Pablo Hugo Reda has found this old site dedicated to the 3D Cube Project (that’s an even older link sent by appc23, with more access to info), as well as a discussion on the net which looks strangely alike the actual exchanges around UD. There are an exe and a database, as well. I reproduce further some parts, taken from the two links provided by Pablo (only boldfaces by me):
3DCube is as far as I know, a completely different way to store and display complex 3D images. The building atoms of the 3D data structure are lines of pixels, not polygons, bitmaps, or voxels. As explained above, the main objective of the project was to create a system that would allow very complex, elaborate models, with hundreds of structures, which could take up to an entire CD-ROM but require only a small portion of the model to reside in RAM. A large portion of the challenge was coming up with data organization that would allow keeping the model on CD ROM but be capable of displaying the perspective of the entire model instantly at any time. This is possible since the high detail of the model is only needed for elements close to the view point origin. Almost no processing is needed to load the model or its parts from the disk (please notice how quickly the demo initializes). Therefore, the disk activity processing load related to tracking the view point movement is very small – much lower than required for playing a digital video for example.
The algorithm required to display the image at any angle from the view point is quite simple. No floating point calculations, trigonometric functions, or even division instructions are needed, and use of multiplication instructions is very limited. A simple custom hardware utilizing my method could render the image with the same ease as a video card hardware displays the bitmap stored in its video memory. […]
My rendering algorithm is essentially a DSP-type algorithm, working with 64-bit data, which generates the image scan-line by scan-line with operations being mostly adding of 64-bit data. If the 80×86 just had a few more registers, the entire rendering algorithm would use no temporary RAM data (just the CPU registers) and would render the entire image by reading the model and outputting the resulting scan-lines. The biggest current problem in implementing the algorithm now is the necessity to swap the temporary data in and out of the registers to memory.
3D Cube project was originated with the intention of making it a new generation 3D game engine, allowing unprecedented detail and complexity of “virtual worlds” created. After seeing the performance and model sophistication possible with this method, I realized that the possible applications of the method are far beyond just video games. […]
In addition to the above, 3D Cube could allow things like taking pictures of some scene from different directions, building a model out of it. 3D Cube storage method
It’s not a mistake, the text ends like this.
From the discussion, a comment by the 3D Cube creator, S.A. Janczewski:
Well, let us try to clarify the term voxel.
– If a “voxel” can have different color/attribute from each of 6 directions, is it still a voxel?
– If it is not cubical — can have sloped surfaces, is it still a voxel?
– If the six colors of a “voxel” are not at all stored together as a unit, is it still a voxel?
– If the data organization of a 3D engine does not have any kind of data structure that would store data through clear (x,y,z) – coordinate
granularity, is it still a voxel engine?
If you answer yes to all the above questions than my engine is a voxel engine but so is any polygon-based engine.
I hope that the above will make it clear that the reasons I did not use the term voxel anywhere were following:
– My method has absolutely nothing in common with commonly known voxel engines or storage techniques
– Since the term voxel is not clearly defined, using it would not contribute anything(rather than confusion) to the descriptions
– Some “voxel engine” techniques are patented — using the term could result in getting myself (without a reason) accused of “patent infringement.”
Please forgive me if you somehow interpreted my previous message as criticism of your input. I did however get accused quite a few time of ignorance (not by you) for not using the term voxel in my description and felt it was appropriate to respond to it.
What do you think?
UPDATE: Is maybe relevant for the discussion to state that the goal is to produce an open source variant of an UD like algorithm. As you can see, supposing of course that 3D Cube was indeed a precursor of UD, the problems are the same, i.e. a probably very good idea, with a lot of potential for cash and also a game changer for an industry. Communicating it means loosing an advantage, but not communicating it leads to disbelief. There is a third way, open source. Right, no direct money from it, but everybody benefits and new possibilities open. So, in case you are working on that, don’t be shy or secretive, that’s not the good idea. Share.