- 27,387
- Toronto
- NewAesthetic
- SlipZtrEm
Two main issues;
1. That'd be a daft way of referring to them.
2. They'd have to be car-specific - one scheme created for an LCC Rocket won't transfer smoothly to something like a Maserati.
The " I want a livery editor" thread.
- Thread starter stucar17
- 1,351 comments
- 116,494 views
Two main issues;
1. That'd be a daft way of referring to them.
2. They'd have to be car-specific - one scheme created for an LCC Rocket won't transfer smoothly to something like a Maserati.
- 1,601
- Vienna
- bmw123_
Man, I hope you are right... this would AWESOME... I can see all the online lobbies "livery show off and trade"
- 1,997
- United Kingdom
- 8,722
Given the texture detail available on the cars, a full livery editor just isn't going to be very nice looking on PS3. Even if they do something like what Forza does, and have high-detail textures for photomode, the contrast to what would be achieveable in-game (during a race / replay) would just jar far too much, in my opinion. I hope to be wrong.
What we may get, though, are decals. Quite what of and how many, how flexible etc., I don't know. But a lot of the incidental detail on cars (and tracks) appears to already be decals, effectively.
What we may get, though, are decals. Quite what of and how many, how flexible etc., I don't know. But a lot of the incidental detail on cars (and tracks) appears to already be decals, effectively.
- 1,997
- United Kingdom
- 202
- Kekistan
- The_Rotorhead
On the second point, not necessarily. With Midnight Club LA's livery editor, even a full livery is quite easy to fix from car to car. That is because it is well divided into 5 sides (left, right, top, front and rear) and it allows to pick all the stickers on a side and move them all at once. After that, each sticker can be moved by it self for a fine tune.
Hmmm, I don't know about that...
Don't premium race cars already have what we can call liveries? And they look just as good in race/replays as they do in the photomode.
I believe it depends solely on how PD implements that.
Last edited:
- 8,722
Yeah, I think PD are using a very sophisticated "UV" mapping system (how the texture is applied to the mesh); possibly several separate maps of different types, all hand-made for the best efficiency, in terms of visible texel density on the car vs. available texel number in the car's textures. That means you can put the texels where they're needed most, where there is fine detail, and not waste them elsewhere.
If we're not going to be given the ability to do that UV mapping ourselves (it's pretty crazy stuff), it presents a problem in that the texel density will have to be more-or-less uniform over the whole car, just in case anyone wants to put detail anywhere on the car. Most liveries have "blank space" that can be exploited in the mapping by not allocating (m)any texels to that area. I doubt that's very easy to automate, and will not be consistent depending on what is actually put on the car.
If PD limit the number of colours that we can have in a livery (bit depth, effectively), or do it using material IDs (paint colours, etc.) instead of specifying the diffuse colour in the texture, they may get more detail (resolution) out of it that way. That's complicated by texture compression...
But it's a fine balance, and only PD knows where that balance lies.
If we're not going to be given the ability to do that UV mapping ourselves (it's pretty crazy stuff), it presents a problem in that the texel density will have to be more-or-less uniform over the whole car, just in case anyone wants to put detail anywhere on the car. Most liveries have "blank space" that can be exploited in the mapping by not allocating (m)any texels to that area. I doubt that's very easy to automate, and will not be consistent depending on what is actually put on the car.
If PD limit the number of colours that we can have in a livery (bit depth, effectively), or do it using material IDs (paint colours, etc.) instead of specifying the diffuse colour in the texture, they may get more detail (resolution) out of it that way. That's complicated by texture compression...
But it's a fine balance, and only PD knows where that balance lies.
- 8,722
Where did I say it wasn't? For a livery editor, you can't really do it the normal way, you have to do it "old-school" (because you just don't know where the detail will be needed). Unless you go really, really sophisticated. The mapping and the unwrapping are different steps, but are linked (i.e. what you can do with mapping might be determined by how you unwrap the mesh, which is why you sometimes need more than one map).
My bet is that the memory limitations on PS3 mean you have to work a little bit harder for that same visual quality. You could probably do everything with one map on PC and not worry about the wasteage, but I expect that'd feel dirty.
