Voxatron!!

For those who haven't seen it already:

 

Looks fantastic and ticks all my 8 bit boxes.

 

looks like allot of fun! !

 

hahaha, looking at the voxel resolution you could say that game is like a 320x240 pixel equivalent... maybe it's 320x240x180?? >__<;

You could say voxel wise we are at the stage we were with pixels in the early 90s! That was 20 years ago! So in 20 years... will the voxels rule the world? :-0

god help us!

 

Awesome!

 

Looks like a top-down shooter 3d dot game heroes. This is a good thing.

 

Awsome! Just shared your vid on facebook, hopefully it kicks starts some Pr for you.

 

any info on the techniques use to display the voxel terrains and effect ?, look really nice btw

 

this is awesome, really looking forward to play this.

 

Haha that is sooooooo cool! Looks like it would be a ton of fun to play.

 

Nice! Is that Bidule's Chipsounds I detect Voxel Music Man?

 

OMG! Looks awesome!!! Truely inspirational!

 

Hehehe good job, looks fantasitc

 

it's so vibrant and so cool! It's retro and yet refreshing I like it very much!

 

awesomeness!

 

really cool

 

what next? voxelvaders?

 

well.. they're including an editor. it probably won't be difficult to have waves of invaders shuffle left/right and have some walls for you to hid behind.

 

+1

 

oWH WOAH! love the customization

 

Amazing!

 

So long as we're talking about voxels lets not forget:

http://www.atomontage.com/?id=dev_blog

 

Why bother with voxels at all? If you just like the appearance of pixelated 3D objects then why not just make blocky pixellated 3D objects?.

Aren't real voxels irrelevant in an isometric game?. And without voxels the viewable area would be able to cover the entire screen.

 

There are many advantages to using voxels not least the fact that they contain volume unlike meshes which are basically hollow. If you punched a hole in a mesh wall it'd be empty inside, if you did the same with a voxel wall you'd see all the contents of the wall. Think minecraft.

The voxels don't use the GPU, all the pixels you see on screen are the only ones that are rendered so you could have a scene with a million voxels in it and it would run at the same speed (theoretically, depends what else you're doing) The cube effect in that game is deliberate, don't be fooled into thinking it's a limitation.

Then there's destructible worlds with voxel scale physics, something you can't do in a mesh based game, the best you can do is move the mesh around and instantiate some debris, but with voxels you can dislodge the actual particles of the ground so you don't get identical craters and debris no matter what you hit.

Of course both methods have pros and cons

 

... you just don't get it, do you...? XD haha, just kidding!

I guess you're right in many ways. I remember some years ago, I wanted to make a 320x240 oldschool game. But I also wanted it to be REAL 8 bits, and it's been a long time since that's not compatible anymore. But weirdly enough I wanted it to be a true oldchool 8 bit game! not 8 bit sprites on a 32 bit display. Weird huh?

I ended up learning that I could just settle for the look and feel of oldschool game, just do blocky pixelated graphics, perhaps even faked scanlines, and it's satisfying enough.

But even so, with voxels, there's more than just the graphics.

It's about the little details that perhaps doesn't even matter on the whole, but couldn't be faked with polygons. I even like that the graphics are clipped on the edges, it makes it special. It'd be very hard to simulate the clipping with polys, and the way the voxels tear apart would be next to impossible to do by subdividing the triangles. All in all it's not just about the looks, it's about knowing it's all done in actual voxels! That makes it very special!

 

What if the programmer later admits that he's faking the voxels. Would you still be happy? Or would you divorce the game?

But seriously, after looking at the video again I can see that the terrain really is voxels as it's clearly destructible.. But I still don't really get it. But then I don't get Minecraft either.

 

Good point, hehe! If it were the case, not sure if I'd divorce the game altogether, but I'd find it much less interesting! :-[
But I for the looks of it, I strongly believe he didn't do stuff the conventional way!

I guess what I crave for is new and original game ideas!

Btw, I don't get minecraft either, I guess there's a thin line between interesting and boring. Why didn't infiniminer get to be as popular? Minecraft was interesting and different, a simple game full of possibilities, and with a bit of luck became a trend in a very short time.

