2020 Project Direction

#21

Hello, I am new here, so let me quickly introduce myself and give some background before I get to the point. Sorry if this post gets too long :slight_smile:

My name is Milan, and I’ve been keeping an eye on the Amethyst project for a while now, but I’ve always been very passive, checking the website occasionally to see if there are any news, and trying out the engine whenever I had a little bit of free time (that happens extremely rarely, since I’m a university student with a part time job).

As such, I’m very much a beginner in both Amethyst and Rust and I don’t have a deep technical insight into the engine like most of you guys seem to have. I barely know there is something happening with Specs and Legion - I admit I have not researched these things well enough, so I’m still not sure what kinds of benefits are there. But that is not the point.

The point is that I’m mostly an Amethyst outsider and a hobbyist game developer with a fair bit of experience in Unity and Godot. I want to give you my perspective on how I view Amethyst, mainly from the point of view of a new-ish user. I may compare Amethyst with Godot along the way, I hope you’ll forgive me for that :slight_smile: Also, I may make some incorrect statements and false assumptions along the way, please correct me whenever I do. It’s the result of not following the project closely enough.

That’s it for the introduction, now for the main part:

It has already been mentioned that the engine is hard to use. I agree completely. But it is not just the API or the documentation. What I miss the most when coming from, say, Godot is the editor. In my opinion, an editor makes a huge difference in usability and I would say it is essential to have it in any real-world project.

I have noticed some attempts at making an editor, but they all seemed both lacking in functionality and abandoned. If I recall correctly, there were two official attempts to make one, with the first one being cancelled in favor of a rewrite which, last time I checked, was inactive and incomplete. I was also stoked when I saw a Blender integration (it was called Arsenal I believe), which I thought was the perfect solution. Unfortunately, that project seemed dead as well.

Amethyst itself seemed to progress very slow in the past months. The latest news were from October last year, which announced Grumpy Visitors. I remember there was also news about WASM, which was supposed to arrive sometime in late 2019 if I remember correctly. Whenever I opened Github, all the latest commits seemed to simply fix typos here and there.

I know this is a project people are working on in their free time, but it got to a point where I feared that the project is dead.

I really hope that one day, Amethyst will reach Godot level of activity. Whenever I visit Godot’s website, it feels like an exciting adventure. Whenever I visit Amethyst’s website, it feels like a ghost town.

Last but not least, and this may just be my feeling, but I think Amethyst focuses way too much on low-level technical stuff. Things like Specs vs Legion, WASM, renderers… I mean, all these things are obviously very important. But after all, Amethyst can be used to make games already, right? Why not push for more ease of use (like focusing on the editor for now) instead of going for, say, the most performance possible?

I guess it’s about tradeoffs. As a programmer, I completely understand why one might fall into the trap of playing around with cool shiny technologies. After all, it’s one of the reasons I’m so interested in Amethyst and Rust. But right now, it feels like everyone is polishing the REST API while the frontend guys are nowhere to be found.

I think that if there is a massive push in the usability aspect, more people will start to use the engine, and as a result, more people will be able to contribute. Refactoring and redoing stuff can always be done later - it doesn’t have to be perfect from the start (and it won’t be, no matter how hard we try). I remember that when Godot started out, I hated its editor. It was really clunky and weird, but it was there. And people have been using it. And they have provided feedback. And slowly, feature after feature, it got better. And now, look at where Godot is.

TL;DR version:
I think Amethyst desperately needs more activity, and way more focus on ease of use. Stop trying to be too perfectionist and focus on features that make the engine nicer to use.

That’s it, sorry for the wall of text. I really hope Amethyst succeeds, but I think it won’t if the bar to entry is too high. I hope this post is helpful in some way.

5 Likes
#22

New here. Just to give some direction advice’s.

I barely can do anything in rust. Or in any language really. But I know that Unreal or Unity is just not something I believe in. Or Godot for that matter. (bad 3D and copy allot of Unity’s bad stuff)

First thing for 2020 might be to update stuff to be current year? I mean if you check on Amethyst Home page under “RoadMap” you might get what I mean. The hole project looks like it is at a stand still. Makes one wonder if there is even a reason to look into joining in you know? 4 months out of date is not good. No direction indeed.

Get stuff to be usable. And by that I mean whatever talk about Legion is a mess as someone not knowing anything about the core systems is scared away from whatever is going on deep down in the engine roots. Like is it worth looking into anything while it is up in the air if anything is going to be the same in 6 months? I’m ready to learn but it looks like stuff is in question to change wildly so should I? Or just go somewhere else? Or just wait (as I have for eeeeh 4-5months now) and check back then?

The main reason to use Amethyst VS doing stuff on your own is listed on the .amethyst.rs Yet why go with Amethyst when your leaving stuff undone and up in the air. : / Why not role your own at that point.

