The way to do that would be to have two diffuse textures, a base color texture, and a customizable overlay (or the other way around). The fragment shader can blend in any way that seems interesting, in fact that's what damage textures are for, and randomizing or customizing each separately you have lots of possible variations.riftroamer wrote: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. ..
Models - Chris Kuhn
Moderator: pyramid
-
- Elite
- Posts: 7243
- Joined: Mon Apr 18, 2005 2:40 pm
- Location: LS87, Buenos Aires, República Argentina
Re: Models - Chris Kuhn
-
- Elite
- Posts: 1363
- Joined: Sat Aug 04, 2007 3:42 pm
Re: Models - Chris Kuhn
Coming back to this... I'm not sure it's completely true you can't use the two together, as long as you bake the hull with a small gap between it and the proxy pod.klauss wrote: 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.
That way, when there is no pod, the area around the connection point looks fairly dark, but this darkness seems like it could by grit, grease, a charred area where the pod was blown off or a slot in the hull where a pod can insert. And when a pod is attached, it still looks close to the way it should because the gap was so small that the hull receives most of the occlusion that it should.
And it's no different from what you have to do with turrets, which have exactly the same issue, only worse (due to them being animated as well as modular).
-
- Elite
- Posts: 7243
- Joined: Mon Apr 18, 2005 2:40 pm
- Location: LS87, Buenos Aires, República Argentina
Re: Models - Chris Kuhn
We're not talking about just ambient occlusion. The "ghost pod" will cast shadows. They'll be fuzzy, and not immediately recognizable as such, but they'll look wrong.
-
- Elite
- Posts: 1363
- Joined: Sat Aug 04, 2007 3:42 pm
Re: Models - Chris Kuhn
Then when you fly into a space station hangar or get close to another object, it will look equally wrong. You always make sacrifices when you introduce baked effects into a dynamic environment.klauss wrote: We're not talking about just ambient occlusion. The "ghost pod" will cast shadows. They'll be fuzzy, and not immediately recognizable as such, but they'll look wrong.
But if PRT shadows are much more sensitive than AOs, you could just use the ghost pods exclusively for the AO bakes. Use nothing for the PRT bake.
-
- Elite
- Posts: 7243
- Joined: Mon Apr 18, 2005 2:40 pm
- Location: LS87, Buenos Aires, República Argentina
Re: Models - Chris Kuhn
That's where you're wrong. The PRT is an image-based lighting technique, that uses the surroundings and a transfer function (the PRT) to compute lighting. When you go into a station, the environment is the station, and you are thus lit by reflected light from it.Deus Siddis wrote:Then when you fly into a space station hangar or get close to another object, it will look equally wrong. You always make sacrifices when you introduce baked effects into a dynamic environment.klauss wrote: We're not talking about just ambient occlusion. The "ghost pod" will cast shadows. They'll be fuzzy, and not immediately recognizable as such, but they'll look wrong.
PRT is a replacement for AO. It does what AO does, but better.Deus Siddis wrote:But if PRT shadows are much more sensitive than AOs, you could just use the ghost pods exclusively for the AO bakes. Use nothing for the PRT bake.
However, we could have two versions of the PRT. One with and one without the pod. Not sure how that scales, though.
-
- Elite
- Posts: 1363
- Joined: Sat Aug 04, 2007 3:42 pm
Re: Models - Chris Kuhn
So then why isn't the attached engine pod considered part of the nearby environment? Performance reasons (when there are multiple pods)?klauss wrote: That's where you're wrong. The PRT is an image-based lighting technique, that uses the surroundings and a transfer function (the PRT) to compute lighting. When you go into a station, the environment is the station, and you are thus lit by reflected light from it.
It would most likely scale terribly, since you'd have maybe a half dozen choices of pods for each mount point and perhaps a dozen mount points for hull sections, engine pods, weapon pods, radar arrays, etc.klauss wrote: However, we could have two versions of the PRT. One with and one without the pod. Not sure how that scales, though.
