Models - Chris Kuhn

[gonzo]

Some Hi-Poly models on Blendswap.com .



Model:
http://www.blendswap.com/blends/view/59575
Texture:
http://www.blendswap.com/blends/view/62139



Model:
http://www.blendswap.com/blends/view/60877
No texture and some 300k polys.

Licens:
CC BY 3.0
http://creativecommons.org/licenses/by/3.0/

[klauss]

Cool models. Good license. A pity the station isn't textured. They both need lots of work (LODs, baking) if they're to be integrated with that level of detail, but they'd fit I think.

[pheonixstorm]

Nice station. Any idea on the scale?

[loki1950]

Scale is does not matter as you can change it either in blender or in units.csv but the download size of the .blend is 15 Mb so it is fairly large I imagine.

Enjoy the Choice

[Deus Siddis]

Scale is does not matter as you can change it either in blender or in units.csv

That's an over simplification; any decent model is modeled to a specific scale (preferably 1 unit = 1 meter). If you try to brute force the scale up or down then you will have all kinds of inconsistencies like a half meter wide hand rail or a tenth of a meter wide window. Scale does matter.

[pheonixstorm]

hence, why some of our in-game units suck.

[Deus Siddis]

hence, why some of our in-game units suck.

Though there isn't anything stopping us from rescaling them to a size that their models properly represent.

[pheonixstorm]

No, but scaling a model made with 1 unit = 1 meter and the model is only 20 units in length, up to a 1km ship is just plain wrong. You loose a lot of detail that way. Also makes the texture look bad as well unless it is very plain to begin with. A few of the Aera cap ships fit this horribly. While they may have looked fine 8-10 years ago (if they are that old) they look arse now with some of the more modern ships.

[klauss]

Indeed. I think we need a remodel-a-thon

[Deus Siddis]

No, but scaling a model made with 1 unit = 1 meter and the model is only 20 units in length, up to a 1km ship is just plain wrong. You loose a lot of detail that way.

That is why I am suggesting scaling models down, not up. If a model only has enough detail to be no more than 50 meters, scale it down to 50 meters. That's a quick fix we could do immediately.

Indeed. I think we need a remodel-a-thon

That will only be possible if we get serious about how many models the game should have. You'll never look anything like star citizen if you want modelers to appear out of thin air and create modern, professional quality models and maps for 200+ ships and stations.

We must settle on a relatively tiny number of ships for the whole game, probably no more than what is in privateer, and settle on a minimum quality and completeness level for their graphics. Once this minimum set of ships is fully in game, all of the old content must be dumped. Then rinse and repeat for stations.

It's the only chance VS ever has of modernizing its content and attracting a user community again with its eye candy. You can do all the graphics programming you want but the poor content quality will keep the project down just the same.

[klauss]

What about some creative way to make the most out of less?

Stations can be modular, for instance. Modules can be hand-placed, but it'd be easy to make variation. Or, they can be randomly placed. Possibilities there are infinite.

Ships are tougher, but maybe something can be worked out for them?

I like variety. I don't want to part with it. I agree, we can't have this many assets with the required level of quality, but we can try to exploit each asset to its full potential.

[riftroamer]

What about some creative way to make the most out of less?

Stations can be modular, for instance. Modules can be hand-placed, but it'd be easy to make variation. Or, they can be randomly placed. Possibilities there are infinite.

Ships are tougher, but maybe something can be worked out for them?

I like variety. I don't want to part with it. I agree, we can't have this many assets with the required level of quality, but we can try to exploit each asset to its full potential.

Modular stations are very convenient and the way most current models in VS are actually modeled, it seems to be no real problem at all to have multipart modeles even with intersecting meshes.

I assume you meant to pre-assemble ship variants inside the modeling app prior to converting them for VS and not on the fly in game (which could be saved for later use), right?

Same goes with ships where applicable. Frigates and Freighters being my personal premium suggestions. Frigates are normally quite specialized as they are a small and vulnerable warship. Ther might be boarding, minelaying, gun-ship and other variants that differ in detail only.

