Regarding Node Webkit

  • Ashley - Here is a capx

    dl.dropboxusercontent.com/u/7765312/Seam/Seamg.capx

    And here is the exported capx as Node-Webkit

    dl.dropboxusercontent.com/u/7765312/Seam/SeamNode.zip

    Here is what I get when I run this capx in PREVIEW and shrink the Chrome window. No seams. Looks just fine <img src="smileys/smiley16.gif" border="0" align="middle" />

    <img src="https://dl.dropboxusercontent.com/u/7765312/Seam/PreviewNoSeam.png" border="0" />

    However, when using Node-Webkit and shrinking (or enlarging) the window, I get seams.

    <img src="https://dl.dropboxusercontent.com/u/7765312/Seam/NodeWebkitSeam.png" border="0" />

    I would be less confused if the seams both appeared in the preview, AND node-webkit, but that isn't the case...

    Is this something to do with "Powers of"? If I colour these sprites in this capx to a solid colour, seams don't seem (no pun intended) to appear...

  • Nope, no preview for node-webkit...

    I'll whip a demo up tommorow or monday, so i'll reply here!

    Cheers!

  • sqiddster - as far as I understand, in full-screen mode, the graphics card only has to draw that one application's window, it doesn't have to consider any of the others, so that means less work for it.

  • Arima ah, I see. JohnnySheffield does node-webkit support proper fullscreen?

  • If not, they should definitely add it as a feature.

  • The browser -> fullscreen action works in node-webkit. Looks awesome to have my C2 games filling the whole screen.

  • That's not true full screen mode though, it's just a borderless window that takes up the whole screen.

  • Isn't that what "true" fullscreen is anyway? Just a fullscreen size window? Native engines might boost the process priority and change the monitor resolution, but they still would do that.

    Jase00 - that's a perfectly typical case where you should be using Tiled Background! One tiled background can often render 30x faster than a grid of 30 sprites. See performance tips. Not only that but tiled backgrounds are much better at rendering seamlessly. If you still want to line up sprites, pretty much the only way to get it seamless is to use Point sampling and letterbox integer fullscreen mode.

  • Try Construct 3

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

    Try Now Construct 3 users don't see these ads
  • From what I've read (I'm no expert on it, though), 'true' fullscreen boosts fps a bit because somehow the graphics card can skip some steps/ignore every other application and throw the rendered image directly to the monitor - with a fullscreen borderless window it still has to check if any other windows are overlapping it, ui elements, os stuff, etc - some also theorize that the graphics card could be drawing the desktop and then the window on top of it, meaning it hits the pixel fill rate and such (I'm honestly not sure of the real reasons, again, I'm no expert and what I've read could be wrong as they were from people theorizing/discussing it on the internet, so the specific reasons could be incorrect).

    However, it's not a placebo effect, as I've seen it on my computer as well, playing the new XCOM with a fullscreen borderless window results in a choppy framerate, and using the 'actual' fullscreen setting makes it much, much smoother - easily perceptibly so. It could just be the difference of tipping it a bit over the point where it can't handle 60fps, but it shows that it can make a noticeable effect, and we're talking about low-end laptops here where it's likely to be a problem.

    I don't know if it's possible or not, but it would be helpful if it is.

    Edit: From gaming.stackexchange.com/questions/107028/is-there-a-difference-between-running-games-in-windowed-or-fullscreen-mode

    "When an application runs in fullscreen mode, it runs in "exclusive mode". That means it has full and direct control over the screen output.

    But when it runs in window mode, it needs to send its output to the window manager (windows explorer) which then manages where on the screen that output is drawn. This takes some additional performance. The performance penalty, however, is greatly reduced in newer version of windows."

    That might explain why it's an issue on my machine, as it's running vista. In newer versions of windows it might be imperceptible.

  • That low-res renderer that stretches to fullscreen sounds like a fantastic idea. I'm glad more people are also concerned about the fullscreen performance on middle-lowerend computers.

  • Jase00 - that's a perfectly typical case where you should be using Tiled Background! One tiled background can often render 30x faster than a grid of 30 sprites. See performance tips. Not only that but tiled backgrounds are much better at rendering seamlessly. If you still want to line up sprites, pretty much the only way to get it seamless is to use Point sampling and letterbox integer fullscreen mode.

    Ashley - Ah the tile has a few different frames and I used separate sprites so that I could use these extra frames, but after reading the performance tips link, I never realised how badly sprites actually butcher performance. Thanks for the advice!

    I still wonder though, howcome with lined up sprites, on preview it looks fine, but node gets seams?

  • sqiddster

    exe

    Here's a really rudimentary demo of the set window width/height function in node-webkit. Just type some values in the texboxes and hit ok. (You can even try with 1/1, for the ultimate fun.)

    To try it yourself you'll need my node-webkit gui plugin from here.

    Cheers!

  • JohnnySheffield really cool! thanks!

  • Ahh, i forgot, regarding full screen and node-webkit...

    Not sure, but i think it's safe to presume that it uses the same method of "fullscreening" as chromium.

    If anybody wants a detailed answer, your safest bet is to ask on the node-webkit mailing list, guys are pretty active so you'll probably get an response!

    Cheers!

  • Guys, regarding node-webkit and "real" full screen performance, there was an interesting conversation on the nw mailing list;

    groups.google.com/forum

Jump to:
Active Users
There are 1 visitors browsing this topic (0 users and 1 guests)