gles.js - a lightweight WebGL renderer for Android

Gianmichele Haxe looks awesome! I have been thinking about it for a while.. I'm gonna try out the Three.js Haxe port and if that can compile that to more than JS I am totally sold on Haxe myself <img src="{SMILIES_PATH}/icon_e_biggrin.gif" alt=":D" title="Very Happy">

Yeah, I have been playing with it for a while. Haxe in itself is pretty bare bone though. You're going to need a framework of some sort.

OpenFL is one of them but even though it mostly works, I don't really like the status of flux which it seems to be constantly in (Stencyl is suffering from that).

I'm using Luxe engine for now, which although is stil in alpha, is rock solid and with a super plan behind it (read editor and consoles too!).

Have a look here for more infos:

http://luxeengine.com/

http://snowkit.org

I'm really waiting for the day Scirra is going to embrace a technology like this.

Joining the C2 performance/rant

To put it simple, C2 is "Jack of all trades, master of none".

The good part is that more and more big companies are going for web apps (HTML5) ...The bad part is that no matter how much Ashley struggles, HTML5 will not even be close to native for at least 2-3 years ... Is still in "development". At least on mobiles where performance is weaker. While C2 has great HTML5 performance (simple games works decent even on 1GHz CPU and 512MB RAM), you can't compare it with native performance.

Like others said, a simple Crappy Fird clone takes 30MB and uses 20~40% CPU and still has some frame drops... Angry Bots (default demo of Unity) takes 38MB, uses around the same % of CPU, no frame drops.

Eli0s's Fancy Benchmark runs at 6~9 fps CPU 25~40% in Chrome, 8~9fps CPU 70~90% in FireFox on my Samsung Galaxy S4 while Epic Citadel runs at constant 60fps on default and 30~55fps at Ultra High Quality (rendered at 1080p), and Angry Bots mentioned above...

So yeah, "HTML5 has close to native performance" / sarcasm

Maybe the CPU part is close to native, but the graphics part is not even close. I've stressed my device to test Epic Citadel and Angry Bots at a lower framerate, and guess what, at 15~20fps they still look more fluid, with no stutter then C2 at 45~50fps

Just a quick remainder that Unity also started to support HTML5 and works quite well (on PC), on mobile is not optimized at the moment ... I know that the grass looks greener on the other side, but in most of cases it is greener

So yeah, the same old rant But all of us were aware (or should had been) when chosen to work with C2

Anywhere I can find a video of a mid-range (under $250 unlocked) android device spitting out 60fps/1080p/1,000s of sprites in the context of an actual C2 game exported with the current crosswalk stable(CW10)? Because really, I would love to believe this is possible, but I just don't. Then again, I've been wrong before...

I would like to see Flappy Bird clone in 60 fps, perfectly 100% smooth in Crosswalk:

3 sprites is not too much, I think...

Egyptoon i couldnt agree more, C2 or C3 needs a custom exporter for mobile.. There's no way around it, simple as that.

Right now perfomance is really bad on thirdparties, and lack of bugs control will ruin this in the long run.

If Scirra could find it in their heart to make it, im SURE it would be the absolute best game/app developing tool for mobiles, period.

Ashley Maybe you can start a crowfunding project to hire a talented team, for a custom exporter for mobiles?

ps. The Haxe project looks awesome! - Scirra should consider something like that!

Ashley: I know you're probably sick of me pointing to this, but I would really love for you to read these two articles and then tell me if this is not a technology worth considering

Given the number of people here criticising us for "depending on third parties too much", I am extremely wary of adding any additional technology dependencies in to the C2 workflow. Even if Haxe works pretty well, we could still get shafted by compiler bugs, missing features and unoptimised parts that kill the performance. In particular I think Stencyl got burned by some of these types of issues. I'd rather stick with HTML5 - it may not be perfect, but it's improving fast and has some huge players backing it, such as Microsoft and Google.

slowly replacing parts of the engine with asm.js code?

IMO this is a much better idea than writing native exporters. I do want to do this eventually, but it still means rewriting a big part of the runtime, and probably breaking compatibility with all existing plugins. So I want to do this eventually, but it would probably come post-C3 release.

Also regarding native exporters - one of the reasons C2 is $129 for all exporters and not $999 is that we only have one codebase we need to maintain. All that extra engineering work would probably push the price up to another level. And "all exporters for one price" is a really nice thing about our licensing IMO.

Yeah, I have been playing with it for a while. Haxe in itself is pretty bare bone though. You're going to need a framework of some sort.

Besides a rendering engine (like Three.js (I can't do WebGL...)) I would much prefer to code my own engine anyways

this might be something for people to read about: (why JS is slow - GUI mostly):

http://sealedabstract.com/rants/why-mob ... -are-slow/

Ashley: You should really think about making a Construct Pro Version for 999,- for all this people that want native exporters so badly!

