<--noob. ya now that i see Adam's example that is a lot of space. It's just that we used 2048 for the mod facade so i was like hmmm...., then realized that we have 4 times that much space now. anyway, it's early...leave me alone
I have a question about blending tiles. I can think of a few ways to do it but none of them are specifically sanctioned or outlawed by the rules...
Are blend materials in 3dsmax/Maya ok. Can we use the vertex color channel to paint a mask or do we need to use a bitmap on another UV channel, so it comes out of our texture pool?
It's fine for a screenshot from far away, but for a real game env that would be a joke
Well, considering that you're making an environment thats basically only going to be seen from one or two viewpoints, you can be very selective about where your detail is going. You're not going to be building an entire level, nor are do you need to build something that holds up to ultra-close-ups either. Considering all this, 1 4096 is just fine. If you really think about it, those 2048s that you guys use for wall textures are getting mipped down to 1024s and mostly 512s 90% of the time anyway.
About the scale and scope - a little vignette would not require 250k polys right? Some of the concept art is massive in scale and some is very enclosed/small. With the textures couldn't you fake the distant features/use smaller texture sizes?
Size/scale is entirely up to you.
Texture usage is entirely up to you.
Its very easy if you want to make a very detailed small scene. Theres nothing there to stop people that would rather focus on tight scene with loads of detail as apposed to a massive scene. This sort of stuff kind of evens itself out.
If you want to make a large scene you're naturally going to have a farther away camera.
If you want a small scene its naturally going to be a tighter camera.
Your max res for your renders is 1100x1100 pixels, if they gave out restrictions of 8 4096's theres no way possible you could even get close to representing that level of detail in such a small render.
So is this a game art environment challenge or a make environment for a screen shot challenge? I'm not complaining at all but i'm just noticing people saying that you should build the environment for the specific shot and not worry about what it would look like if you walked around in it....which kind of flies in the face of "game art".
Thats a valid point but i think you could look at it more like this: If you wanted to gear your scene more to FPS standards, you'de probabbly want to make a smaller, more detailed scene, if you'de rather show off a larger, rts-style massive scene, you'de be better off to keep it less detailed. I think that this mentality transfers itself well to what actual games use. Of course this isnt some sort of absolute statement, that all RTS games have to have little detail or all FPS games are take place in cramped corridors(tho most do =P).
do people not know what tiling textures are anymore? Earthquake hit the most important nail on this one; just because everything you make is 1024 doesn't mean that it is ever displayed at that resolution in game...All this asks is that you make your textures "pre-mipped."
Think about what size they would REALLY end up being. You'd be shocked how many textures are actually rendering out at 256. The amount of time you'll actually see an in game texture remain 1024 is slim as hell.
Remember, 90% of environment assets aren't actually made like the Eat3d tutorials...
I imagine if someone did a bunch of facing planes with pretty textures they wouldn't place. After all in the Domwar you have to worry about the character being viewed from all angles, not just the front.
I can see optimizing the scene for a specific camera angle, that happens all the time, every game panders to the player/camera. But I think there should be some kind of a line if people are deleting back faces that would normally be viewed if you stepped forward or turned around?
Remember, 90% of environment assets aren't actually made like the Eat3d tutorials...
Why have that mentality? I think most environment assets SHOULD be given that much love, i know more than 90% of the environment assets I've made have.
maybe this mentality is the problem? I think most environment assets SHOULD be given that much love, i know more than 90% of the environment assets I've made have.
Yes, but your experiences alone do not represent the entire gammut of the industry, this sort of stuff varies heavily based on what type of game you're working on. Now you cant honestly tell me that every RTS game should get the same amount of detail put into it that Gears has for example, i mean really, thats just logistically impossible, not to mention performance wise.
Now i would love to be able to have the time, resources, and hardware support to be able to create unique 4096s for every asset in an entire game, but you have to realize this just is not realistic for every project.
It's not a question of "how much love" they get, it's a matter of the fact that it's totally unrealistic to think that you can give every single object a unique unwrap job and a normal map where every single high frequency detail was placed by hand in zbrush.