But back to voxels. I guess it's a different method for representing a world. When I was little, anything with pixels fascinated me. Maybe that's what you don't find so interesting? What's so interesting about a square? I don't know, maybe it's because it's something abstract yet full of life? Hard to explain :-[

 

Perhaps today's games players coincidentally got their first woody while playing Donkey Kong and now they find the sight of pixels strangely arousing.

Pictures of Mario may replace Viagra when today's younger generation reach their 60s.

 

Hot stuff!:

 

Donkey Kong?? Try Atari or Commodore 64! haha XD

j/k!!

 

One of my all time fave games: http://www.youtube.com/watch?v=j_3rmoA1JKY

I would play it all night and go to work the next day hallucinating and hearing "YEEEARGH!" everywhere.

It was the first ever computer game with decent graphics and sound.

 

Beat 'Em and Eat' Em, best game ever:

 

I rest my case..

 

It really does depend on a person. I can play games like Minecraft with regular textures or HD textures all day, same for Terraria and old-school games on emulators. I played games since the Atari when I was 4...but my fiancee got into gaming when she was older, and had prettier graphics...now she prefers games like Dragon Age and the like, and when I get her to try a new game, I'm aware it needs to be fairly pretty to engage her (though she's the biggest critic over story).

Some people simply don't have the imagination to fuel simplistic representations such as voxels or pixelated graphics. Rather depressing...

 

I doubt that it's a lack of imagination. Perhaps it's more a case of some people feeling like they shouldn't have to use their imagination in today's games.

And don't be depressed. You would find it more depressing if everybody decided to use their imagination and only play text based adventures.

 

Hah, the first game I ever played was a BASIC game called "Wizard". I remember being like four years old and my dad having to write down the commands to get it to load up and play.

After talking to a few people, I think I should word it better as a "lack of voluntary imagination". Immediately when the games start, I can get into them, and others can too. But along with what you said, my fiancee has a great imagination, she's an artist and likes to write for a hobby. But for the life of her, she just stares at me when I play older games, or new games like Terraria or Minecraft. Sometimes she may call me insane too...so I think it's kind of a mesh between what I worded poorly earlier, and what you mean, they don't want to volunteer their imagination to interactive media.

Man, I miss books now.

 

That stuff about voxels looking blocky isn't necessarily what voxel graphics are like, the atomontage link I posted shows what high res voxel games look like and that's just one guy, image what an engine like that could do if it had the same development as has been put into polygon graphics. When you can have voxels the same res as your screen then you wont even know the difference.

 

Yeah, voxels are definitely the future for terrains. I'm a 3D Coat user and so I'm a little aware of the pros and cons of voxels.

At the moment most voxel based engines are converting the voxel data into a mesh for displaying. But there will come a time when the voxels are rendered directly onto the screen via ray tracing. The main thing holding them back today is the memory requirements of voxel data. A large high res voxel scene would require a ridiculous amount of memory.

 

Commanche used voxels for the terrain and the maps were as large as in Battlefield size maps

 

but very low res

 

You can get very complicated scenes with the memory available today - especially if the terrain is optimised and procedural.
2GB of ram is a lot of memory - if I remember correctly, comanche ran on 1MB of ram, on a 40mhz cpu. I used to play sentinel on my spectrum - I think it had 10000 3d landscapes in only 48k of RAM. - I don't think RAM's a limit, it just has to be utilised properly.
We sit here saying we have so little capacity and yet look at how much could be done on computers that had 1/1000th of the capacity we have today. Comanche came out almost 20 years ago - it's older than many people on this forum!

If noise can be used to generate a heightmap, then it can be used to generate the voxels below the surface too for when the user digs. This minimises the overhead of voxel memory needed in a very big way.
And voxel groups can be indexed - the same way as a tile map works, just in 3D. I have seen demos where this saved them 80% of the memory space that would have been used up by normal arrays - and this was a very simple implementation where the priority was speed as the terrain was dynamic.
Octrees would also add a huge memory saving as I recon only about 10% of volumetric terrain will actually be utilised in 90% of all games.