Transport ships (not exclusively freighters actually) might offer cargo bay extensions, external drive pods to compensate for additional mass, external container fittings and whatnot. I once presented an example of such (though as part of another discussion):

Without modular upgrades:


With external drive upgrades:


And with external drives and cargo bay extension:


While we used to talk about representing installed upgrades on the model, in this regard that would be three variants of a ship (for the model above for example a Wotan Freighter, Wotan Long Range Transport and a Wotan Maxi-Hauler).

Even more variants are possible with - let's say - a set of cockpit sections, body sections, drive sections and wings and detail parts with matching attachment provisions like this one by SolCommand:

Ten ship variants based on the same hull
http://www.solcommand.com/2013/02/updat ... -pack.html

So I'm all in and probably can help (re-) modeling a ship or two.

So personally I think thats a good way to go and on top of that LOD-wise the external addons could be used only in selected LOD levels. I like that approach.

[klauss]

What about some creative way to make the most out of less?

Stations can be modular, for instance. Modules can be hand-placed, but it'd be easy to make variation. Or, they can be randomly placed. Possibilities there are infinite.

Ships are tougher, but maybe something can be worked out for them?

I like variety. I don't want to part with it. I agree, we can't have this many assets with the required level of quality, but we can try to exploit each asset to its full potential.

Modular stations are very convenient and the way most current models in VS are actually modeled, it seems to be no real problem at all to have multipart modeles even with intersecting meshes.

I assume you meant to pre-assemble ship variants inside the modeling app prior to converting them for VS and not on the fly in game (which could be saved for later use), right?

No, I mean by the game. Otherwise, the human resources are still tied to maintaining those meshes.

Now that you mention upgrades and sections, imagine having a unit have a list of meshes, and for each mesh, a list of alternatives. Something like:

{{body.bfxm}{body2.bfxm}{body3.bfxm},{engine.bfxm}{engine2.bfxm}{engine4.bfxm}}

Or whatever other format makes sense.

Of course, each component would have to have also some transformation values, so you can translate, rotate and scale.

Then, when a unit is created, you pick the unit row in units.csv, plus a random seed. With that random seed, of of each alternative is picked.

Problem here is ambient occlusion and PRT maps, they depend heavily on being able to be baked with an accurate description of the surrounding meshes. How do we solve this? I'm thinking...

[gonzo]

...imagine having a unit have a list of meshes, and for each mesh, a list of alternatives. Something like:

{{body.bfxm}{body2.bfxm}{body3.bfxm},{engine.bfxm}{engine2.bfxm}{engine4.bfxm}}

Or whatever other format makes sense.

When you posted yesterday I was fully occupied imagining precicely that.

Of course, each component would have to have also some transformation values, so you can translate, rotate and scale.

You could go much simpler and have fixed hulls that you can splice together. Some basic shuttle components that have the same rotation and scale.

Problem here is ambient occlusion and PRT maps, they depend heavily on being able to be baked with an accurate description of the surrounding meshes. How do we solve this? I'm thinking...

I've never baked anything but if they're done (baked) the same way (light and rotation) I don't think this would be a problem.

Transport ships (not exclusively freighters actually) might offer cargo bay extensions, external drive pods to compensate for additional mass, external container fittings and whatnot. I once presented an example of such (though as part of another discussion):

Yes. Simple stuff like this would make a difference.

[klauss]

Problem here is ambient occlusion and PRT maps, they depend heavily on being able to be baked with an accurate description of the surrounding meshes. How do we solve this? I'm thinking...

I've never baked anything but if they're done (baked) the same way (light and rotation) I don't think this would be a problem.

Think of it like this. A hull has no engine pods, so if you bake a PRT, it would say light comes from all directions, and compute occlusion of hull-on-hull features. But if you add an engine pod, now the engine pod blocks some light. If it was all a single mesh, the PRT would reflect this, and areas of the ship in the pod's shadow would not receive light from the pod's direction. But since they're separate meshes, now they'd get lit from that direction and look wrong.

Now, if you absolutely know a hull will have engine pods, you can bake the PRT with a "proxy pod", a pod you use only to block light, but one that you don't include in the mesh. That'd work. But then you cannot use the hull without the pod.