That kind of time-budget and texture budget is totally ridiculous. Did you happen to notice that EVERY single chair, concrete barrier, and 99.99% of the other breaking objects and simple props in Gears were IDENTICAL? I think all that "love" is lost when that chair you spent two weeks carefully making and custom unwrapping is used 40 times in one scene because there is no texture budget or time left to make additional variations.
Tinman - good question! Here's your answer: Make it for a videogame, but only worry about your 2 beauty renders and how they'll frame. Don't worry about camera fly throughs and all of that. Pretend you're making it for a videogame but showcase it with 2 interesting compositions & setups.
Which videogame you make it for is irrelevant, thats your choice.
EDIT: Remember guys, this is a challenge. A contest based on who creates the most compelling videogame environment based on the limitations at hand.
The 'Mod Facade' challenge was 1x2048x2048 and 20,000 and look how well those turned out. If you wanted to, you could do 4x that amount of work for your challenge.
Or, you could tighten the shot up of your facade and put more details in to that. Or pull it far away and spread your detail & pixel density around.
And, that's just for a building facade or street scene. Your options are endless here.
It's completely up to you how to use those resource but to ASSUME that some amazing environments cannot come out of these limitations makes me worry how creative some of you are or if you're stuck in a mindset. I know... I know you guys will do a damn fine job if you break out of this linear way of thinking some of you are showing and really think about what this is: A challenge.
Beyond the opinions of the specs, I'm insanely stoked to see what you guys produce. Both as a fan of you guys individually and as a fan of environment art.
So is this a game art environment challenge or a make environment for a screen shot challenge? I'm not complaining at all but i'm just noticing people saying that you should build the environment for the specific shot and not worry about what it would look like if you walked around in it....which kind of flies in the face of "game art".
This is also what I'm trying to get at , it says in the brief that it is best if the environment:
Definition of Playable: Making a pretty image is great! But making it look like it can fit in the universe of a game is a bonus!
If a camera is looking at 250k of polys, with nothing behind it, then its not playable. Going to make the environment encapsulated anyway, but its good to know whats best.
bah i'll leave this thread alone, somehow me saying environment art should be given more love was turned into me saying each asset should be given a 4096 and every game type should be given the same amount of detail per square ft be it an rts or a fps.
I will say I dont understand why you're hating on building a highpoly mesh for each asset sandbag. You see repetative assets in a game regardless but you're using that as an excuse for not taking the time to detail an asset. You're also assuming it takes two weeks to create a "nice" chair. You didnt post an alternative so what are we to assume? just slap a tileable wooden texture on there?
at any rate I feel like I'm detracting from the contest so i'll stop.
I'm of the same opinion as well but I fail to see how you couldn't fit it all on a 4096.
Either way, we've got some ideas throwing about at the moment after spotting a few awesome concepts, really excited at the possibilities
What I am saying is that in the context of an argument that 2 4096's do not provide enough texture space, saying that every object in a scene could realistically receive the "eat 3d approach" is preposterous under the development time lines that the vast majority of games are allowed.
Would I "slap on a tiling wooden texture?" Of course not. Would I take it into ZBrush and model the wood grain? Absolutely not.
The chair was just an example because I distinctly remember being bothered by the fact that they made one chair that broke one way and put it everywhere. Then they did the same thing to the vast majority of the rest of their props. My main objection is to the Eat 3d method of taking an object and unwrapping the entire thing as if it were a single character. I realize that it was an exorcise, but in my opinion it is terrible form to suggest modeling and texturing a fence and the concrete it is in front of (not even going on the concept of that) as one object is ridiculous. Do you really not have any other concrete or fence in the rest of the scene? I would think the normal thing to do would be to make the concrete part using a nice tileable concrete texture (as surely there are other large stretches of concrete walls in the area made of the same material), then place the fence in front of that using its own tiling fence-chain texture (since logically there will be other chain-link fences in the scene to go along with this chunk of one), etc. Regardless of whether or not its an exorcise, you would never do it that way professionally, so why do it that way at all?
