Originally posted by BIGPoJo
Yeah but when you add some overlapping patterns things start to look more random. You could use this tech to create huge terrains with realistic features. For NPCs you could still use polygons to do the hard work. To say that it will not be used is almost laughable.

Nobodt is saying it won't be used. Just that it's not unlimited, and it's not currently going to give you better results that a polygonal engine (in an actual game).

Nobody is saying that voxel technology is not real. Just that his claims about it are hyperbole.

Originally posted by Uncinus

Originally posted by Limbo
Yep but I realise is isnt really an issue now since on the vast majority of modern CPUs mult/divs are not so expensive. I don't know why he made this point - seems to add another argument that his tech or him being in the past. (It might have been when he first started working on the tech.)

I think he was actually talking about ray tracing against an arbitrary polygonal soup, vs ray casting into a octree. Ray tracing uses more math

Let's assume his world his world is made up of instanced octrees he still has to intersect a ray with the tree.
Since the distance to the tree is not a power of 2^n he could use a line stepping alg like you say. or he has to use multiplies to find where the ray hits the tree. (I know which one I would choose)

Speculating (not really thought about it)
What happens if he doesn't do it that way, he collects all the trees in the view somehow and figures out what nodes he can throw away by some kind of sieve?

Then he does the ray calcs based on the nodes per pixel (He already knows the tree from the sieve just needs to compute the ray hitting the solid cube.)


.

Originally posted by flexy123

And I doubt that ATI or Nvidia are going to make something just to accommodate this...

ROFL!

They will welcome this! NEW HARDWARE! Everyone will run and want one, and YOU know you would want one too More money for NV and AMD... its NOT their goal that people are sitting on outdated hardware for years and years.

Your right...
Would be probably be a year though at least.

Originally posted by flexy123

And I doubt that ATI or Nvidia are going to make something just to accommodate this...

ROFL!

They will welcome this! NEW HARDWARE! Everyone will run and want one, and YOU know you would want one too More money for NV and AMD... its NOT their goal that people are sitting on outdated hardware for years and years.

nVidia and AMD contribute a great deal of research in the field of computer graphics. Of course they want people to buy their latest hardware, that's their business model. No one's forcing you to shell out, but if you want to run the latest, cutting edge algorithms, you'll have to buy the latest, cutting edge hardware. Why does everything have to have to be some big, grand conspiracy?

Originally posted by Limbo

Let's assume his world his world is made up of instanced octrees he still has to intersect a ray with the tree.
Since the distance to the tree is not a power of 2^n he could use a line stepping alg like you say. or he has to use multiplies to find where the ray hits the tree. (I know which one I would choose)

Speculating (not really thought about it)
What happens if he doesn't do it that way, he collects all the trees in the view somehow and figures out what nodes he can throw away by some kind of sieve?

Then he does the ray calcs based on the nodes per pixel (He already knows the tree from the sieve just needs to compute the ray hitting the solid cube.)

You could build a world from a collection of arbitrarily scaled SVOs. But then you no longer have the single render per pixel, as they would overlap each other, and you've have different rays in local space.

just out of curiosity but couldn't this be an example of people not thinking 3 dimensionally (as far as things like size and direction are concerned)? if the level of detail is atomic, perhaps the original rendering is extremely small in size. for example, you may create a size standard for the rest of your game world, by creating a standard model of a tree, which will take up x amount of space of the screen and around which you construct the rest of your height models. perhaps the height models he's using are VERRRRRRRRY small because he's working with atoms not polygons and he's developed a way for the very small models to be magnified, if you will, to whatever the screen resolution is? does that make sense?

i do believe he said this entire idea came out of his work in medical software, for depicting anatomical models of the brain and so forth, in extremely high detail. let's say one of the things he had to do was depict molecular structures in detail, with features like an infinity zoom, from macro to micro level. magnification and so forth, are all part of using medical equipment

Originally posted by undo
just out of curiosity but couldn't this be an example of people not thinking 3 dimensionally (as far as things like size and direction are concerned)? if the level of detail is atomic, perhaps the original rendering is extremely small in size. for example, you may create a size standard for the rest of your game world, by creating a standard model of a tree, which will take up x amount of space of the screen and around which you construct the rest of your height models. perhaps the height models he's using are VERRRRRRRRY small because he's working with atoms not polygons and he's developed a way for the very small models to be magnified, if you will, to whatever the screen resolution is? does that make sense?

Not really. Units are arbitrary in computer models. So if some things is 1x1 that could be 1 meter or 1 nanometer.

Atoms here just means indivisible units, not literal representations of atoms.

i'm thinking all software today, owes its existence to software that came before it, so saying new hybrids of older tech is an example of stealing other people's ideas is perhaps pushing the envelope. pretty much everything today is beholden to something else, yesterday

Originally posted by undo
i'm thinking all software today, owes its existence to software that came before it, so saying new hybrids of older tech is an example of stealing other people's ideas is perhaps pushing the envelope. pretty much everything today is beholden to something else, yesterday

To some degree yes, but there's a big difference between "we've invented a rendering engine that's 100,000 times better than everything else", and "we've modified a standard voxel rendering algorithm to handle instances, but it's still not really practical as a game engine, or even really all that different from other people's voxel engines, and Id Software probably have a better version anyway, but we want more funding"

Originally posted by Uncinus

Originally posted by Limbo

Let's assume his world his world is made up of instanced octrees he still has to intersect a ray with the tree.
Since the distance to the tree is not a power of 2^n he could use a line stepping alg like you say. or he has to use multiplies to find where the ray hits the tree. (I know which one I would choose)

Speculating (not really thought about it)
What happens if he doesn't do it that way, he collects all the trees in the view somehow and figures out what nodes he can throw away by some kind of sieve?

Then he does the ray calcs based on the nodes per pixel (He already knows the tree from the sieve just needs to compute the ray hitting the solid cube.)

You could build a world from a collection of arbitrarily scaled SVOs. But then you no longer have the single render per pixel, as they would overlap each other, and you've have different rays in local space.

edit on 4-8-2011 by Uncinus because: (no reason given)

but this is how he says it is done in his videos. The storage space would be impossible otherwise and Notch would be right. Since he is overlapping trees how does he pick the right pixel as he claims.
his would contradict the one point per screen pixel that he claims thus

He is either using one tile per space block OR as he says in his grains of dust comment each grain of dust is
an atom etc. (In which case overlappying) Since 2 trees can occupy the same space it is undefined what he picks so he would ahve to sample 2 or more together.

If they figured out a technology to create infinitely deep RPG stories I'd be very happy.

Or a bit of time travel to go back a couple of years.