pretty neat indeed! awesome work!

yeah, I got an invite to Google Wave, which I already had

so um... you have to add me to the wave so I con see it. I think.

that's all the mail I got.

"Is Writing" and "on character written conditions" do not fire for the last character.

Using a font object with an autostarting fade behavior seems to break the edit box

or am I just doing it wrong?

EDIT: it seems that "show text" does not create objects until the next frame, as I can't pick them within the same event... and that's not really useful (I want to start a fade behavior on them)

A little further googling yielded this gem:

http://channel9.msdn.com/forums/TechOff/22266-Side-by-side-screwup/

Wooops. Maybe this manifest was enabled when compiling Construct? seems like a MS screwup. Wouldn't surprise me, given how they don't bundle DX9 updates with DX10.... urgh.

EDIT: uh... I hadn't read the whole thing. I hate these deployment nightmares. There's a lot of info in that thread, but it seems its related to linking to this or that redistributable. Perhaps the redistributable didn't install correctly? Dunno, read through the whole thing and look for clues

EDIT2: very complete explanation and solution by codeproject: http://www.codeproject.com/KB/winsdk/sidebysideerror.aspx

As for actual depth sorting on a pixel level using normal or heightmaps, that would require a change in the Construct renderer I believe. Dave and Ash said they were thinking about doing this for C2 iirc.

Sweet. Yeah the main problem with the current renderer is that you cannot reference more than one texture from an object (foreground tex sampler) and there's the background that can be used as a second texture (background text sampler) but that doesn't always make sense. Also, the render target is always the framebuffer.

So if one could customise the shading pipeline and have objects that hold more than one texture per frame, one could do this kind of stuff (and more!)

Not an easy thing to accomplish, though.

Still, I'm excited about the posibilities seen on those videos.

the System object has an Angle difference expression that always works correctly. You could use that.

for punching holes in canvas, paste an object with the erase effect on it.

and, of course, then update the collision mask of canvas

funny, I did that but set the ratio to be a percent of the object's size. Found out that bounding box changes when the object is rotated (to contain it) and just left it there.

Pixel height and width worked fine for me though and I could define the disk using them. It just didn't match the object size sometimes and I figured it would be hard to use.

And the funniest thing is that I called it Disk too. It's always centered though, so yours is more complete. Cheers!

I'm pretty sure this is doable in Construct, but editing wouldn't be as easy and events to keep everything in synch (with moving objects and all) would be hell.

Also, there would be VRAM issues, since I imagine any game using this technique would have to stream textures in-game (doubt they you can fit a whole level's textures in VRAM at once).

I had not seen this technique before, only imagined it and it looks much much better than I had imagined. With all this data you could even do SSAO, as it relies on normals and depth and that's all in there already. Moving objects near others would look awesome.

Edit: lots of people doing it, apparently

try selecting the offending layers and checking the "force own texture" checkbox at the left panel.

I bring a laptop to my presentations.

Can't go about installing DX9 on campus hardware.

ROFL