next »

[Rebusmind]

I love voxel graphics and this looks like a lot of fun. Following!

[bithead]

Absolutely gorgeous, great job!!

[Lee]

Cool stuff, can you say a bit about the direction of the game/gameplay experience?

[djr]

Cool stuff, can you say a bit about the direction of the game/gameplay experience?

The back bone of the game is Ikari Warrior/Heavy Barrel style gameplay. Lots of running/driving around and shooting. There will be a story but it's not critical to the experience, so no cutscenes! The story is there to create a coherent world but who cares when you're going to destroy most of that world anyway!

In addition to the basic gameplay we're adding an exploration element. Because the player can explode the world we want them to want to explore to reveal pickups, paths, secret areas, easter eggs and anything else we can think of. What better way to explore than with a grenade or rocket launcher!

Multiplayer will be 4 player coop on the same screen and over the network.

We're very early on though and we keep discovering new ways to play with the voxels so it might change a bit but we're going to stick to the Twin Stick arcade style experience because that's what we enjoy playing.

We might release the editor as well so people can create levels and share but we're not sure about that side of things yet.

Hope that's enough to whet your appetite?

[Seaport]

We keep forgetting to mention we'll also be adding controllable vehicles in to this game, Dazza's working on a prototype of a tank so we'll hopefully be able to bring you some of that of soon

[Seaport]

As mentioned in the previous post we're working on drivable vehicles, here's a quick peek at one of our new tanks.

[Danmark]

Looks like tons of fun, and by far the best visual style I've seen in a voxel-based game.

How are you rendering the voxels?

[deroesi]

seriously, this looks amazing! following.... 

[djr]

Looks like tons of fun, and by far the best visual style I've seen in a voxel-based game.

How are you rendering the voxels?

Charlie will be pleased to hear you like his art direction

We use PolyVox to manage the voxels. The world is chunked in to 16x16x16 voxel blocks and we build models from that to render. PolyVox provides a mesh extractor so it's pretty easy to get up an running pretty quickly.

We've not experimented too much with the chunk size yet, we've just got it working. It's likely that 16 isn't the most optimal chunk size but it's good enough for now.

The characters are rendered slightly differently, they are built with voxels using Qubicle, exported as OBJ files, imported in to 3DSMax, rigged and skinned, animated and exported at DAE (Collada) files. The game currently does the runtime skinning on the CPU. The downside of this is that the characters are very high poly for their size on screen and since all the voxel data is gone they are not destructible like the world is. This is fine for the characters but for bosses this isn't good enough. To fix this we are implementing a hybrid system where we animate voxel objects and they are destroyable (just like the landscape is) and that will play a big part in the bosses I think. Imagine fighting a boss and revealing it's skeleton as you shoot it's cube like flesh and armour off

Hope that answers your question?

[Batowski]

I've always loved voxels (Outcast is still one of my favourite games) and this looks awesome! 

I often wonder about a parallel universe where 3D accelerator cards hadn't favored polygons, but voxels. Imagine the games we had today...

[sidbarnhoorn]

This looks great! :-)

[djr]

I've always loved voxels (Outcast is still one of my favourite games) and this looks awesome! 

I often wonder about a parallel universe where 3D accelerator cards hadn't favored polygons, but voxels. Imagine the games we had today...

There was a guy on reddit a few days ago doing a voxel Ray Casting engine (all of the CPU). It reminded me of Outcast. I didn't have a machine fast enough to run Outcast back in the day A few devs are working on GPU voxel raycasting as well. Your parallel universe might be merging with our polygon universe 

[Batowski]

There was a guy on reddit a few days ago doing a voxel Ray Casting engine (all of the CPU). It reminded me of Outcast. I didn't have a machine fast enough to run Outcast back in the day A few devs are working on GPU voxel raycasting as well. Your parallel universe might be merging with our polygon universe 

That would be a good day to die.

On a more constructive note, the pacing of your game seems spot on (speed of main char, bullets and enemies). What it needs a little more in my opinion is dynamism. May I suggest giving this talk a watch:

.
I know it sometimes borderlines obvious territory, but still a good source of inspiration.

Another question: have you tried forcing the voxel cubes onto the low res voxel grid? I mean, what if the cubes can not roll and move in the full resolution, but instead they can only move on the 16x16 cube grid and can not roll at all. I think Voxatron does a similar thing, where the smallest component of the playing field is a voxel and there is no further detail or movement inside 1 unit. I hope you get what I mean. 

