Havok's Forum Posts

    Moot Yeah and to be frank the last thing I care about currently is making Browser games.

    Does that then mean that I'm not Construct's target audience?

    Looking at the Newgrounds competition and Ashleys comments in this thread it does seem that Construct is more of a "fill the void Flash left" and "one day all games will be played in a browser" movement rather than to be a game engine for the desktop / native phones etc.

    That might all be correct "someday".

    Anyway I'm still here looking at where it's going but I'm not going to lie, I'm also very keenly checking out Fusion 3 that seemed to have fixed most issues people had with mmf2 and fusion 2.5. I.e it's a complete re-write.

    I really love Construct, the quickness of getting something up and running, the cool behaviors, the excellent event system but I need to look at where I eventually want to publish too and if the games I make is in the scope of Construct.

    I appreciate all the benefits Chrome / html5 / Webgl brings too but it doesn't help we champion a technology and also rent the privilege to do so.

    I guess color me cautiously optimistic at this point in time.

    We can argue all we like about the tech and why it's this or that...

    If the game you want to build for your target platform does not perform as it should on the client machine you are dead in the water. End of story.

    The customer sure as hell doesn't care if it's HTML5, C++ or freakin Basic.

  • Thanks Guys, will check a bit later.

    > C3 runs in the browser, with all the benefits that brings, but also with it comes all the negatives. i.e Construct 3 won't be as fast in performance as Fusion 2.5-3 , GameMakerStudio 2, GameSalad, Stencyl, Unity.

    >

    That's not actually exactly right: in some cases the browser tech is very advanced and can exceed straightforward native code. For example Chrome has a sophisticated parallel rendering architecture, so all WebGL calls are forwarded to another thread and run in parallel while the next frame starts rendering. This essentially lifts the performance overhead of the graphics driver off to a different core, which can actually result in improved performance vs. a native engine that does not implement such parallel rendering. It's a difficult and complex feature to get right, and we certainly didn't try for the CC or C2 editors. So this aspect is probably faster because it's in a browser. On top of that we moved from OpenGL 1.1 in C2 to OpenGL ES 3 (WebGL 2) in C3, and C2 has a lot of JS <-> native transitions which is expensive performance wise (and can more than undo any performance gains in the native code), and C3 has no transition overhead because it's all in JS.

    So you can't just say "it's slower because it's in a browser". There's a lot of tradeoffs going on, and in some cases leveraging the browser tech brings improvements even over native code.

    I hear what you are saying Ashley and yes there are benefits.

    Real world usage looks different currently, although unfair since C3 is still in Beta.

    It's one thing running benchmark set up for specifics. It's another when a "Non Coder Game Developers" starts using or comparing the engines back to back and make the conclusions based on what they experience first hand.

    So far I'm feeling quite confident that it will be fast enough for my needs although I'm obviously not the only one with concern looking at some of the posts. I'm just wondering where the ceiling lies and if the memory usage are so high (Chrome + Content exported to a Runtime be it Desktop or Phone) that there would be a higher Minimum Requirement on end users to play our games.

  • https://tinyurl.com/lddg862 <-- link to small screen recording.

  • Try Construct 3

    Develop games in your browser. Powerful, performant & highly capable.

    Try Now Construct 3 users don't see these ads

  • Very cool!

    Keep it up.

    Delenne - I opened Kiwi Story in C3 with Chrome Canary and it only used around 300mb of memory. It might be worth checking your browser addons, some can add a lot of extra memory usage, especially ad blockers.

    Yes add ons can cause it (check Helper Objects on Mac in activity monitor - I've tested it, disabled adblock plus and https everywhere for instance, then the memory usage drops a bit). So now if someone is a Chrome user they will have to disable all addons etc to get the performance in Chrome. Not so bad if you use another browser for regular browsing etc, but irritating if you are a Chrome user. Unless get a beefier PC.

    Not an ideal situation but we are beating a dead horse here.

    C3 runs in the browser, with all the benefits that brings, but also with it comes all the negatives. i.e Construct 3 won't be as fast in performance as Fusion 2.5-3 , GameMakerStudio 2, GameSalad, Stencyl, Unity.

