next »

[William Chyr]

Looking great dude :D

Thanks you!

[William Chyr]

Devlog Update #218 - 05/31/2015

Started playing around with generating a menger sponge because I wanted to see what it would look like.

Well, it looks crazy awesome, but the way I'm making it is insanely intensive.

400 million tri. 800 million verts. Inside the Unity editor, it runs at a whopping 0.5 FPS...

I think generating ProBuilder objects inside Unity for this stuff is not the right way to go. Probably better to generate it in a 3D software and then import it into Unity. It's just insane right now.

[marcgfx]

looks absolutely insane and very busy. looks like you also turned off shadows, to get it to run. I can't imagine the scale very well, just based on the screens. can you get through the smallest gaps? I was thinking you could possibly replace the last level of holes with a simple rectangle texture (maybe a darker floor-color). it might still give the feel you are trying to achieve but keep the poly-count lower, reduce performance requirements and make the screen less busy.

[ProgramGamer]

A menger sponge is just a fractal, right?

[William Chyr]

looks absolutely insane and very busy. looks like you also turned off shadows, to get it to run. I can't imagine the scale very well, just based on the screens. can you get through the smallest gaps? I was thinking you could possibly replace the last level of holes with a simple rectangle texture (maybe a darker floor-color). it might still give the feel you are trying to achieve but keep the poly-count lower, reduce performance requirements and make the screen less busy.

I actually didn't turn off shadows. I think it's because there's so much geometry, that everything you're seeing *is* in shadows.

The smallest gaps are 2 unit wide, so just enough to pass through standing upright.

I think replacing the last level of holes with a rectangle texture would work really well actually. I'd have to do some work to get the coloring and line thickness right, but I don't think it should be an issue.

I'm sure there's also a lot of culling that can be done, since so much of the geometry is actually hidden.

A menger sponge is just a fractal, right?

Yes, it's a fractal.

[ProgramGamer]

I should just have googled that tbh.

My bad...

[Mark Mayers]

I heavily use fractals in my game to generate procedural geometry, it's essentially how I make all my meshes.

I made a post many moons ago regarding combining procedural meshes into a single game object.

It's possible to combine procedural geometry into a single mesh and save the GameObject overhead in Unity.
Of course, this won't reduce your tri/vertex count. You could potentially implement a LOD or culling system for the fractals, but that could be difficult with a combined mesh.

You might find that code of use (and feel free to use it).  It would be cool to see the fractals in game
Also Willy, I hope to the GameDev gods you're using C#

[Mark Mayers]

Also, I've spent a stupid amount of time with Mandelbulber.

It's an awesome 3D fractal rending program.

You might get inspiration from that!

I made this about a year ago:



This was one of the built in Menger Sponge functions, although I changed the colors a bit.

It's a ton of fun!

[William Chyr]

I heavily use fractals in my game to generate procedural geometry, it's essentially how I make all my meshes.

I made a post many moons ago regarding combining procedural meshes into a single game object.

It's possible to combine procedural geometry into a single mesh and save the GameObject overhead in Unity.
Of course, this won't reduce your tri/vertex count. You could potentially implement a LOD or culling system for the fractals, but that could be difficult with a combined mesh.

You might find that code of use (and feel free to use it).  It would be cool to see the fractals in game
Also Willy, I hope to the GameDev gods you're using C#

Really neat! I'll check it out. David (the programmer who I've brought on to help with the game) has implemented some really sweet editor tools for merging meshes.

We're doing something similar to what you did.

With the menger sponge though, it would require more than that. There are a lot of hidden faces that are being drawn, and they don't need to be. Will be writing more about the system we are using in here soon.

[William Chyr]

Devlog Update #219 - 06/05/2015

This one is really more of a journal post. It's 2:38 AM at the moment on a Friday morning.

Got a ton of stuff I need to do for E3, many due in less than 24 hours.

Definitely feeling super overwhelmed and stressed out, but also really really grateful.

I honestly cannot believe how far the game has come...

Two and  half years ago, I didn't even know what Steam was, and now there are people who are actually excited about Relativity. It's pretty amazing.

Been working mainly on infrastructure and optimization. Realizing that actual game making is incredibly different from prototyping. Can't just run head first and try to make a bunch of things. Need to plan out pipeline and how it will all fit into the project and work with other mechanics.

Of course, it doesn't make sense to that stuff too early, and I'm at exactly the right point to do that, but it was a bit unexpected, and it definitely is much more work than I had expected.

Writing editor tools, rethinking asset pipeline. Thankfully, David, the programmer I've brought up, is an absolute wiz at this stuff.

Ok, enough late night devlogging. Gotta get some work done and submit some assets!