Ow and sort out Physics. The pain of most open source game engines are the physics and 3D : c

=== Tip about working with Arsenal ===

I have played around with allot of stuff trying to find something open and powerful enough to stick with and grow with. And out of all game engine related stuff I have tried Armory3D was VERY interesting

Basically due to Blender integration and that everything was done in one language more or less made it stand out as something worth investigating. (and with Blender/a editor your in the right place to do that)

Armory3D is a bit more of a framework then a game engine with a editor if you ask me.
Everything code related was easy to open up and read, and due to being inside of Blender one had more or less everything but Code editor and Gimp at ones finger tips in the “engine”! Big+. So powerful yet way easier to do stuff solo. You can borrow allot from Blender and it is a given a game dev is going to learn some kind of 3D or 2D tool. And right now Blender is king in at least on the modeling side.

Blender UI is not that bad considering all the tasks a game engine needs to be able to do. Agen Armory3D and 2.8 Blender was very much something I had loved to see happen. (UBGE exist I know)
It might not fit everything but it is like the next gen of Unity/Game engine. Having so much of the tool chain integrated together and working together strength to strength while at the same time learning Blender fluidly is grate!

Video editor. 3D modeling. Sculpting. Game engine. Animation. Grease pencil. Blender have ALLOT going for it right now. Open source. Free. EVEE. Visual scripting with Blender Nodes? Not that I like VS.

Blenders biggest drawback was fixed in 2.8. Now only thing naggin me that it sometimes seems slow and probably not the easiest thing to integrate a game engine into. And video editor I find lacking in usability and things vs a real video editor.

Arsenal. <=== Make that thing happen with Amethyst. I’m quite sure that doing it that way vs doing your own editor and stuff from scratch really helps the usability and “beginner” friendly side. But that is just my thought. Not a magic code wizard so do not know the feasibility of things. Just know that Arsenal grinned to a stop due to Amethyst not being there and help it to come of the drawing board. So? Be there pretty please and make that stuff happen! I’m sure advances in Arsenal helps Amethyst to develop. If there is difficult to do stuff in Arsenal improving Amethyst (being the core) is a given no?

#23

First, I think there’s a real chance I’m about to kick a hornet’s nest, so I’m sorry in advance if this steps on any toes or offends anyone. There is clearly a lot of love and thoughtfulness invested in project - both in terms of technical/code and community.

Above all, I think the most important asset that this project has produced has been a community… I hope this is preserved regardless of how the technical side of the project goes.

I think the biggest issue is a lack of focus. Even commercial engines generally pick ONE of these three categories and focus on it:

  • Full-sized Native Games (i.e. PC, full-fat consoles): These use to-the-metal rendering APIs, GBs of memory, and multiple threads. Battery life and asset sizes are not much of a concern.
  • Mobile Titles (i.e. iOS, android, handheld consoles): These can use multiple threads, but usually have to deal with a less modern rendering API (or buggy implementations/inconsistent support), more limited memory, and assets need to be right-sized to be appropriate for a small screen-size and not wasteful/slow to download.
  • Web: These are using very limited resources, immature and poorly supported standards, and need to be even more compact than mobile games.

Trying to target all three of these completely different environments seems out-of-scope for a project with amethyst’s resources. In particular, I believe targeting WASM brings too many compromises, which would lead to a niche product that is only good at games for a very specific environment. (Limited GPU/memory/CPU resource access, web standards that lag the state-of-the-art, etc.)

The legion migration either needs to be completed or cancelled. I think it is a good change to make as long as there is consensus in the community to make the change. I don’t see this as a matter of performance, but as a matter of being more user-friendly, more ffi-friendly, and being designed with streaming data in/out of the world in mind.

I’m hesitant to bring up the editor/UI… partly because whatever I say will likely be controversial and because the first issue - choosing a target - is way more important. First, there needs to be a decision on if there is going to be an editor or not. Designing an engine around having an editor but not actually having one is going to cause problems. If there is going to be an editor, it needs to be easy and low-friction to build new tools into the editor/UI. That means the implementor should be able to write their feature in rust, and the editor tools for that feature in rust. If the contributor has to be familiar with web technologies, this vastly reduces the number of people who can easily contribute improvements to the editing experience. It should be based on a stable and proven API like dear imGUI. The current approach brings a lot of complexity to an already high-risk component and in my opinion will make contributing to the project more difficult.

I think all of these issues have led to the technical aspects of this project stalling out. Every project has a “risk-budget” and these issues have clearly led to that budget being exceeded. Even a funded commercial game studio would struggle with this level of technical risk.