-
- Elite
- Posts: 7243
- Joined: Mon Apr 18, 2005 2:40 pm
- Location: LS87, Buenos Aires, República Argentina
Re: Models - Chris Kuhn
Technical considerations. The pod is in the near-field, the environment is supposed to be in the far-field, where distance is high enough that its direction doesn't vary much across the model's surface. In essence, the pod is too close to be considered "environment".Deus Siddis wrote:So then why isn't the attached engine pod considered part of the nearby environment? Performance reasons (when there are multiple pods)?klauss wrote: That's where you're wrong. The PRT is an image-based lighting technique, that uses the surroundings and a transfer function (the PRT) to compute lighting. When you go into a station, the environment is the station, and you are thus lit by reflected light from it.
The phrase aimed more at how to make it scale, somehow. I don't think there's a way to make it scale, but who knows.Deus Siddis wrote:It would most likely scale terribly, since you'd have maybe a half dozen choices of pods for each mount point and perhaps a dozen mount points for hull sections, engine pods, weapon pods, radar arrays, etc.klauss wrote: However, we could have two versions of the PRT. One with and one without the pod. Not sure how that scales, though.
-
- Elite
- Posts: 1363
- Joined: Sat Aug 04, 2007 3:42 pm
Re: Models - Chris Kuhn
That is quite unfortunate then. Each ship will have to be fully assembled offline before the PRT bake and must remain in that arrangement permanently, so no in game graphically represented upgrades or modular damage modeling. Or else no PRT.
What about UV maps, would each module get its own separate UV map or would multiple modules share the same texture space?
What about UV maps, would each module get its own separate UV map or would multiple modules share the same texture space?
-
- Elite
- Posts: 7243
- Joined: Mon Apr 18, 2005 2:40 pm
- Location: LS87, Buenos Aires, República Argentina
Re: Models - Chris Kuhn
Well, it's not so black and white. If a solution isn't found, it could simply be that the PRT will not account for them, and it won't look totally realistic.Deus Siddis wrote:That is quite unfortunate then. Each ship will have to be fully assembled offline before the PRT bake and must remain in that arrangement permanently, so no in game graphically represented upgrades or modular damage modeling. Or else no PRT.
But I do have ideas.
Each module would be a separate mesh. It would get its own everything.Deus Siddis wrote:What about UV maps, would each module get its own separate UV map or would multiple modules share the same texture space?
I'm betting on the fact that they're reusable, so overall performance won't be hurt, since the engine will be drawing many of the same meshes, which is considerably cheaper than many of different meshes.
LODs might work against that. I guess we'll have to try and see what happens.
At worst, the engine will have to "bake the mesh", to make a wholesome mesh out of its parts. That's also entirely possible without the artist noticing.
-
- Elite
- Posts: 1363
- Joined: Sat Aug 04, 2007 3:42 pm
Re: Models - Chris Kuhn
Intriguing stuff... Tempts me to try putting something together in blender.
Why would modular ships and LODs not play well together?klauss wrote: LODs might work against that. I guess we'll have to try and see what happens.
-
- Elite
- Posts: 7243
- Joined: Mon Apr 18, 2005 2:40 pm
- Location: LS87, Buenos Aires, República Argentina
Re: Models - Chris Kuhn
Modularity makes meshes reusable, which means a single batch might draw many using instancing, or at least many batches would be chepaer because they'd need less state changes (drawing many of the same mesh with same textures makes it cheaper).
However LODs will counter that, because farther instances will in fact use a different mesh (and possibly different texture and shader set).
It's difficult to predict whether the expected speedup of reusing meshes will be achieved given that LODs will reduce the likelihood of actual mesh reuse.
However LODs will counter that, because farther instances will in fact use a different mesh (and possibly different texture and shader set).
It's difficult to predict whether the expected speedup of reusing meshes will be achieved given that LODs will reduce the likelihood of actual mesh reuse.
-
- Elite
- Posts: 1567
- Joined: Tue Jan 26, 2010 2:03 am
Re: Models - Chris Kuhn
Could the shadows not be computed by the engine? For older graphics and slower systems they would have to use precomputed shadows anyway, but for more modern systems we should be able to generate shadows in the scene either in the engine or via shader.
