I don't blame you; without a game as a side project to keep the project interesting, I think i'd be insane by now!

The topic is "Object Deformation" (currently using voxel tech to extract a skeleton for animation with some other nice features - i've got a method working but trying to get rid of all parameters for robustness.. kinda stressful), but later planning to focus on other types of deformation and have some ideas using voxels

Can I ask how you encoded the volume? I literally just found this a few hours ago so haven't had much time to look/search around yet.

EDIT: Ah the comments in Volume.h are more than suffice to my needs

Sorry for the slow reply, I was away for the weekend...

Yes, the Volume class is actually one of the few which has good documentation so hopefully you could figure out how it is storing the data. It's going to be changed fairly soon though, to get much better data compression.

Yeah it'd be nice to have a decent RLE in the future with a 2d (or 3d?)-paged-grid (north-south encoded?) and a Frustum uncompressed/block compressed (both?) central section.

I guess in the 2d case the files would be like:

0,0.page
0,1.page
0,-1.page
0,-2.page
1,0.page

..etc..

But yeah, its a bit weird having to balance the parameters knowing when to page/how big to make the chunk size.. taking into account hard drive speed, and processor utilization.. ouch!

Yep, I imagine I'll compress blocks based on a Least Recently Used scheme, and just allow user code to control settings like how many compressed/uncompressed blocks are available at a time, and how big they should be, etc. Well actually the blocks size is already controlable for exactly these kind of reasons.

Sounds good. Looking forward to idling here and seeing it progress

Offtopic/for fun: attached is a picture from my old voxelizer if interested - but I won't get back to programming fun things until after May.. ah well.

Oh, very nice. I did write a voxelizer in the past (the castle model in the movie on the Thermite3D front page was voxelised from a mesh) but it's been abandoned for a couple of years and won't work any more. I've also had sucess with this and this but haven't some anything serious with either. It would definately be nice to expand the tools aspect on PolyVox in the future to include something like this.

There is also: http://www.cs.princeton.edu/~min/binvox/

binvox. Haven't tried it myself but I know a lot of people use it to import models into Minecraft, and that it was linked to on realtimerendering blog so it's good

Hey guys!

My name is Sergio González. I am a mexican CS student currently working as a research assistant at my university. I'll be using Polyvox to implement an algorithm my supervisors have developed to represent 3d meshes with strings of numbers and parentheses. There is currently no code written so I am the one to implement it all. Couldn't ask for a cooler project. I might graduate thanks to this.
This isn't the point of their paper, but my hope is that the algorithm can be used to compress voxel data. My hopes aren't too high for practicality since they haven't done asymptotic analysis of the algorithm. Hopefully I can finish soon and have a linear algorithm with a good compression rate =). In the worst case, I'll have a lot of fun working with voxels.

Thanks to this library I can start working on the algorithm right now without worrying about representing/visualizing the voxels, which would have taken a lot of time and probably would be filled with bugs; so thank you so much!!

Welcome to the forums!

If you look back a few posts you'll see there has been some discussion about implementing some voxel compression. Although I haven't officially announced anything yet this work is actually already underway in a seperate branch which you can get here:

https://thermite.svn.sourceforge.net/sv ... anches/RLE

It needs some tidying up (note that the class documentation is completly wrong becaue it describes the old system) but I think it will be ready to merge with the main trunk within the next week. Maybe it can provide a starting point for your voxel compression. In particular the volume is built of blocks, each of which has a compress() and uncompress() function. These are currently implemented using RLE, but you should be able to put a different scheme in quite easily.

Hello everyone!

My name is Shanee, I am a student of Biology & Computer Science in the Open University of Israel. In my free time (what's that?) I make games!.. ok, so far I made a game engine, but a game should come next

I have recently got my engine to what I felt is "ready for game" state and started working on a game but my boyfriend who is my partner for the project (when he has the time, even less time than me he has!) really inspired me to look into volumetric terrain so we can have cave system and cities which could resemble the Underdark (for you D&D lovers out there). Oh, him and Minecraft inspired me

One of my desires is to make a game which is sort of like Ultima and Ultima Online in style, will most definitely have multiplayer in it and a lot of exploration, quests, dialogs but because of our huge team (two people) and vast amount of assets available (nearing zero) we are looking into cool algorithms to have a lot of the features generated somehow.

So what am I doing here? Ha! main reason for me to post after all, isn't it? Well, I been looking into marching cubes... googling, gamedev'ing, searching, reading, oh what's this? PolyVox... looks interesting... next! But then I remembered, "don't re-write what's already there for you to use" and made my way back here. I went over the license, free to use for commercial, feel free to put credits if you like (I like, you deserve it ), blah blah blah, oh a small guide.. super simple to use actually. Hey! this could be it.

The story from the previous boring paragraph results with: "I'm here, hello!" or well, this post. So... Hi!

I am looking forward to trying to implement PolyVox into MLE (the engine my bf and I are working on) tomorrow, after I finally get some sleep. Or the more sensible choice - sometime after my biology exam on monday (that will be hard to wait so long so don't count on it!).

That's about it!

Except the questions!
MLE is using DirectX 9, I saw the example uses OpenGL. As far as I know the OpenGL and DirectX triangles/indices are pretty much flipped, aren't they? Can I just copy the vertices and indices given by the SurfaceExtractor to my DirectX vertex buffers and index buffers as is?

Second question, I tried to look but didn't find much (btw parts of the manual seem to be broken?) - How does LOD work here? and I heard there were some issues with holes?

Question number three, texturing. How do you get the UV on the surface?

Last but not least, question number four! I saw the videos show real time editing of the voxels, how will this apply for terrain? Does it mean you should use relatively small "volumes" so when updated the vertex buffers are easy to update? What sizes of voluem should I consider?

Thanks for reading! And even more thanks for making this awesome project.