Ashley's Forum Posts

  • Hmm... I don't know. Try running it from the official Microsoft website?

  • Have you tried this on the latest build, 0.99.42? If it still happens, report it to the tracker again.

  • Originally it was intended to allow organisations of complex animations, such as:

    Stopped

    Stopped (with pistol)

    Stopped (with machine gun)

    Walking

    Walking (with pistol)

    Walking (with machine gun)

    In this case each weapon-state would be a subanimation to Stopped, Walking etc...

    I guess it's too complicated/not useful enough though, since nobody seems to use them. I guess we'll remove it in Construct 2.

  • If the timescale is 0, then timedelta is 0 and it's impossible to retrieve the unscaled timedelta - unless you use my method

  • Yes, obviously expressions don't compute themselves magically out of thin air with no cost, but I think compared to running conditions the cost is small.

  • It's perfectly plausible to custom code independent timescales for objects - even jumping - you just have to adjust the gravity accordingly as well. Gravity is a uniform acceleration applied to all objects, and acceleration is dependent on the timescale. If you're read the article on timedelta, you'll know that adjusting a speed by an acceleration involves speed + acceleration * timedelta.

    The main reason timescale is global is that it was easy to code Per-object timescales is possible, but would have to wait until Construct 2.

  • Everything in Construct should be 1-based (loops, lists, arrays etc) except where manually specified (eg. you can still specify a For loop to run from 0 to 9, it's not going to change that). If you find anything operating as 0-based, report it as a bug - obviously we don't want a mixed indexing standard!

  • Can you explain better for those of us who haven't tried Opera or its notes feature?

  • Yep, this has annoyed me before too, it's Davo's area to fix though.

  • Try Construct 3

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

    Try Now Construct 3 users don't see these ads

  • Maybe the DirectX display inteferes with it. I had to tweak the other window controls (buttons, edits etc) to get them to show right even in windowed mode.

  • Anything which will reasonably fit in memory. There is no predefined limit.

    You're probably creating several hundred thousand array elements (what exactly is the display width/height?) which I definitely would not recommend. Why do you need data per-pixel?

    Using lots of memory generally does not kill the FPS - are you using events to loop over the array every tick? That's what would kill the framerate.

  • Yes, this is yet another instance of someone being confused by Microsoft calling certain DirectX 9 updates "optional" and not including them in the default install.

    Even if you have DirectX 10, you might not have the latest DirectX 9. They're updated separately. I know, it's mad, and drives me mad too! Follow the instructions to run the DirectX update like it says and you'll get the "optional" components installed (which includes those DLL files you mentioned).

  • It doesn't have per-pixel collision for text objects. It uses the rectangle.

    You're using the wrong object. Use the File object - it has an "execute file" action which is the one you should be using.

  • If you want precalculated you could do a lookup within Construct, with an array

    That'd be much slower due to the overhead of running events. A sin or cos call can usually complete in under 200 cycles, which is about the overhead of running an event that calls an action that does nothing (let alone computing sin or cos).

    Yeah, it'll be faster for those who don't care of too much precision.

    Actually I doubt there will be any performance gains at all. Since performance improvements in 0.99, sin and cos don't appear as significant CPU users in profiles of the runtime, therefore there is little (read: none) gain to be had from further optimisation here.

    Did anyone notice performance improvements in 0.99? They substantially improved the runtime performance, and I don't think anyone mentioned they saw a difference just yet, so I'm not sure anyone would actually notice even if I replaced sin and cos with approximations

  • If I understand correctly, you just want an attribute that undoes 'solid' and makes its opaque area count as non-solid space?

    I'm not sure using hundreds of anti-solid sprites for destructible terrain is a good idea performance-wise - it might be better just to use lucid's suggestion, especially given it could be tricky to program an antisolid attribute.