in GTF.

 

One evidence that moving package into internal is a bad choice

 

Hello !

I begin the development of an open source tool that require a graph view. The GTF API seems very promising, but it also looks very unstable at the moment. Do you know for which unity version GTF would be made available for public and if the upgrade from the preview version currently available would be easy? I wonder if I should start working with the current state of the API, or if it would be preferable to go with some open source packages like NodeGraphProcessor that rely on the forever experimental GraphView API.

Thanks !

 

It seems that a stable version of GTF is still far away from us. A few months ago I asked a similar question, and Unity’s reply was to suggest using GraphView instead of the current GTF.

 

Thanks for the answer, I will stick with GraphView for the moment. I hope we will have some news on the state of GTF api soon

 

I guess I'll have to use GraphView then...

 

I walked down the "path of pain and tears" with GTF API too, but ended up using Unity's VisualScripting in the end.

https://docs.unity3d.com/Packages/com.unity.visualscripting@1.8/manual/vs-advanced-topics-intro.html

Creating custom nodes, controls and events required some rethinking, especially handling the flow myself, but I have to say it's working very nice! And the previous experiences with GTF helped a lot, e.g. creating nodes with configurable inputs. Maybe that's an alternative for some people

In case you're interested, how I used custom VisualScripting, you can check out my project + code here (code is in the FmvMaker/Assets/FmvMaker/Scripts/Graph/ folder): https://github.com/FireDragonGameStudio/FmvMaker

 

UnityCsReference/Modules/GraphToolsFoundationEditor at master · Unity-Technologies/UnityCsReference (github.com)

 

Li Hai Le !!

 

Wait ?
You were able to test it ?

...How ?

 

Just download the Beta/Alpha versions of Unity Editor via Unity Hub. Haven't tried it myself, but the previous post had Unity 2023.1.0a21 installed.

 

这是一个简单的案例：
iLoveNiu/GTF-Test (github.com)

 

Thanks, I had a lot of fun playing with this !

It's missing a lot of stuff like the ability to create connections that are neither an input or an output and I couldn't find a way to had logic to the graph or use it in component.

But still, it looks good.
Can't wait for the official release !