If a programmer looks at a gig of ram and says "that limits me to a landscape of only 1000x1000x1000 voxels" then they seriously need to do some homework.

 

I don't know about Comanche 2 and 3, but the first version didn't use true 3D terrain, instead it used a 2D height map to store the terrain. So with the Comanche terrain there couldn't be caves in the terrain or overhanging cliffs etc. So it was basically no different to a low res mesh terrain. And Comanche 4 gave up on voxels completely and used a mesh terrain.

But of course you can always use tricks to reduce memory requirements, but even with the tricks I'm not convinced we have enough memory in today's machines to store a truly dynamic high resolution voxel terrain.

 

Fractals!

 

A real-time fractal terrain could give you the detail, but you won't be able to modify it.


The only solution I can think of to create a large modifiable, dynamic and detailed terrain with today's desktop computers would be to use something like dynamic tessellation. So either something like Sculptris' approach or DirectX 11's real-time tessellation.. The latter should start to appear in this years games and sculpting programs.

Maybe an even better and more optimal approach would be to use real-time CSG on terrains. This method would use less polygons as the deformations/holes will have a cleaner topology.

 

What about.. you make the terrain with fractals, then store it as voxels, and you render it as polygons? Can't go wrong with that!!

 

There isn't enough memory in today's desktop computers that can hold a large modifiable high resolution voxel terrain

That's why ImaginaryHuman suggested fractals, so then you don't have to store any terrain data, but instead have it generated in real-time.

But a real-time fractal terrain can't be modified.

 

There are many imaginary limitations coming through here.
It sounds like you're saying "You can't measure a line that's longer than 1m if you only have a 1m measuring tape."

There is no rule saying your terrain can't be based on an algorythm e.g. fractal, then when a piece is modified, that piece is stored seperately.
There's also no rule saying you have to use only one method.
I don't think I've ever heard anyone even mention things like ROAM on the unity forums.

You could a basic heightmap for the general layout of the world - like looking at a single image of the earth, basically defining continents, seas, mountains etc
Then as you zoom in, details could be fleshed out using a procedure eg fracals or noise.
Then, when a chunk gets modified, the actual details get stored.

"tessellation" is a fancy term that basically means detailed meshes are generated from less detailed ones - i.e. procedural lod. That type of approach will work well for voxels too, you just need to think voxels instead of vertices.

What do you consider "large high resolution voxel terrain "?
I recon 10k x 10k x 1k is big enough for the working area the user is currently in and I'm sure you'll get that into 1GB with ease, and today's computers have several times more than 1GB.

 

I don't mean to be stepping on anyone's toes here.
It's just that when I hear "x is a limitation", I push people to challenge that limitation.

Perhaps the person that initially said X can't be done because we just don't have the memory, just needs a little push, needs to be asked What about compression or What about spatial trees or Just skip data that's not relevant to the user.
Even Eienstein placed imaginary limitations on things (e.g. for many years he thought the universe was a constant size) and that held him back in a big way.

All we really need is someone to stumble across some new ideas. I think the hardware we currently have has incredible power - we just need new thinking in terms of software and approaches. Most people are only limited by what they think they can't do and therefore never try anything different.

 

You're in the mindset of "Anything is possible", which I do tend to agree with. But I differ in that I don't think that anything is possible in any given situation. Instead we sometimes have to think outside the box and change the situation to make something possible.

So it is possible to create a large modifiable and detailed terrain.

...but just not with voxels in today's computers!

And you can only take a discussion/disagreement so far before you end up getting lost in the fuzzy and subjective world of words.

So let's just agree to disagree.

 

I agree.
An entire world in voxels is asking too much.

I have been thinking about a hybrid solution.
Using a standard world (traditional meshes) and then using voxels for their destructability.
e.g. A wall would start off as a normal mesh, then if something crashes into it, use voxels to calculate the damage, then rebuild the mesh(es) using something like marching cubes.
I think that's a feasable alternative to having pre-designed sub-meshes for everything you want to be destructable in the world.
Obviously the models would have to be built in a specific way, but I think it should work.