Euclideon Geoverse

Every so often I see an update on voxel technology. Jo Yardley caught a new update from Euclideon. I first saw a demo of Euclieon tech a year or two ago. At the time it was very impressive. Today the recent announcement of Geoverse is even more impressive.

About a month ago I mentioned this tech in relation to Philip Rosedale’s High-Fidelity project. It is supposed to be using the voxel tech Euclideon is developing. (See: High Fidelity News) The comments in this new video about lag may be the reason that Philip’s people think they can build a no lag virtual world.

Geoverse would seem to suggest it is possible.



7 thoughts on “Euclideon Geoverse

  1. Euclideon isn’t using voxels. It’s making use of “point clouds”, which is to say that it packs many different tiny points into a given density that it would be difficult to tell the difference between a surface made up of points vs. a polygonal surface.

    High Fidelity uses sparse voxel octrees, which is to say 3D pixels that can be of variable size. You literally have blocks that make up the world.

    There can be some degree of confusion between the two since at their most basic levels, they can be used for the same general purposes. Voxels however require less data to constitute a surface than, say, Euclideon Geoverse which requires a massive data set as input.

    Euclideon was declared to be many things back when it was unveiled; some people were claiming it was a hoax, others were claiming that it was voxels. A colleague of mine at time correctly deduced that was it was just densely packed point clouds, and its overall practicality now days in video games and simulations is pretty low. There’s many reasons for this, but the biggest problem that Euclideon was trying to solve when it was initially demo’d was the issue of detail in video games. That has since become a non-issue as GPUs move more and more into tessellation, and efficient voxel rendering starts to gain a bigger presence in the industry. It’s gotten to a point where point clouds are only really useful for storing arbitrary data that can be used in the rendering pipeline for techniques like global illumination that require large sets of data; and even then voxels have largely displaced them for that as well (Unreal Engine 4 makes a good example of this – their realtime GI and reflections are all done with voxels).

    So sure, you could say it’s impressive that they’re able to push a large set of data to the screen at once. But with the limits Euclideon, and to some vague degree point clouds in general, has (no realtime lighting, shading is all pre-baked and can’t be updated) it’s pretty far from practical for anything but the geospatial industry where this isn’t a problem. There’s a number of good reasons why the game industry stopped using point cloud visual representations of objects years ago.

  2. (I might be double posting here… not sure, delete this if first one came through.)

    Glad Geenz came in and cleared things up. (So I don’t have to. lol)

    A lot of developers are staying away from Euclideon because they had been trying to twist things around when it come to such technology where it’s mostly misleading. Thus this whole “hoax / scam” thing came about. Generally speaking Euclideon are completely ridiculous, trying to pass themselves off as true leading developers for voxel tech. Only causing more embarrassments by getting all high on hope by using raw point cloud data. With no real ways to clean up the gaps and holes and other graphical problems in the same data.

    if you wanted to see some true marvel tech about voxel engines. You picked the worst of them all. :/
    Here are some of the best example of Voxel engine.
    Atomontage have been trying to make it work with artist input in 3D form brushing, physic, Polygon-to-Voxel interaction, and some 3D scan conversion. Another voxel sandbox game, which are very similar to Minecraft. Except for the terrain, he has developed them to have smooth / connecting round surfacing and more definded details than typical 1m^3 cube size.
    Another Minecraft clone… except this one is writen in C#, which demonstrate a speedy result in light shading, water volume, fast & further viewing distance and no lag terrain destruction. Basically what Minecraft would have been if it wasn’t for Java.

    As for High-Fidelity project, I think Philip is making the right choice to follow up on next “Second Life” platform. Voxel is the future.

    • Euclideon never claimed to be using voxels. That’s a general misconception that people tend to make about them. Their founder outright claims to be using point cloud data for it.

      But I’m not entirely convinced that voxels are the future. As a storage medium for more complex effects? Sure. For general static geometry? Yeah, it works for that too (marching cubes and transvoxel are both great examples for voxel to geometry conversion algorithms). But as a general purpose replacement, I’m not entirely convinced yet.

      • Thanks Geenz.

        Euclideon’s ability to filter large data sets and quickly load needed data still seems impressive.

  3. Point cloud data is just another way of saying Voxels. Both are tehcnically the same.
    And as for Voxels being used for games, there is a huge advantage using voxels. Try to look to Atomontage Engine. I have been following it since years and that guy made some nice progress on it. It goes beyond graphics, he also reached some nice goals with illumination and interativity with terrain and objects. Interactivy that we cant see in any other real time 3D engine at all.

Leave a Reply

Your email address will not be published. Required fields are marked *