Regardless of where the project goes from a technical side:

  • I think the amethyst community has been and continues to be a huge asset to the rust game development community. While I personally have chosen not to use amethyst “the engine” for some of the above reasons - this community has produced and continues to maintain many crates that are important to the ecosystem. Many of these crates start out as single-person projects, and having a place to transition them as they mature is incredibly important and valuable to the broader rust community.
  • Additionally, this community has provided a path for both new-to-rust game developers and new-to-gamedev rust developers. It’s not a perfect path - but in this case I think the effort and care from the community that can be clearly felt is more important than the result.
6 Likes
#24

Really amethyst should not have a editor. And not cater to Mobile Titles. I mean if the engine is efficient and not very taxing to begin with it, more or less is will be catered to the mobile already. Just not purpose made like some more out of the box stuff out there. (Unity) If someone care enough about making a good mobile game they are going to seek out Amethyst or something powerful low resource. If they are lazy they go with something easy instead that just works. :laughing:

If focus is on high end and Web I think everyone is a winner. As making a powerful Web engine is not a bad fit for Rust and WASAM or what it is called. (not interested in Web so not any good at it) I rather think that Mobile = Web. The first Iphone was meant to not have apps outside of calculator and such. They imaged stuff being web based and not application based.

Amethyst is a really bad search therm btw. (are we games yet => I get to Amethyst)
I really think Amethyst should be used to build something on top of. And by that making categories possible and still keep focus on a a single thing. Also it can boost the contribution and advancement all around if one project is about the frame work and the other on a game engine.

If it is made to be a big tasty framework for stuff to be built into something like a Godot/Arsenal, yet leaving room for someone to build there own game from a nicely product ready framework instead of everyone running off reinventing the wheel. Profit? If the framework can not make a game engine possible then what use is the framework. And what is the point of working on something like a editor if all someone wants is to not role there own Vulkan and Web stuff. Amethyst should make stuff work cross platform and fast. The “magic” if you will. And something else should be using it to make a game engine that then is used to make games!
FR = ENGINE = Game!

Hits all the right tick boxes making amethyst not a big unnecessary big pile 1 fit all thing. Better to branch and use Amethyst as a base building block for specific purpose jobs. (that being FW or full blown ENGINE.)

No?

Have a nice day!

1 Like
(Thorlucas) #25

Also a newbie at Amethyst — posting this here not to be a dick, but because I care about this project and I would like to see it viable for use in game dev soon:

Amethyst has some really strong points. It’s biggest selling point in my opinion is the ECS system. It’s a delight to work with most of the time, and it feels good to develop in. Whenever I write code in amethyst it feels tight and well structured. That’s a very, very good feeling for any programmer to have, and it really instills confidence.

The biggest downside is unfortunately rendering, and it’s the reason I just can’t use amethyst for any project I hope to release (even if just a game jam). With a lack of a variety of built in rendering components (basically we just have sprite and ui?) it needs to be as easy as possible to implement new rendering components. Unfortunately, its ridiculously difficult.

I implemented a custom shader and it was a nightmare. I had to define a new rendering pass, copy a ton of boilerplate code I didn’t quite understand, build up my own tris, etc. It really needs to be as easy as possible to define custom shaders, and unfortunately right now in amethyst it’s about as low level as it can get. I never want to do it again, and to be honest it’s the main reason I haven’t used amethyst since.

Second is 2D hierarchical rendering, and from what I understand amethyst does not support this. For example, if I have some entity, and two child entities, these two child entities need to be drawn on top of the original entity in 2d. AFAIK right now you have to do this by explicitly setting the Z of the transform? That’s bad. It takes a long time to get this working, and that’s valuable time wasted for a developer doing a game jam or something. In fact, needing to specify a Z position in general is not great. Why don’t 2D objects use 2D transforms?

Third is the 3D currently looks like garbage. Again, I don’t mean to be mean, but honestly it does not look good. This NEEDS to be a high priority. Look at Evoli — its a super cool project, but it looks bad. Nobody will opt to use amethyst for 3D because it just does not look good.

There are several more grievances I have with amethyst, but those are probably the top 3 and they all come from the same category: rendering. The ECS right now is very good. Legion I’m sure would be great, but I don’t think it needs to be priority. WASM is fairly important because without it game jams are impossible (nobody, I mean NOBODY will download your game if it doesn’t run in the browser). But really, the top priority in my opinion should be sorting out the rendering engine to make it more end-user friendly.

4 Likes
(Martin) #26

Could I suggest working on documentation and examples? Please look at Amethyst from the viewpoint of a game author who wants to turn a game concept into a working game. They need to understand all of the game engine, from the point of view of how do I make Amethyst do this? And that documentation is just not there at the moment. Also, we need documentation and examples for all the features. Otherwise, if a feature is not documented, from the game author’s point of view it does not exist.

3 Likes