So... how to resolve that without adding those constraints? I imagine the engine will have to compensate somewhat, but I'm not sure yet how to proceed.

And no, we can't ignore the issue. PRT-based lighting is awesome, you haven't seen it yet because we have no ships with PRT maps, but I had produced some WIP screenshots of a hornet a while back, and you don't want to break this technique because it really adds a lot of realism.

[gonzo]

Think of it like this. A hull has no engine pods, so if you bake a PRT, it would say light comes from all directions, and compute occlusion of hull-on-hull features. But if you add an engine pod, now the engine pod blocks some light. If it was all a single mesh, the PRT would reflect this, and areas of the ship in the pod's shadow would not receive light from the pod's direction. But since they're separate meshes, now they'd get lit from that direction and look wrong.

Now, if you absolutely know a hull will have engine pods, you can bake the PRT with a "proxy pod", a pod you use only to block light, but one that you don't include in the mesh. That'd work. But then you cannot use the hull without the pod.

So... how to resolve that without adding those constraints?

I think I get it. What I mean is for some "production line" where you give a constraint that, for instance, the hull cannot have a shadowing mod/pod and only be allowed to have various lengths, with/without windows etc.
You wouldn't be able to vary a model infinitely but enough to make a noticable difference from the amount of work you put into it.

[pheonixstorm]

Two things. First, I like the number of ship types we currently have. It feels more organic and alive to have such a variety. BUT we do have a problem with having that many ships and having them HIGH quality. So, what we really need is a core set of high quality non-modular ships. Say 12 or so per race, and the other 48+ can be modular randomly created by the game or by the artist from whatever building blocks we have. Should solve both problems.

Second. Has anyone ever played galactic civilizations? Its a 4X game so maybe not.. but all ship designs are completely modular. You have a number of parts you can use in the ship designer (can be scaled etc) to create whatever style ship you want. You get a base hull as a starting point, the rest is up to you.

Anyway, my 2 pence

[klauss]

Two things. First, I like the number of ship types we currently have. It feels more organic and alive to have such a variety. BUT we do have a problem with having that many ships and having them HIGH quality. So, what we really need is a core set of high quality non-modular ships. Say 12 or so per race, and the other 48+ can be modular randomly created by the game or by the artist from whatever building blocks we have. Should solve both problems.

I agree in general. The particulars might vary. 12 HQ ships per race might be too much for our capabilities. Scratch might. I think it is too much.

Second. Has anyone ever played galactic civilizations? Its a 4X game so maybe not.. but all ship designs are completely modular. You have a number of parts you can use in the ship designer (can be scaled etc) to create whatever style ship you want. You get a base hull as a starting point, the rest is up to you.

Anyway, my 2 pence

Yep, I have. They always looked like crap though, stuff bunched up together with superglue. We have to avoid that sloppiness, it's fine for a 4X game, but not for VS.

[Deus Siddis]

What about some creative way to make the most out of less?
Stations can be modular, for instance. Modules can be hand-placed, but it'd be easy to make variation. Or, they can be randomly placed. Possibilities there are infinite.
Ships are tougher, but maybe something can be worked out for them?

For aerodynamic or organic looking craft, modular designs would be more difficult to pull off or just less modular. For everything else that should work fine.

And then you would want to seriously think about making upgrades visible additions to a craft and localize damage modeling to modules. Because the work would be half way done for these features.

[klauss]

What about some creative way to make the most out of less?
Stations can be modular, for instance. Modules can be hand-placed, but it'd be easy to make variation. Or, they can be randomly placed. Possibilities there are infinite.
Ships are tougher, but maybe something can be worked out for them?

For aerodynamic or organic looking craft, modular designs would be more difficult to pull off or just less modular. For everything else that should work fine.

A shader-based sculpting method could be used for those. Imagine this, all such models have a displacement texture, so by randomizing the texture, you randomize the shape. Displacement can be applied on the GPU or on the CPU, it wouldn't be difficult either way, and it would allow some shape customization for aerodynamic models. Not sure what that would do to baked textures, there's displacement-friendly normal mapping methods, but we don't use them ATM.

And then you would want to seriously think about making upgrades visible additions to a craft and localize damage modeling to modules. Because the work would be half way done for these features.