[Join Indies]

Great job on the detail!  I've noticed a lot of voxel games skimp on voxels. A lot of their models have low voxel counts and look pretty boring.  That's not the case here. The spider shown looks amazing.

[djr]

On a more constructive note, the pacing of your game seems spot on (speed of main char, bullets and enemies). What it needs a little more in my opinion is dynamism. May I suggest giving this talk a watch:

.
I know it sometimes borderlines obvious territory, but still a good source of inspiration.

Another question: have you tried forcing the voxel cubes onto the low res voxel grid? I mean, what if the cubes can not roll and move in the full resolution, but instead they can only move on the 16x16 cube grid and can not roll at all. I think Voxatron does a similar thing, where the smallest component of the playing field is a voxel and there is no further detail or movement inside 1 unit. I hope you get what I mean. 

You're totally correct. Charlie and I have talked about the Vlambeer video before. We're just not at that stage with the player yet, we will be trying out/stealing all those ideas though. There is a lot in the game but we still feel like we're feeling our way around what we want to do and what the tech can do. We've also go so many weapons to prototype and they'll all need that special feel 

We haven't tried locking the voxels to the grid. Right from the beginning we wanted the more free form explosions that a physics system gives you. We do lose some of the 8bit charm that that style offers but we feel we gain a lot by exploiting the feel of the physics engine.

[djr]

Great job on the detail!  I've noticed a lot of voxel games skimp on voxels. A lot of their models have low voxel counts and look pretty boring.  That's not the case here. The spider shown looks amazing.

Thanks, he'll look even better when you can shoot his nuts (and bolts) off 

[Seaport]

Thanks everyone for your comments 
@ Batowski   Cheers for the link, I've probably watched that video about half a dozen times now, its a great guide and set of tips from someone who knows their stuff. I'm also a big fan of their game Nuclear Throne, which according to steam I've played for 32 hours now, yipes, its bit hard though.

[Batowski]

Yes, Vlambeer does very simple stuff, but they are very expressive with it. I admire them for their drive for simplicity, I tend to always go for the deeper and more complicated stuff... 

[Danmark]

We use PolyVox to manage the voxels. The world is chunked in to 16x16x16 voxel blocks and we build models from that to render. PolyVox provides a mesh extractor so it's pretty easy to get up an running pretty quickly.

We've not experimented too much with the chunk size yet, we've just got it working. It's likely that 16 isn't the most optimal chunk size but it's good enough for now.

The characters are rendered slightly differently, they are built with voxels using Qubicle, exported as OBJ files, imported in to 3DSMax, rigged and skinned, animated and exported at DAE (Collada) files. The game currently does the runtime skinning on the CPU. The downside of this is that the characters are very high poly for their size on screen and since all the voxel data is gone they are not destructible like the world is. This is fine for the characters but for bosses this isn't good enough. To fix this we are implementing a hybrid system where we animate voxel objects and they are destroyable (just like the landscape is) and that will play a big part in the bosses I think. Imagine fighting a boss and revealing it's skeleton as you shoot it's cube like flesh and armour off

Hope that answers your question?

Yes, thank you for such a thorough response!

Have you considered using hardware skinning? With effectively one bone per vertex, it seems a trivial case for the technique.

Also, in my tests simple DDA raycasting on the GPU is very fast with modern hardware. Might be worth experimenting with for boss characters & the environment, though you'd lose MSAA and it's not straightforward to reconstruct depths.

[djr]

Yes, thank you for such a thorough response!

Have you considered using hardware skinning? With effectively one bone per vertex, it seems a trivial case for the technique.

Also, in my tests simple DDA raycasting on the GPU is very fast with modern hardware. Might be worth experimenting with for boss characters & the environment, though you'd lose MSAA and it's not straightforward to reconstruct depths.

Hardware skinning is on the books , and yeah the vertex shader would be pretty simple, but at the moment I think the CPU skinning accounts for (worst case) 2% of CPU where as voxel collision is about 60% so I'll be focusing there first

You know, I hadn't even considered Raycasting. My initial thoughts are: Intel GPUs are so bad that it's hard enough to get them to render triangles correctly let alone run sophisticated Vertex/Fragment shaders in real time. But I may be wrong  I'll add it to my research pile.