- 13,892
- Adelaide
- Neomone
You think they did this for all the standard race cars as well? They have liveries of an acceptable level for a livery editor.
I don't see any reason why it's not images mapped directly to the surface of the car. Memory limitations don't come into it, because it'd be the same size images that are currently required on the cars.
- 8,722
I don't know. They used decals even then, for some things, and even with GT3 there was some nifty masking going on given how the paint was shaded.
It's interesting that you think the quality of the Standards is fine, when the main issue most people have with them is the texture quality. But these things are very subjective.
I fully expect that Standard-level liveries are possible, but I just wouldn't have expected that to be acceptable.
It is images mapped directly to the car. It's just how they're mapped that's important. If all you have is a little logo you want to put on the door, you can use all of your texel budget to make that as nice and sharp as you can, and not waste it on the rest of the car where you have nothing. If, on the other hand, you want a full-body job, like the BMW art cars, say, then you have to spend that texel budget on the whole car, and the relative quality (taking that small area your logo took up on the door) will be much lower as a result, using the same number of texels (which is what determines how much memory you need). You could forgo the masking, but I don't know how that would affect the shading and hence the final quality.
If you just use one mapping for all liveries on a given car, that small logo will have the same quality as the full-body scheme. Most artists would prefer to optimise the usage of texels just a little bit more than that, but how do you do that for an arbitrary "livery"? And how do you make that work for every car without ugly artefacts, or "good" and "bad" cars emerging?
HKS racer
(Banned)
- 4,802
Lame cars don't need liveries at all. They don't even need to be in the game if you ask me.
Just use common sense. What are the race cars people would like to paint? All the premium Super GT's, all the Race Converted, all the premium LMP's , Nascar, GT3 and stuff like that. Is it to much work? If so PD should really quit gaming and make wine.
If your resolution is high enough you can have all the tiny details you want wherever you want. If common sense 2D artist thinks 2048*1024 resolution is no longer high enough they jump to 4096*2048. There is no guru sophistication yada yada BS, just higher resolution, higher details.
You should see a couple of SimBin templates. Here as reference a Zonda R template from Race 07, which was a 2048*1024 template, but was upsacled by the author and then painted at 4096*2048. The final result is an HD livery, zoom on the decals, you'll realize it's the resolution that give you "details". UV mapping has already been done, they optimized the space as much as possible, they surely didn't waste space and this is normal routine for common sense 2d artist, nothing "sofisticate". If you want more details it's just a matter of doubling the resolution, that's what he did on that skin.
You can be sure GT5 Super GT and NASCAR skins have a fairly decent resolution as well.
Last edited:
- 13,892
- Adelaide
- Neomone
Be careful. I didn't say that the quality of the Standards was fine. I said the the quality of the Standard liveries would be acceptable for a livery editor.
The liveries on the Standard GT300 cars are fine for what might be output from a user created livery. Maybe they're optimising each section of those liveries so that it has only the quality that it needs and no more. If you've got some evidence that they're using these techniques, or that they've used similar techniques in the past then please link it to me.
Otherwise I'll assume they're using the simplest solution of one image mapped to the whole car. And that technique provides no obstacles to a working livery editor.
- 327
- Essex United Kingdom
- Davey BoyEssex
Livery editor is a must along with more flexibility in modifications, Because there is one thing i want to do when GT6 arrives and that is to buy a TVR Cerbera Speed 6 and race mod it and bring the Cerbera Speed 6 LM Race Car from the original game back.
- 8,722
When I said quality of the Standards, I meant the liveries. That's why I said what I said about their texture quality, as that's what the liveries are - textures, which you know.
I'm not sure what I'm supposed to be linking to. UV mapping is pretty diverse. I can link to some academic papers on the subject, if that's what you're looking for. It's likely to be a bit mathsy, though.
The "simplest" solution may not be adequate for the image quality required, and that complicates things. If you're happy with Standard quality liveries, that's certainly possible, but I doubt most people will be happy with that. As I said, it's up to PD to make that decision (whether the quality's good enough), but it'd be nice if they just did it anyway.
ITCC_Andrew
(Banned)
- 18,532
- Kitchener
Can I just say, 43 pages later... There definitely seems to be more talk about a livery editor than almost any other feature for GT6.