Indeed.

[Deus Siddis]

That sounds pretty cutting edge. But maybe it isn't such a bad thing that aerodynamic stuff be less varied, in the first place.

Either way, you'll have to decide whether to pull the trigger on modular ships, since it will be a whole new direction for content creation. Instead of contributing whole ships that have been (in theory) optimized to render as a single model, 3D content creators will need to build parts of ships while following new specifications that insure everything fits together in the end.

[klauss]

That sounds pretty cutting edge. But maybe it isn't such a bad thing that aerodynamic stuff be less varied, in the first place.

Either way, you'll have to decide whether to pull the trigger on modular ships, since it will be a whole new direction for content creation. Instead of contributing whole ships that have been (in theory) optimized to render as a single model, 3D content creators will need to build parts of ships while following new specifications that insure everything fits together in the end.

It doesn't have to be fully random. An artist can provide a custom ship, and variations of its pieces. That's the whole point, that models are custom, but up to a point.

Stations are another story. What I described for ships retains the overall structure, but stations would vary wildly, with interconnected pieces and little common structure. Thing is, it's far easier to do this for stations, since that's the way stations are supposed to grow: you need a new place to store cargo, attach another cargo area.

[Deus Siddis]

I don't mean the ships need to be procedurally constructed, but you will get way more mileage out of your ship parts if they are built to some specifications. The parts used to build the ship the artist intended, could be rearranged and mixed with parts by other artists to create dozens of ships. And ships will have a higher degree of consistency, which the game is often lacking at the moment (we have ships supposedly built by the same faction that look nothing alike).

[klauss]

I don't mean the ships need to be procedurally constructed, but you will get way more mileage out of your ship parts if they are built to some specifications. The parts used to build the ship the artist intended, could be rearranged and mixed with parts by other artists to create dozens of ships. And ships will have a higher degree of consistency, which the game is often lacking at the moment (we have ships supposedly built by the same faction that look nothing alike).

Oh, yes. Didn't read it that way, but yes.

[riftroamer]

I don't mean the ships need to be procedurally constructed, but you will get way more mileage out of your ship parts if they are built to some specifications. The parts used to build the ship the artist intended, could be rearranged and mixed with parts by other artists to create dozens of ships. And ships will have a higher degree of consistency, which the game is often lacking at the moment (we have ships supposedly built by the same faction that look nothing alike).

Oh, yes. Didn't read it that way, but yes.

I think I saw a design guide or style description somewhere on the wiki and I do agree that a) model parts by different artists help to create a more immersive and still consitent variation and think that b) each faction should have their own bits and pieces.

For my (currently entirely theortical) mod I jotted down variants of vessels all following a common style as a guide for my later modeling. Here are two examples...

Ships belonging to the Antares Clans Union should for example follow these rough lines:


while Ships belonging to the T'Quar race should for example follow these rough lines:


In addition to the different ship classes bits and pieces to detail or modifie the existing design are planned to be made interchangeable to a certain degree: patrol craft components to be shared by fighter, bombers and corvettes, those designed for sub-capital ships to be shared by frigates and destroyers and those for cap ships to be shared by cruisers, battlecruisers, battleships, carriers etc.

As you see this is entirely non Vega Strike but may serve as examples.

[Edit] With careful design and cleverly selected attachment areas (slots) and an automatic component placement as well as the possible option to manually buy hull upgrades based of the same parts would be awesome.And expanding on this possibly to the ability to tell the engine to ignore selcted slots (so these may be available for manual upgrading) - if it could create such ships procedurally in the distant future would be beyond this world.

I'm practically drooling over the remote possibility that it might be considered. So I personally think this is going into the optimal direction content wise. Variation. Consitency. Player control over a custom ship design.

And a working easy to use content creation pipeline would be really great. But that has been stated before.

That being said say two artists produce models following the same (forgive me for staying with the terminology of my post) the slot design or scale but chose different coliring of their textures. Is there a way to automatically blend a texture with a base color (similar to vertex shading in blender)? I think for procedural geneeation thw color sceme of the hull shoild match its upgrades. ..