It is that kind of process failure that bothers me and the same kind that leads to people thinking 2 4096 maps are insufficient texture space for an isolated scene.
This is gonna rock, hard to chose from all these concepts (especially in a 3 man team, gotta please everyone).
Sandbag, who said anything about 2 4096 maps, it only 1 no?
Guess that could sound a bit small when you're thinking of giant environments, but it'll just be a good excercise in making modular props and textures.
I like the specs, and Sandbag is right. A 4096 is PLENTY.
I dont really like most of the concept art, it has that tutorial video paint style where you cant tell shit about the scene. You pretty much get a color palette, with random scribblings of shapes that in no way will help me develop a scene fast and efficiently.
I like the specs, and Sandbag is right. A 4096 is PLENTY.
I dont really like most of the concept art, it has that tutorial video paint style where you cant tell shit about the scene. You pretty much get a color palette, with random scribblings of shapes that in no way will help me develop a scene fast and efficiently.
FFS! Everyone's bitching about being forced to use a concept. USE ONE OF THOSE BLURRY ONES! Do what you want with it, an just use the color palatte/overall shapes. Then you're free to do whatever the fuck you want in there.
Im not bitching. If the concept is so unreadable that you have to say "Do whatever the fuck you want", then why am I using concept art in the first place.
Anyhoo, opinions & bitching is perfectly fine. I just hope nothing deters people from showcasing their skillset as I think that would be pretty silly of them.
Just to be clear - I didnt see all the concepts until the moment we went live, and no one on my team was told what I knew. I want to compete as evenly as possible :P
My team is myself and 2 other fellas from work who'll register and post in our thread once we get it going.
Doesn't matter what team Adam is on, I need to get better at environment art anyway. KJ steroids are popular at the moment, emitation is a form of flattery and all musicians know that learning from pros will improve their own skills.
All this hate... We truely are the evil side.
I don't understand why locking the camera off from 1 or 2 viewpoints makes the environment "unplayable". Games like Silent Hill, Resident Evil, Prince of Persia have done this and its what made the games so stylised and memorable. If you want to full encapsulate your entire environment, then by all means.
But if it comes to needing an additional 1k tris, I'll be backface culling without hesitation.
If a camera is looking at 250k of polys, with nothing behind it, then its not playable.
If the camera is locked to one spot. Pick two.
In many games, a level can be very small, but included a reasonable duration of game play. Think current adventure games. Dreamfall comes to mind. Events take place within a very small path of a level, before the next level is loaded. With a small level, and clever use of a skybox, the environment can appear enormous and highly detailed without demanding resources. Anything that can't be seen from that short path, doesn't need to be modeled. Gears of War is the same way, if you've seen any of the glitch videos on youtube. Pick two points along a short path, and build around them for the beauty shots. Put the most thought and detail around them. For now, we'll call it Marketing.
I'm just going to construct it as something that can be used in the average third person game like an mmo. With objects the player can get more close to being more detailed than backdrop stuff, and wide enough corridors etc..
What are the rules on using vehicles, Brome? Are they allowed at all? If so, are they allowed to be operating as long as theres no operator visible?
Also, skeletons and corpses. Good/Bad? The rules seem to be dissuading the use of anything that looks even slightly like a character, but nothing says evil quite like a good skeleton.
Just to be clear - I didnt see all the concepts until the moment we went live, and no one on my team was told what I knew. I want to compete as evenly as possible :P
Still had a leg up on the specs and other details noone else was privy to, I'm sure.
No worries though, look forward to competing with ya! :] My team should have a thread started soon enough.
Also with some questions regarding the rules. When it comes to exterior environments do we take the skybox out of our texture resource? Also is a full skybox required or can it be a simpler plane for a backdrop that incompasses both our camera shots?
Replies
yeah i was hoping for double this with everything but I can manage to be creative with this.
I know the game i'm on 4096 would be eaten up quickly....
Are blend materials in 3dsmax/Maya ok. Can we use the vertex color channel to paint a mask or do we need to use a bitmap on another UV channel, so it comes out of our texture pool?
