Planet Chili

Unfortunately, it seems that all the transposing and dot products needed to multiply matrices by vectors is still slower. I was able to simplify a lot of the transformation math by not using matrices so I'm not surprised it was slower with matrices. But if I recall correctly the rotations suffer fr...

Sorry for the late reply, been very busy last few days. Just tried your code and it does seem to boost the frame rate by at least 4 or 5 frames. Imagine how much faster it could be if we could skip entire groups of pixels at the same time or if we had an algorithm like the one Euclideon has. I was a...

Hmm well that's odd, I don't seem to be getting any measurable change in frame rate with your code, but I'm seeing the effects you described when the matchbox is too close. Are you sure the increase you're measuring isn't simply due to being too close to the matchbox and having lots of the points ig...

At that point you don't have to transform it into view and screen space and can skip filling in or interpolating the space between points on the back facing points when the object is close to the focal point. Very nice, I'll have to give it a go later, I'm on my linux system at the moment. I'm stil...

Anyway, the point I wanted to make is this. Basically, what you are doing is turning every texture point into a vertex. You apply vector offsets for transformations from model to world to screen space on each point. Once they are transformed, you do per pixel lighting. I'm not sure if using matrix ...

It is a hybrid system though, so like I said it's not using raytracing for everything in the scene. That's the beauty of hybrid ray tracing, it only needs to send out rays from the surfaces visible to the camera but it still produces high quality lighting on par with traditional ray tracing. There ...

Ultimately, lighting is still the biggest obstacle in realistic games. I know that wasn't ever your point, just went off on a tangent. The problem with trying to do things like shadows in a polygon engine is that you have to use all sorts of hacks to make it work efficiently and look good, and it s...

From what I remember about the hints of explanation with Euclidean, is they take the point cloud data, sort or index it. Once sorted, you should be able to mathematically calculate which points correspond to which pixels on screen. So as far as levels of detail, it wouldn't be needed. The closer yo...

Actually now that I think about it, the description I gave for this engine is slightly misleading, probably because it's been a while since I actually wrote most of this code. When it loops through each point in the scene the point is transformed to camera space so that we can easily map it onto a X...

Hmm, actually there is a closed source project called Unlimited Detail or Infinite Detail, by Euclidean. Yeah I've been following that project for a while now, their lookup algorithm certainly is impressive. Their main problem when it comes to games is that they're trying to use 3D objects scanned ...