And please, if somebody wanna argue now that Unity is free..look first the history of this company, for how long they sold their exporters for a good price, before they could afford giving it away for free...

— Games

Yes I'm ready to pay for that up to 2000$ because if i have the amazing and easy-to-use C2 editor and Native exporter together i can make AWESOME games and gain what i paid in just 2 months.

I can't believe that there is no one can understand the importance and necessity of making native exporters !

we are making GAMES, and games = performance in the 1st place.

No performance = no playing, this is it.

Wasting the CPU power and memory to run a HUGE buggy browser and inside this browser there is a game that needs the whole device power to run!

Again look at this example (flappy bird clone) VERY very simple game.

https://easyandroidcoding.wordpress.com ... gdx-part-1

I followed this tutorial and the result was : 2mb APK, the image quality is perfect, constant 60 fps. JUST PERFECT!

Someone tell me please how to achieve that using C2 ?! simply there is no way to do that.

the APK size will be at least 16.5mb, installed on +20mb, up to 70% cpu usage, frame rate = 35 fps, and the battery will die very soon!

NO COMMENT

BTW

I'm not a developer, I'm a cartoonist, animated film director and game designer, and i like C2 because it allows me to CREATE, no time to study coding from the beginning and I'm not even a developer.

I want to imagine, draw, design and play, i want to make my games without depending on a developer, it is very hard to find a developer who can understand animation , composition and directing as i do.

This is my job.

C2 developers job is to allow me to achieve that and they can, I'm sure they can, but they don't want to do.

Not sure how this flappy bird look like cause there is no image of it or whatsoever, but if that kind of games takes up 70% of cpu and runs only at 35 fps then obviously you are doing something wrong here.

Crosswalk adds ~17MB to the game (which is a bit idiotic and wasteful) but it's not that bad! Bare in mind that crosswalk work only with Android 4.0 and higher. And even if some devices support 4.0+ it does not mean they are great for games.

And yeah, I'm bit concerned about current situation with exporters too. It comes and goes with waves. Unfortunately there were more bad times then good times. Looks like when everything works as it should then it lasts for a couple of months and when it breaks it takes twice longer to fix it.

shinkan

I'm not doing anything wrong, we tested a flappy bird clone already, check this thread please

the results was horrible!

i switched to CocoonJS just to save my game on Google Play (OpenSSL issue with crosswalk 7 Alert) and it runs now at 60fps and the APK size is only 11mb

and I'm learning Unity to save my future.

Learning Unity isn't the first threat here, but it is the funniest.

My game should work perfectly over 6 FPS, but these lag spikes ...

Here's my threat, I am learning Corona SDK.

Just wanted to chime in on a couple of these to put them in perspective; and not so much to refute them.

Like others said, a simple Crappy Fird clone takes 30MB and uses 20~40% CPU and still has some frame drops... Angry Bots (default demo of Unity) takes 38MB, uses around the same % of CPU, no frame drops.

Unity has a 10mb over at the core. Which is about 7mb less than Crosswalk. Unfortunately there does seem to be some unwanted overheard on the browser aspects. I keep on saying it. But I truly believe Scirra should take Chrome xyz. And customize a version of that. Remove the DOM as a priority.

Eli0s's Fancy Benchmark runs at 6~9 fps CPU 25~40% in Chrome, 8~9fps CPU 70~90% in FireFox on my Samsung Galaxy S4 while Epic Citadel runs at constant 60fps on default and 30~55fps at Ultra High Quality (rendered at 1080p), and Angry Bots mentioned above...

So yeah, "HTML5 has close to native performance" / sarcasm

Fancy benchmark is nice. But in refer to my comment above. Scirra should just use a stripped verion of Chrome. that would solve a lot of the overhead issues.

Epic Citadel is gorgeous. However please note that 99% of Epic Citadel is

• static environments,
• quadtree collision detection,
• binary spacial tree drawing determination.

in comparison

• C2 poor performances demos are batches of moving objects. Check out Goo Create for a proper 3D static environment comparison
• C2 uses Cell collision detection. however don't let the name fool you. It's still brute force collision detection with in the cells.
• C2 probably uses the Cell's to determine drawing or does brute force.

This all comes down to NOT HTML5 or the browser, but just not well optimized critical performance pipeline of C2 engine. Which I have been requesting overhauls for 3 years now. It's why I requested focus on core malleable World Object with dynamic attached behaviours rather than C2 rigid Plugin system. This would allow for more flexibility to overhaull the choking points in C2.

Maybe the CPU part is close to native, but the graphics part is not even close. I've stressed my device to test Epic Citadel and Angry Bots at a lower framerate, and guess what, at 15~20fps they still look more fluid, with no stutter then C2 at 45~50fps

I have a Unity game on mobile where the start runs at 15 to 20fps to start on 2 year old devices. Unfortunately that's unoptimized due to time constraints. The game is very flappy birds like. 1 character moving to a tap. Though we 3 more animations on the screen.