It's fine for a screenshot from far away, but for a real game env that would be a joke
2) How can you NOT get an environment done with 64 512x512 textures? That's more then standard for games today...
Well, considering that you're making an environment thats basically only going to be seen from one or two viewpoints, you can be very selective about where your detail is going. You're not going to be building an entire level, nor are do you need to build something that holds up to ultra-close-ups either. Considering all this, 1 4096 is just fine. If you really think about it, those 2048s that you guys use for wall textures are getting mipped down to 1024s and mostly 512s 90% of the time anyway.
EDIT:Bah, too slow.
I think If I combined all the textures that I've done from all the games I've worked on it still wouldn't add up to 4096^2.
Texture usage is entirely up to you.
Its very easy if you want to make a very detailed small scene. Theres nothing there to stop people that would rather focus on tight scene with loads of detail as apposed to a massive scene. This sort of stuff kind of evens itself out.
If you want to make a large scene you're naturally going to have a farther away camera.
If you want a small scene its naturally going to be a tighter camera.
Your max res for your renders is 1100x1100 pixels, if they gave out restrictions of 8 4096's theres no way possible you could even get close to representing that level of detail in such a small render.
Think about what size they would REALLY end up being. You'd be shocked how many textures are actually rendering out at 256. The amount of time you'll actually see an in game texture remain 1024 is slim as hell.
Remember, 90% of environment assets aren't actually made like the Eat3d tutorials...
I can see optimizing the scene for a specific camera angle, that happens all the time, every game panders to the player/camera. But I think there should be some kind of a line if people are deleting back faces that would normally be viewed if you stepped forward or turned around?
eh not really, its more than i generally get to work with, working on a 360/ps3 game..
64 512 tex!!
Why have that mentality? I think most environment assets SHOULD be given that much love, i know more than 90% of the environment assets I've made have.
Yes, but your experiences alone do not represent the entire gammut of the industry, this sort of stuff varies heavily based on what type of game you're working on. Now you cant honestly tell me that every RTS game should get the same amount of detail put into it that Gears has for example, i mean really, thats just logistically impossible, not to mention performance wise.
Now i would love to be able to have the time, resources, and hardware support to be able to create unique 4096s for every asset in an entire game, but you have to realize this just is not realistic for every project.
That kind of time-budget and texture budget is totally ridiculous. Did you happen to notice that EVERY single chair, concrete barrier, and 99.99% of the other breaking objects and simple props in Gears were IDENTICAL? I think all that "love" is lost when that chair you spent two weeks carefully making and custom unwrapping is used 40 times in one scene because there is no texture budget or time left to make additional variations.
Which videogame you make it for is irrelevant, thats your choice.
EDIT: Remember guys, this is a challenge. A contest based on who creates the most compelling videogame environment based on the limitations at hand.
The 'Mod Facade' challenge was 1x2048x2048 and 20,000 and look how well those turned out. If you wanted to, you could do 4x that amount of work for your challenge.
Or, you could tighten the shot up of your facade and put more details in to that. Or pull it far away and spread your detail & pixel density around.
And, that's just for a building facade or street scene. Your options are endless here.
It's completely up to you how to use those resource but to ASSUME that some amazing environments cannot come out of these limitations makes me worry how creative some of you are or if you're stuck in a mindset. I know... I know you guys will do a damn fine job if you break out of this linear way of thinking some of you are showing and really think about what this is: A challenge.
Beyond the opinions of the specs, I'm insanely stoked to see what you guys produce. Both as a fan of you guys individually and as a fan of environment art.
This is also what I'm trying to get at , it says in the brief that it is best if the environment:
Definition of Playable: Making a pretty image is great! But making it look like it can fit in the universe of a game is a bonus!
If a camera is looking at 250k of polys, with nothing behind it, then its not playable. Going to make the environment encapsulated anyway, but its good to know whats best.
