Yes, they ARE an essential type of 3D data, however, comparing with GIF and
JPG is invalid I believe. First, GIF and JPG are COMPRESSION standards for
images. True, the RECTILINEAR, and OCTREE types are a form of compression,
but a) they wouldn't have to be included if it were too difficult for a
browser to deal with, and b) I don't think they compare in difficulty with
GIF or JPG encoding whatsoever. Second, VRML is already able to deal with
images, so saying that voxels should not be a part of the standard because
they are more related to images doesn't make sense.
> a) Voxels can benefit highly from fourier (or wavelet) encodings such as
> jpg. b) Voxels can (and should) be viewed with a variety of associated
> transparency and colormaps, ideally manipulable on the fly. c)
> Manipulable planar and spherical slices through voxel space are another
> useful analysis tool. None of these concepts or techniques fit
> comfortably in VRML, but should be derigeur for a robust voxel viewer.
If I am getting the point of most people's comments on putting in a voxel
node is not concern for what people would want to do at the high end of
voxel manipulation, but making sure that there existed a default method for
dealing with them for browsers that did not want to implement the full
voxel toolkit. I think the worry is that someone browsing with a
pentium might come across a world with a voxel node in it, and have to
stare at an hourglass for two hours before it fell over. As for (a) I
believe that most browsers have the ability the deal with GIF images as
textures, and my guess is that they will be gaining JPG functionality if
they don't already have them. (b and c) Yes, the idea is that there will
be one or two browsers which would fully implement a voxel toolkit. To say
that they would not fit comfortably in VRML is not true either. Voxels can
be dealt with just like any other 3D object, and VRML is a 3D spec. To not
include them pushes a not insignificant segment of the 3D rendering users
to another standard instead of VRML.
> On the other hand, the most convincing argument for their inclusion is
> the potential for combining geometry (VRML) with fields (voxels) in a
> single scene.
YES! Although I believe that fields are already implemented in VRML, they
just happen to be missing that 3rd dimension. I think that the
possibilities of this combination are what makes this node so interesting,
not to mention the odd cool effect you could have when behaviours are
implemented ;-)
Regards,
tim
---------------------------------------------
Timothy Ritchey [email protected]
Jesus College tel: (01)223 576-822
Cambridge fax: (01)223 576-822
CB5 8BL
England