TiAm's Forum Posts

  • Yup, that's a big problem on android. If you want to help, go here, and star this bug:

  • Basically, this issue seems to have gone into limbo on google's end; they don't seem to have nailed down why decoding is taking so long, and there is no talk of trying to do what FF does (uses it's own decoder for ogg instead of the one built into android).

    An audio-rich android game is basically off-the-books for now, because the load times would be utterly insane.

    Ashley mentioned trying to integrate an ogg decoder into the C2 runtime; an asm.js implement should be nearly as fast as native, which would hopefully bring perf in line with firefox. But...there haven't been any updates on that.

    I'll drop a new msg on the bug report to see if there are any updates.

  • These are great. I hope Somebody is okay though; about 11 days ago he stopped posting and hasn't even visited the site.

  • Really cool project rojohound!

    Anyway, here's some errors I got:

    8 bit issue (seems to happen with spritesheets, but not single images. 24bit png works fine):

    *** EXCEPTION ***

    Failed to load image "images/enemyspawner-sheet0.png". Reason: PNG not supported

    : 8-bit only

    File "<string>", line 261, in <module>

    Press any key to continue . . .

    Also, anything with keyboard commands won't run. Maybe you could allow such projects to boot, even if the controls don't work yet?

    EDIT: curious about newts question. Does your framework support effects?

  • TiAm

    Unfortunately my PM inbox is empty ! i didn't get your email address.

    Here is mine:

    ashraf egyptoon .net (without spaces)

    and thanks for your help.

    That's odd, must've disappeared into the system. Anyway, sent you an email.

  • Egyptoon

    PM'd you my email. You can send me the capx if you want; if not, at least two arm builds with v10 crosswalk: one with WebGL renderer, one with Canvas2D. I'll give it a go and see what happens.

    BTW, your test device is a good match for mine; different hardware, but the performance should be very close (CPU a little slower, GPU about the same, screen rez the same).

    Your specs: http://www.gsmarena.com/samsung_i8190_g ... i-5033.php

    Mine: http://www.lg.com/us/cell-phones/lg-L41 ... ifications

    Here in the states, both would be considered 'low-end'; not impossible to develop for, but you do have to be more careful to optimize.

    FYI, crosswalk is basically a special version of chrome which is packaged up with our C2 exports to run as apps. Unfortunately, chrome is still primarily a browser, and sometimes 'regressions' sneak thru that don't have much effect on browsing, but hit games hard.

    It's getting better with time, and the C2 community has been very vocal on the chromium bug tracker. It helps that other major engines are starting to support webGL exports too (Unity, Unreal), which is bringing major developers into the same space as us.

    Ultimately, this is the tradeoff we live with: efficient, rapid development / tricky exporting and opts.

  • Ok, that is the version I DL'd off google play. Really nice graphics BTW. Runs good on my 'torture-test' low end android, an LG Ultimate 2.

    Do you have the crosswalk 10 version anywhere? There's really no good reason I can think of that a game this simple should be giving you any trouble with performance, except on really, really low end devices, or if something is wrong with your code.

    Do you have any logic that does dynamic reformating of the canvas size or resolution?

    Have you tried debugging with a cpu meter readout? If you do, use a spritefont, and only update it every 0.5 sec (IIRC that's as often as it changes).

    From what I've read, crosswalk 10 is definitely a step down in performance from v7, but your game should still be playable, and those results you are getting with WebGL are way out of whack (it should never, ever, be slower).

    FYI, the performance drop from cwalk v7 to v10 goes all the way back to chromium/blink base, so don't get too mad at Intel on this one.

  • Egyptoon

    What version is on Google Play right now, and how is it config'd? Crosswalk version? WebGL or Canvas2D?

  • If it's a problem in my project why it works just fine if I use Crosswalk 7 ?! Note that I'm testing on Galaxy S3 mini. in my country it's a good mobile phone and it's my target device.

    please check my game on Google play "Propaganda" with a brain picture on the icon, and tell me what to do.

    My Results:-

    Canvas 2d> Crosswalk 7 = 45 fps

    Webgl > Crosswalk 7 = 8 fps

    Canvas 2d > Crosswalk 10 = 8fps

    Webgl > Crosswalk 10 = 5 fps

    Thanks for your reply

    Those results are hard to believe; cwalk10 has certainly exhibited lower performance than 7, but I've never seen anyone report the kind of discrepancy you're seeing. Generally a 10-15% regression, not 500%, as with your canvas2d readings. Speaking of your C2D readings, something is very wrong there: you should not being getting faster framerates with canvas2d, which suggests your device may have a GPU that is blacklisted, and unable to run webGL properly.

    Are you sure you've exported the same project and built it with v7 and v10? What do you get if you preview your game over wifi with chrome or firefox (yes, there is firefox for android)? Have you managed to test on multiple devices, or have a friend test on their devices?

  • I'm really hoping we get support for google saved games at some point; it would be crazy awesome to have a reliable, device-agnostic, cloud based solution with tons of hack-proof storage.

    https://developers.google.com/games/ser ... savedgames

    http://android-developers.blogspot.com/ ... cloud.html

  • The dirty little secret with Apple devices is that their specs are often very underwhelming vs their cost; case in point, no iPad or iPhone (except for the iPad Air 2) has more than 1GB of memory, the same amount I have in my $20 android phone. Most decent androids have 2GB now, and flagships usually have 3GB.

    iOS doesn't have as bloated a footprint as android, which is part of why it runs so well, but that still leaves you with precious little memory. In combo with being a 1 layout game, you're playing with fire. What is your estimated footprint for your project in C2? If it's over 250MB, I'd be worried.

    — might be able to give some guidance too; I wonder how much memory Star Nomad uses?

  • Having actually used OBS a bit now, I can say that it really does work great, and is the first screen cap software I've tried (including fraps, which used to work) that is able to capture 60fps gameplay with dropping frames. That it encodes on the fly into h264 is even better. And while it's very powerful and flexible, it's dead easy to use and understand (often the weak point of open source).

    sqiddster

    You have an SSD and an i5/i7 right? Just go to 'encoding' and set the bitrate really high, like 25,000, 30,000 or 50,000. The quality will still be really, really good...enough for promo videos, and overkill for youtube. Better than capturing a microscopically sharper image with another program that lags and drops frames.

  • > We didn't enable Crosswalk 11 until recently in the XDK, so my guess is you are building it yourself via the command line.

    >

    >

    Wait, what? I just updated XDK, but I didn't see Cwalk 11 as an option. How do we build with it?

  • Try Construct 3

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

    Try Now Construct 3 users don't see these ads

  • IE11 does not support webAudio. There are workarounds using the basic audio capabilities of html5, but as you've found, performance is neither good nor predictable.

    The next IE/Spartan will have support for WebAudio, along with a host of other html5 standards M$ has so far ignored. It's unlikely that such features will be backported to IE11...though I'd love to be wrong about that.

  • Ashley

    Over on the chromium bug report for this issue, rtoy brought up the following issue:

    [quote:3ps4qm9u]One additional item that can cause slow decoding. If the sample rate of the encoded audio files does not match the sample rate of the AudioContext, the audio file is resampled to the sample rate of the context. A high quality sinc resampler is used for this and can consume quite a bit of CPU.

    So for those who reported slow decoding using the example URL, it would be good to know what the context sample rate was. (You can go to http://rtoy.github.io/webaudio-hacks/tests/osc.html which plays a tone and also displays the context sample rate.)

    Could you work up a new version of the audio decoding test with 44.1khz and 48khz samples, so that test takers can eval both?