[Torchkas]

oh it's called R3LATIVITY now

[William Chyr]

oh it's called R3LATIVITY now

Ha ha, it's actually more for legal and search reasons that I've decided to make that the full title.

[William Chyr]

Devlog Update #220 - 06/07/2015

Just hit a major milestone for E3 prep. Feeling really great.

The latest build of RELATIVITY is definitely the best one by far. Great framerate, and the feel of movement has improved dramatically.

Did a stream last night making a bunch of crazy fractal levels. They turned out to be even better than I had expected.

Some of the visuals are due to a glitch with Unity's mesh merging function. I believe there is upper limit to that and I hit it, so it created a bunch of angled pieces.

Below are some of my favorite screenshots from the session. Some of them really look like ink drawings straight out of a manga!

[sam_suite]

Gorgeous! Fun stream, too. Good luck at E3.

[qwetro]

It seems like a wonderful game, I wish I could play it soon. 

[Mark Mayers]

Really neat! I'll check it out. David (the programmer who I've brought on to help with the game) has implemented some really sweet editor tools for merging meshes.

We're doing something similar to what you did.

With the menger sponge though, it would require more than that. There are a lot of hidden faces that are being drawn, and they don't need to be. Will be writing more about the system we are using in here soon.

Some of the visuals are due to a glitch with Unity's mesh merging function. I believe there is upper limit to that and I hit it, so it created a bunch of angled pieces.

I think I know what your problem is regarding the visual glitches if David/you are using a similar mesh merging approach.

I wrote a high-level explanation about how I generate my fractals.
It's more primitive and abstract than what you have here, but it's a similar methodology.

I was actually going to originally post something similar in *your* DevLog.
I didn't want to hijack the thread, so I rewrote it from a general perspective, ha.

My glitches might have been unique to my approach, but it seems like you're getting similar visual artifacts.
I don't know if Unity has an upper bound, outside of exceeding the limits of datatypes (ex. buffer overflow).

Anyway, all of this looks great. Keep up the good work 

[William Chyr]

I think I know what your problem is regarding the visual glitches if David/you are using a similar mesh merging approach.

I wrote a high-level explanation about how I generate my fractals.
It's more primitive and abstract than what you have here, but it's a similar methodology.

I was actually going to originally post something similar in *your* DevLog.
I didn't want to hijack the thread, so I rewrote it from a general perspective, ha.

My glitches might have been unique to my approach, but it seems like you're getting similar visual artifacts.
I don't know if Unity has an upper bound, outside of exceeding the limits of datatypes (ex. buffer overflow).

Anyway, all of this looks great. Keep up the good work 

Whoa... I think this might actually be it! I'll forward this to David.

However, I do think there is an upper limit to the number of vertices in the total mesh. For example, in the last repeating pattern, when I tried to merge a whole bunch of tiles, it just stopped at a certain point. There weren't any of the weird angle pieces. It looked like it just ran out. 

Regardless, thanks so much for this Mark! Really appreciate it.

Will update here once we have more information about it.

[William Chyr]

Gorgeous! Fun stream, too. Good luck at E3.

Sweet. Thanks for checking out the stream! :D

It seems like a wonderful game, I wish I could play it soon. 

Any chance you'll be at E3?

[Mark Mayers]

Whoa... I think this might actually be it! I'll forward this to David.

However, I do think there is an upper limit to the number of vertices in the total mesh. For example, in the last repeating pattern, when I tried to merge a whole bunch of tiles, it just stopped at a certain point. There weren't any of the weird angle pieces. It looked like it just ran out. 

Regardless, thanks so much for this Mark! Really appreciate it.

Will update here once we have more information about it.

Did a quick Google search based off of a hunch; and you're definitely right about the set limit.

Unity only supports meshes with up to 65,000 verts.
This makes sense due to data types.

I'm assuming in the Unity engine, the vertex data is stored as a 16 bit integer.
This means it can store 2^16 vertices, which is 65,535.

What I think might be happening is an integer overflow with your vertices.
This could cause corruption, as vertex data could be overwriting data in memory.

You could probably bypass this in your mesh algorithm by splitting it into separate meshes.

So imagine you have a hard limit of 65,535, and you're dealing with cubes of 8 vertices.
Except objects must be processed as triangles and there are separate verts for UVs, so it's actually 24 verts per cube.

So using that math, you have an upper limit of 2730 cube primitives in a single object.
Perhaps set that as your 'magic number' to separate meshes with?

You could also build an occlusion culling system with this, and only render active chunks at a time, so that works out well to divide them!

I'm guessing the problem is the overflow I described or the mesh transform glitches I mentioned in my post.

Either way, good luck and looking forward to seeing what you come up with!

[qwetro]

It seems like a wonderful game, I wish I could play it soon. 

Any chance you'll be at E3?

No, simply just too far for me, I'm here in China.