quad fills: bug? hardware issue?

[qubodup]

I suppose that sort of thing is inevitable...

It is possible by making the function automatically align the points in a ccw manner. I don't know if the costs are not worth it.

Perhaps it's a good thing to be limited to one way of drawing quads. Do you feel a limitation in the way right now?

[rude]

Ok, how about this. In 0.3.2, we'll change that function to draw the polygon no matter what, then we can benchmark 0.3.1 vs 0.3.2. Of course, I'll make sure the performance isn't completely horrible (unlikely) before release.

[mike]

So if anyone wants to make a benchmark program that does horrible things with polygons then you are free to do so...

[cag]

My answer:

Code: Select all

function quad_ignore_shit(type, x1, y1, x2, y2, x3, y3, x4, y4)
    love.graphics.quad(type, x1, y1, x2, y2, x3, y3, x4, y4)
    love.graphics.quad(type, x4, y4, x3, y3, x2, y2, x1, y1)
end

... Though for most applications on most modern setups, the doubled up quads are not going to affect performance much . It does show that if you're really that picky, you don't need the developers to give you a solution as you can always make your own.

I would be interested in making this benchmark program, but I actually want to start making this adventure game I've been thinking about...

EDIT: Ah, I see you said automatically align... that's just plain spoon-feeding. :/

[Merkoth]

Hmm... maybe I'm missing the point... but what's so hard about remembering that you have to draw the vertices (is that the plural of vertex?) in a given order?

[cag]

There's nothing hard about that, I would think.
._.

[rude]

Nice one cag, didn't even occur to me.

I was thinking about disabling culling, drawing the quad, then enable it again for each rectangle. We'll see what is faster in the next version.