- 13,892
- Adelaide
- Neomone
You made the claim that they're using some sophisticated hand-made UV mapping system in GT5. I'm not entirely sure what you're claiming for GT3, but it seems to be something non-standard.
I'm asking for what evidence you have that these techniques are being used.
- 4,548
- Australia
Sadly I see nothing to base your optimism on.
- 8,722
Again, someone else who likes to jump to conclusions based on their fantasies of my motivations for posting.
Nowhere did I say these things are non-standard. Sophistication is the norm in graphics tech. Sophistication, of the nature I'm referring to, doesn't translate well to the general case, because by definition they're specific optimisations.
You can use one image for the whole game if you want to. It's the mapping that is important, how that image is applied to the mesh(es). To make best use of the texel count you have, you want to put those texels where there is the most detail, and not where there is no detail (on the mesh). That is what you do especially if your texel budget is limited, relatively speaking. In fact, this is one of the criticisms of the MegaTexture approach, in that the texel density is constant for everything, when artists are used to being able to use it where it's needed most (but there are ways around that even then, apparently).
How do you ascertain where on the car the detail will be, in advance, in the case of a livery editor? You can't, you either have to make the mapping unform over the whole mesh (constant texel density), or you have some algorithm make the mapping for you to put the texels only where you've drawn stuff on the car.
I'm sure they exist, but I don't know how "good" they are, and whether they work in "post-production". The mappings are stored in the model files, often using dimensionless units, not in the textures; this is so the textures can be replaced easily, or be a different size and still "work" (and also for generality of code in the renderer). As such, the mappings are done when the model is made, not whilst the game is being played, generally. But procedural content is made somehow, I guess, so...
All I'm trying to say is that the quality of a custom livery will probably be nowhere near what PD has managed with the Premium cars, because they were able to hand-tweak the unwrapped mesh and the mapping to focus on the parts with the detail, and even use multiple unwrappings / mappings to leverage any advantages they might have. We probably won't be given that ability, and will likely have to work with a uniform mapping applied to a "safe" unwrapping (recall that, in Forza, some cars have weird areas that don't work well when painted over), which will in all likelihood not look "good" on PS3, in terms of texture detail (but maybe OK, it'll depend).
I hope they can magic it, because obviously a livery editor is a great feature to have, so this is not me arguing against one, either, in case someone wants to dream it.
- 13,892
- Adelaide
- Neomone
What conclusion am I jumping to? You made a claim, I'm asking for evidence.
You misunderstand how I used the word sophisticated. I used it to simply mean "more complex than normal". If you just want to use it to mean complicated, that covers pretty much every computing concept since WW2.
Nonetheless, constant texel density is the norm for PC sims where you can see what the mapping is doing. iRacing uses a simple image file in which different parts of the image map to different parts of the car. So does pCARS. So does rFactor. So does Shift and Shift 2.
That is the norm. If you want to prove otherwise, provide examples of games that use other techniques, as I have provided examples.
So again, you've come up with a huge reply claiming that PD are doing something more complex/sophisticated than simply mapping a single image. But still without addressing the question.
Provide some evidence that this is the case, or retract your statements.
- 8,722
Oh my word. I never said custom liveries were done any other way but uniformly. Preferentially tweaking the way the unwrapped mesh sits on the atlas is common practice for textures that are not meant to be drawn on by the player (because you don't know where they're going to draw). In the case of LFS' liveries, you'll notice that the templates you have do not include things like the headlights etc. (although they used to, and they can still easily be found now).
Alternatively, you could just watch one of the thousands of videos on YouTube, or equivalent, and see it in action for yourself. Here's one example, skip to 13 minutes if you don't want to watch it all, and notice the difference in texel density on the different parts of the model:
What I'm saying is, if PD are doing any of that redistribution of texels to maximise the detail only where it's needed in order to improve the apparent overall texture detail (and hence perceived quality) of the cars they're making themselves, then it won't look as good (overall) if they use uniform mapping for the custom liveries we'd create. That might be OK, I don't know.