I will say I dont understand why you're hating on building a highpoly mesh for each asset sandbag. You see repetative assets in a game regardless but you're using that as an excuse for not taking the time to detail an asset. You're also assuming it takes two weeks to create a "nice" chair. You didnt post an alternative so what are we to assume? just slap a tileable wooden texture on there?
at any rate I feel like I'm detracting from the contest so i'll stop.
I'm also on team tinman i believe: give each asset the love it needs in order to be successfully portrayed for what it is. Ya?
*high five*
Either way, we've got some ideas throwing about at the moment after spotting a few awesome concepts, really excited at the possibilities
What I am saying is that in the context of an argument that 2 4096's do not provide enough texture space, saying that every object in a scene could realistically receive the "eat 3d approach" is preposterous under the development time lines that the vast majority of games are allowed.
Would I "slap on a tiling wooden texture?" Of course not. Would I take it into ZBrush and model the wood grain? Absolutely not.
The chair was just an example because I distinctly remember being bothered by the fact that they made one chair that broke one way and put it everywhere. Then they did the same thing to the vast majority of the rest of their props. My main objection is to the Eat 3d method of taking an object and unwrapping the entire thing as if it were a single character. I realize that it was an exorcise, but in my opinion it is terrible form to suggest modeling and texturing a fence and the concrete it is in front of (not even going on the concept of that) as one object is ridiculous. Do you really not have any other concrete or fence in the rest of the scene? I would think the normal thing to do would be to make the concrete part using a nice tileable concrete texture (as surely there are other large stretches of concrete walls in the area made of the same material), then place the fence in front of that using its own tiling fence-chain texture (since logically there will be other chain-link fences in the scene to go along with this chunk of one), etc. Regardless of whether or not its an exorcise, you would never do it that way professionally, so why do it that way at all?
It is that kind of process failure that bothers me and the same kind that leads to people thinking 2 4096 maps are insufficient texture space for an isolated scene.
Sandbag, who said anything about 2 4096 maps, it only 1 no?
Guess that could sound a bit small when you're thinking of giant environments, but it'll just be a good excercise in making modular props and textures.
I dont really like most of the concept art, it has that tutorial video paint style where you cant tell shit about the scene. You pretty much get a color palette, with random scribblings of shapes that in no way will help me develop a scene fast and efficiently.
FFS! Everyone's bitching about being forced to use a concept. USE ONE OF THOSE BLURRY ONES! Do what you want with it, an just use the color palatte/overall shapes. Then you're free to do whatever the fuck you want in there.
Anyhoo, opinions & bitching is perfectly fine. I just hope nothing deters people from showcasing their skillset as I think that would be pretty silly of them.
My team is myself and 2 other fellas from work who'll register and post in our thread once we get it going.
Is my nose brown already? damn.
Fuck You Ror!
I don't understand why locking the camera off from 1 or 2 viewpoints makes the environment "unplayable". Games like Silent Hill, Resident Evil, Prince of Persia have done this and its what made the games so stylised and memorable. If you want to full encapsulate your entire environment, then by all means.
But if it comes to needing an additional 1k tris, I'll be backface culling without hesitation.
-caseyjones
If the camera is locked to one spot. Pick two.
In many games, a level can be very small, but included a reasonable duration of game play. Think current adventure games. Dreamfall comes to mind. Events take place within a very small path of a level, before the next level is loaded. With a small level, and clever use of a skybox, the environment can appear enormous and highly detailed without demanding resources. Anything that can't be seen from that short path, doesn't need to be modeled. Gears of War is the same way, if you've seen any of the glitch videos on youtube. Pick two points along a short path, and build around them for the beauty shots. Put the most thought and detail around them. For now, we'll call it Marketing.
'We build our scene as if its for a videogame, but all we're specifically concerned about is what's in our 2 final beauty renders'.
Also, skeletons and corpses. Good/Bad? The rules seem to be dissuading the use of anything that looks even slightly like a character, but nothing says evil quite like a good skeleton.
Still had a leg up on the specs and other details noone else was privy to, I'm sure.
No worries though, look forward to competing with ya! :] My team should have a thread started soon enough.