    The question one should ask what is fast enough for your specific game you are trying to make. I think for most it will be fine. I really hope so. I like Construct and got along with it better than the other applications. However bad performance is obviously a deal-breaker if your game doesn't run well on a platform even though you did optimize it etc.

    I suppose when you wrap it up so that Chrome runs on the desktop for the C3 editor it won't be submitted to the addons, ad blockers, cookie blockers etc and will get better.

  • I also don't see how limiting the events changes the focus of the bug squashing.

    Limiting the behaviors and such, yes that I can see as a way to focus the bugs that may occur.

    DiegoM Yeah I saw it didn't destroy and just carry on.

    Now of course with only about 290 or so objects on it stays at 60fps.

    I guess we know what the limitations are now for target platforms - safe to say no more than a 1000 objects on screen.

    I suppose a bit less for PC's with no dedicated GPU.

    I can confirm what Havok has said about the "Weather Demo". On my iMac it runs very badly, yet no problems in windows under parellels.

    Thanks Ethan, but mine runs badly under Parallels as well.

    Can anyone on a Windows machine run the weather demo and see if it runs smoothly? Don't even walk or move etc.

    Looks like it's the reflection eye candy.

    But then what does that mean? It's fast as long as you don't use effects?

    Running some https://threejs.org/ demos, things are fine. Fast and running well.

    I want to add (after TrollHunter's deleted comment) that I don't post criticism because I like to argue. I post here because I want people to change my mind. I want to discuss these issues with reasonable minded people who might have a better perspective on it. And by reasonable, I mean people that can see both sides of an issue. Not someone who is who is obviously trolling and has a vendetta against Scirra. We are not on the same side.

    It's not whether its HTML5 or not...its whether Construct 3 is fast enough or not.

    The editor is not snappy like C2 but it's not bad until you load demos, other than that though which is more concerning to me:

    Trying to run the "Weather demo" on my MacbookPro i7, 8gig ram, nVidia 650 gpu.

    I do nothing, just run it. It hits 12 fps average when it gets about 1500 objects. It fluctuates to 8fps, then to 3 fps then back up again a bit and so on.

    Running other game makers I get 50-60fps, running in a Parallels windows vm with half the system resources dedicated to it running physics and particles that without breaking a sweat until objects absolutely fills the screen.

    I'm really hoping this is a MacOS bug or because it's still in Beta.

  • Thanks for the feedback!

  • Was there a loose time-frame mentioned perhaps?

  • Yeah I hear you — You make a good point. I'm certainly not averse to learning. Just would make a useful time saving feature found in other engines. Just a pity that it does nt exist in Construct.

    When the time comes I'll have to dig in to these kinds of things.

    Yea I'm pretty sure at this point I'm on board as well.

    With the history of c2s updates, I'm confident c3 will be completely different over time with all the updates.

    ONLY concern was really exporting to consoles but according to their blog posts, they're looking into options for that.

    subs aren't a concern for me especially if it ensures that the team is staying afloat to keep updating c3.

    They're reworking the runtime and all. So, yea. Pretty happy with the future plans of it so far.

    count me in

    I read your concerns in the post that was then subsequently locked.

    Do you feel that C3 aims to fix the issues you had with performance and porting?

    I know Scirra is working to be able to port to UWP for XboxOne. Great imho!

    The issues you described is also my number 1 concern tbh. Even though I'm not near publishing anything.

    Great...you can port to all platforms, but not so great if the games won't run well.

    Like — I looked at Unity + PlayMaker and did a few demos with it, top down 3D shooter, suedo FPS shooter and a 2d level with some running and jumping etc. It's good and obviously gives you 2D and 3D options but yeah, it takes a lot longer. So pro's and cons.

    Fusion 3 stands out as the main competitor. Construct's event system and the community makes it a good place to make games with. It just feels better with a better UI etc.

    NotionGames Do you feel that this is realistic to do with C3 going forward. I would like to know what changed your mind back to staying since I'm evaluating it as well.

    I wouldn't like to get where you were and then blocked by the engine.

    I see The Next Penelope had similar issues, they decided to get it ported to C++.

    I wonder how C3 will do on the Switch later if at all possible?

    Ashley, is this an ok question to ask you? Where do you see Construct 3 in terms of mobile and console performance, say, at the end of 2017?