i've been thinking that the best way to recreate reality in 3d would be to use transforms on multi dimentional primitive shapes in a complex number system. basically you would use most of the memory to contain an acurate representaion of the complex number the primitives would then use only a little memory. i guess it would be difficult because of the bit ness of computing. the complex number could be stored in binary tho. i'm just wondering if we have anywhere near the speed required to do this. it would sort of work fractorily aswell. i know fractals and ray tracing can take a while with many primitives, but the performance scales over time. i think from what i know the PS3 would be best suited for this task. the hard part would be creating artists tools so they could work in 3d but get a multi dimentional result. or you could just find the right math to do what you want. much like in fractals, and elite where they used the fenobacci series. i'm wondering what the closest example of this is that is availiable now really, or if there are papers dealing with similar ideas out there.
complex numbers, multi dimentional primitives, fractals
- Thread starter Danalys
- Start date
Either you're talking about the procedural generation of content / detail, or you don't have a clue and are just blabbering on.
Procedural Synthesis has been one of the features taunted in advance for the 360, but in reality very little (read: nothing I know of) has actually seen it in a shipping product, probably because of how it interferes with the rest of the renderer (tiling etc.).
Now explain the following to me: your idea of a "complex number", "performance scales over time", "multi-dimensional results".
kthxbye!
Procedural Synthesis has been one of the features taunted in advance for the 360, but in reality very little (read: nothing I know of) has actually seen it in a shipping product, probably because of how it interferes with the rest of the renderer (tiling etc.).
Now explain the following to me: your idea of a "complex number", "performance scales over time", "multi-dimensional results".
kthxbye!
I get it.
Complex numbers can be used to simulate matrices.
Basically you would calculate behaviours of renderers and use a scene with over 2^36 vertices and have them follow a behaviour.
Instead of using imaginary numbers you could use rotations. Under a rotation the inverse is proportional to the intersection of the integral, because of that you could easily invert the imaginary number thereby saving you an extra memory if you were to store the results in memory.
Technically you could do this to generate a painting and it has been done with only 100 lines of code but the algebra is so advanced that you'd need 3 years to do a single object.
Complex numbers can be used to simulate matrices.
Basically you would calculate behaviours of renderers and use a scene with over 2^36 vertices and have them follow a behaviour.
Instead of using imaginary numbers you could use rotations. Under a rotation the inverse is proportional to the intersection of the integral, because of that you could easily invert the imaginary number thereby saving you an extra memory if you were to store the results in memory.
Technically you could do this to generate a painting and it has been done with only 100 lines of code but the algebra is so advanced that you'd need 3 years to do a single object.
SoftwareGuy256
Newcomer
I'm guessing the 3rd and 1st post are authored by the same person because both make no freaking sense
Mate Kovacs
Newcomer
Are you sure about this? AFAIK the square roots of negative real numbers were, and still are, called imaginary numbers. But this doesn't mean that complex numbers (in general) were called that, IMO.K.I.L.E.R said:
Imaginary numbers were "discovered" as a byproduct of the cubic formula, AFAIK. Then they developed complex numbers so that imaginary and real numbers can be dealt with in the same framework (which has the same identities as real numbers alone).
EDIT: I know there were times when e.g. the number 2 wasn't considered a prime (because it's even), but 1 was. So it could be that you're right, I just haven't read anything that called complex numbers that, that's why I'm asking.
Last edited by a moderator:
oeLangOetan
Newcomer
K.I.L.E.R said:
Isn’t it, complex numbers consist of a real and an imaginary part (the "imaginary" name is arbitrary, invented by some French mathematicians)
edit:
rereading the first post, did you mean mathematicaly transfroms .. like laplace or fourier or Z .. becouse afaik everything in 3d is lineary, not periodicaly so that would make little sense and make things even harder.
Last edited by a moderator:
wel i've been looking into things and found that most fractals use an 2d plane imaginary number system to draw the fractal. so going into more dimentions than a 2d plane would probably lead to even more processing time required. fractals seem slow enough as it is.
the orginal idea is something like you can treat a moving object as a crossection in a four dimentional solid. taking that idea further you could probably get alot of effects alot easyier. buy doing calculations in multidimentional mathematical space and then taking an apropriate cross-section. anyway i haven't got it all figured out by any means. but i'm trying to learn more and combine the two in some way.
i've basically just been thinking about hyperspheres i pi and phi and the change in those dependant on shift through xyz and what ever the fouth dimentional coordinate would be. as well as the angle with in that. seems to fit with computer graphics because most of what is happening is trying to get a 2d view of 3d space. i've also been trying to combine phi curves and sheres in 3d images. gives results that look like body shapes.
the orginal idea is something like you can treat a moving object as a crossection in a four dimentional solid. taking that idea further you could probably get alot of effects alot easyier. buy doing calculations in multidimentional mathematical space and then taking an apropriate cross-section. anyway i haven't got it all figured out by any means. but i'm trying to learn more and combine the two in some way.
i've basically just been thinking about hyperspheres i pi and phi and the change in those dependant on shift through xyz and what ever the fouth dimentional coordinate would be. as well as the angle with in that. seems to fit with computer graphics because most of what is happening is trying to get a 2d view of 3d space. i've also been trying to combine phi curves and sheres in 3d images. gives results that look like body shapes.
Last edited by a moderator:
you have to have some grounding in theoretical physics or 3rd year degree level maths to understand it.
i'd be doing transforms from 8D shapes to 2D. an example i've heard of is that the wing shape of a jet is a hypersphere with a particular transform applied. basically it's a rotation in multidimentional space. but by taking the inverse the forces are calculated aswell.
No, Danalys, the problem is that you have yet to explain your technique at all.
Similar threads