Because of YOU Arbiter, MY kids? can't get enough gas. OR NIPPLE! How does that mkae you feeeel? ~ Halo
-
- Elite
- Posts: 7243
- Joined: Mon Apr 18, 2005 2:40 pm
- Location: LS87, Buenos Aires, República Argentina
Re: Models - Chris Kuhn
I don't think it can be done in realtime, not even in modern or future GPUs. Read the paper I linked, it's a lot more than just shadows.
-
- Elite
- Posts: 1363
- Joined: Sat Aug 04, 2007 3:42 pm
Re: Models - Chris Kuhn
Doesn't this still mean a net gain for batch count though, because if a ship uses four unique meshes that repeat a number of times, that is still four times more than a ship with a single piece mesh?klauss wrote:Modularity makes meshes reusable, which means a single batch might draw many using instancing,
What if you force all modules of a ship to use the same level of detail, based on the ship's distance from the camera instead of letting each module set its own LOD based on its individual distance?However LODs will counter that, because farther instances will in fact use a different mesh (and possibly different texture and shader set).
It's difficult to predict whether the expected speedup of reusing meshes will be achieved given that LODs will reduce the likelihood of actual mesh reuse.
-
- Elite
- Posts: 7243
- Joined: Mon Apr 18, 2005 2:40 pm
- Location: LS87, Buenos Aires, República Argentina
Re: Models - Chris Kuhn
LODs are a net win when the vertex pipeline is saturated. With unified architectures, it's probably less of a concern, but I know you simply can't make that tradeoff for onboard GPUs for instance.
In any case, that will be simply tweaked by the "graphic detail" setting in vssetup, so no concern there. I'm sure we can make this work.
In any case, that will be simply tweaked by the "graphic detail" setting in vssetup, so no concern there. I'm sure we can make this work.
-
- Elite
- Posts: 1567
- Joined: Tue Jan 26, 2010 2:03 am
Re: Models - Chris Kuhn
Even using PRT and shadow maps we should still be able to use real time shadows for some things. They are expensive but not impossible. Its something for the high end cards, or even cards made in the last 3-5 years. The book Game Graphics Programming (2008) talked about them being used in games, though their use being limited due to the expense of the computations. I am beginning to think its more about the limits of the VS engine rather than the limits of the hardware.klauss wrote:I don't think it can be done in realtime, not even in modern or future GPUs. Read the paper I linked, it's a lot more than just shadows.
Because of YOU Arbiter, MY kids? can't get enough gas. OR NIPPLE! How does that mkae you feeeel? ~ Halo
-
- Elite
- Posts: 7243
- Joined: Mon Apr 18, 2005 2:40 pm
- Location: LS87, Buenos Aires, República Argentina
Re: Models - Chris Kuhn
To what "realtime shadows" are you referring to?pheonixstorm wrote:Even using PRT and shadow maps we should still be able to use real time shadows for some things. They are expensive but not impossible. Its something for the high end cards, or even cards made in the last 3-5 years. The book Game Graphics Programming (2008) talked about them being used in games, though their use being limited due to the expense of the computations. I am beginning to think its more about the limits of the VS engine rather than the limits of the hardware.klauss wrote:I don't think it can be done in realtime, not even in modern or future GPUs. Read the paper I linked, it's a lot more than just shadows.
There are a myriad of techniques, and there are just as many shadow types. There's the diffuse shadowing usually called "ambient occlusion", there's soft and hard projection shadows, there's volumetric shadows... the subject is huge.
In all direct shadowing cases, where an object casts a direct-lighting shadow onto another object, VS has little use of it. It's an uncommon occurrence, except for planets and stations shadowing nearby smaller units. The case of planets is handled on the CPU already.
Far more common and unsettling is the absence of self-shadowing: a unit casting shadows onto itself. Most techniques are terrible at this, and here's where PRTs come in handy, because PRTs can model self shadowing. Very diffusely, but enough to make a difference.
Hard self-shadowing can only be done with shadow maps, stencil shadows are a thing of the past, their performance is abysmal nowadays. VS has no shadow mapping. That's our limitation. But if I were to implement it, I would only do it for self-shadowing, or, at most, self + nearest occluder (ie: station). But I predict the motivation for it to be greatly decreased if we start using PRTs. Their shadows are already quite something.