It's especially noticeable on the Standards, where some cars look OK, but others have really dire texture detail because there's just too much going on, and the texel budget is not sufficient to maintain the detail over the whole car. Cars are mostly paint, paint has been handled by a pixel shader since GT3; theoretically, you could map all those polygons with paint on to a single pixel of a special colour in the texture (and render any screen pixel that is that colour using the paint shader), but you won't get very nice edges (masking) that way.
What I would do is cut out those parts that are only paint (leaving a buffer to any non-paint areas for masking) and overlay them on one pixel (actually, you would just set their material ID to not use the texture at all; i.e. just exclude them from the map altogether). Then you have a larger area (texel count) to do your painting on. Of course, you need to know where you're going to paint first, and where to "cut out". This is the hand-made part. The sophistication is in the method of projecting the polygons onto the map for minimal distortion and for controlling uniformity of texel density in world space (where required), and the use of multiple maps (strictly, the UV map is not the texture, although the interchange is common and confusing) for different parts of the mesh to leverage benefits of different projections etc.
Take this example:
Why should the UV map be uniform over the whole car when the texture detail is confined to a few very small parts of it? Most artists would preferentially scale those polygons where the decals are to sit over the largest possible area in the texture, so as to maximise the apparent texture detail (texel density).
Actually, in that case, you would just use a decal projection for each actual decal, which is much more efficient (it's still a UV map).
I can see a system like what GRiD 2 has as being a tempting option: a few hand-made templates with tidy maps and optimised detail etc. to choose from, plus the ability to project a few decals onto the body. (Which is what I said in the first place).
- 13,892
- Adelaide
- Neomone
Because they have to be able to deal with the worst case scenario, a field of 16 premium GT500 cars/NASCARs with full liveries.
If your system can cope with that, why bother tweaking to optimise less complex cases?
Anyway, you've answered my question. You have no idea what techniques PD are or aren't using. You're making some educated guesses, couching them in complex technical language to make it more difficult for others to refute and presenting them as facts.
Bravo.
- 22,551
- Arizona
- HamiltonMP427
- 35,335
- Addison,Texas
- GTP_RACECAR
I'm not at all sure myself but all I know is going all the way back to GT3, they've been using Maya.
- 8,722
My system? What are you on about?
You're the one pulling stuff out your arse: what worst case scenario is that, when the texel budget is determined by the memory allocated for textures? It is fixed, per car. What the artists do is put it to best use, per car. This is standard practice.
You were trying to refute this based on some misguided fantasy that I was trying to big up PD... for what, exactly?
No, I'm trying to ground expectations, as usual. If I just so happen to introduce some actual information, is that so bad? I only added more technical stuff because I was challenged (because of a misunderstanding of my intentions, as usual), and I don't know enough about it to streamline my explanations. If you want to participate, you have an obligation to educate yourself on the issue as well, I can't do it all for you.
Anyway, at least my guesses are educated, if only minimally.
I still don't have a clue what it was that you thought I was trying to say, and I don't think you know either - just another opportunity to take a jab; bravo, indeed. 👎
- 13,892
- Adelaide
- Neomone
You misinterpreted that. I referred to PDs system. If PD has a graphical system that can deal with 16 premium NASCARs with liveries that cover practically the whole car on track, what advantages are there in optimising something like that GT Academy car?
This is my point. You can't present an educated guess as a fact.
Let me point out again:
Evidence for PD using these techniques? Your educated guesses are all we have so far. If you've got more to present then be my guest, or you can continue getting pissy that I called you on presenting opinions as facts.
Last edited:
- 8,722
It'll look better as a result? I suspect you're still not fully aware of what is going on with UV maps, and how fundamental they are in the modeling - it is standard procedure to "optimise" the usage of available texture space; it's pretty straightforwardly sensible no matter how you look at it, when UV maps are unique for every model anyway.
Never did any such thing.
It's the industry standard to optimise the UV map in such a way - you'd know that if you even did the tiniest bit of investigating. If you're going to "call" someone, at least be sure you have the capacity to understand the subject at hand. It's not difficult, in principle. My mistake for assuming it was common knowledge, or at least search engine friendly.
Although, actually, there is a move to more automated methods, and directly painted on textures (using the 3D viewport) with fully automated mapping (but still the option of hand-tweaking things), but this is potentially going to be sub-optimal in the high performance arena - i.e., big budget.
Similar threads