-
- Elite
- Posts: 1516
- Joined: Sat May 10, 2003 5:26 pm
- Location: into the depths
Re: Models - Chris Kuhn
a quick one this one, regarding this guy awesome work:
http://www.blendswap.com/user/KuhnIndustries
he says: I could probably be convinced to do it for free
Maybe asking him, with your ideas, can give a good surprise?
http://www.blendswap.com/user/KuhnIndustries
he says: I could probably be convinced to do it for free
Maybe asking him, with your ideas, can give a good surprise?
I see dead polygons....
-
- Elite
- Posts: 1363
- Joined: Sat Aug 04, 2007 3:42 pm
Re: Models - Chris Kuhn
Giving modular ships a shot...klauss wrote: 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?
Consequence:
Pacifier:
Admonisher:
Six sub-models used. A couple fuselage sections, an engine pod, a radiator, a 16x medium missile pod and a 2x heavy missile pod:
Thoughts?
-
- Elite
- Posts: 7243
- Joined: Mon Apr 18, 2005 2:40 pm
- Location: LS87, Buenos Aires, República Argentina
Re: Models - Chris Kuhn
That in your particular examples, AO will be a nightmare. I guess I really need to solve the AO/PRT baking problem somehow... I need to brainstorm, and for that I need time away from RL
-
- Elite
- Posts: 1363
- Joined: Sat Aug 04, 2007 3:42 pm
Re: Models - Chris Kuhn
Well there is always SSAO.klauss wrote:That in your particular examples, AO will be a nightmare.
-
- Elite
- Posts: 7243
- Joined: Mon Apr 18, 2005 2:40 pm
- Location: LS87, Buenos Aires, República Argentina
Re: Models - Chris Kuhn
I was thinking exactly the same. I don't like SSAO, but it's an option.
-
- Merchant
- Posts: 62
- Joined: Tue Jun 21, 2011 10:33 pm
Re: Models - Chris Kuhn
I think modular stuff is good for mockup purposes if nothing else. There's some interesting shapes you can knock out pretty fast in comparison to ground-up design.
It looks like It'd be restricted mostly to Andolian style vessels, and there's a certain amount of playing about you'd have to do with each base chosen, but I think there's room for expansion/use.
It looks like It'd be restricted mostly to Andolian style vessels, and there's a certain amount of playing about you'd have to do with each base chosen, but I think there's room for expansion/use.
-
- Elite
- Posts: 1363
- Joined: Sat Aug 04, 2007 3:42 pm
Re: Models - Chris Kuhn
Our deficiency is not in concepts but completed content.Primordial wrote: I think modular stuff is good for mockup purposes if nothing else. There's some interesting shapes you can knock out pretty fast in comparison to ground-up design.
This is the situation:
- Every model in the game looks strikingly outdated. Nothing in game looks younger than a decade.
- Making a modern looking model takes at least 10x more work than one from a decade ago.
- There is no skilled, content creator manpower still working on VS.
- Do nothing. Easiest solution but certainly contributes to the continued slow death of VS.
- Dramatically reduce the number of ships. Like 8 unique ship designs for each major species- Aera, Human and Rlaan.
- Make all ships modular and create a larger pool of ships out of a small number of recurring parts.
The technique would have to work for all factions to be worth the effort. Whatever current aesthetic a faction has would need to be adapted to modularity. The part designs might also be made more intricate and tighter fitting than the above examples, to make the ships look a bit more fluid and unique. But not by too much or else you will either have too small a number of possible, good looking ships designs or so many parts that the advantages evaporate.It looks like It'd be restricted mostly to Andolian style vessels, and there's a certain amount of playing about you'd have to do with each base chosen, but I think there's room for expansion/use.
-
- Elite Mercenary
- Posts: 1329
- Joined: Thu Mar 27, 2003 11:51 pm
- Location: Manchester, UK
- Contact:
Re: Models - Chris Kuhn
Here's another set of ship building modules:
http://www.blendswap.com/blends/view/17601
This idea is not new btw. I guess reality [of lack of artist power] has sunk in more by now though.
http://www.blendswap.com/blends/view/17601
This idea is not new btw. I guess reality [of lack of artist power] has sunk in more by now though.
Free Gamer - free software games compendium and commentary!
FreeGameDev forum - open source game development community
FreeGameDev forum - open source game development community