A Question for You Designers...
polycounter lvl 18
Do you feel the learning curve for all the new technology is too much for most designers to handle? I know there's tons of talent here at Polycount, but can most companies and designers keep up with the next gen of normal maps and other engine tricks? I remember a lot of amazing texture artists and stylistic people on the boards, but the next wave seems to be more geared towards a long modelling process. Not too many years ago there was a wide range of available tutorials, many for tools that most people could pick up on and produce half decent results. Now there seem to be the elites only left, as the mediocre people have given up due to frustration with all the new techniques and tools. As much as I've tried to get back into design, it seems to be a futile attempt as the next gen technology comes out as soon as I learn the current technology.
I guess my final question is : Do you feel the level of elite talent will drop? Do you also feel that normal mapping is the only solution for next gen? IMO, most of the normal mapped games just look bland and about as exciting as a box of stale crackers. Do you think a better product could be produced using high resolution well done texture than simply using engine tricks?
Sorry if it's another repetitive thread, but I've been thinkin about it all day and felt like I needed to get it out of my head.
Thanks!
I guess my final question is : Do you feel the level of elite talent will drop? Do you also feel that normal mapping is the only solution for next gen? IMO, most of the normal mapped games just look bland and about as exciting as a box of stale crackers. Do you think a better product could be produced using high resolution well done texture than simply using engine tricks?
Sorry if it's another repetitive thread, but I've been thinkin about it all day and felt like I needed to get it out of my head.
Thanks!
Replies
The thing to realise that is inevitable is that we are rapidly moving toward a model of making games that utilizes highly specialized talent. I work on a project where designers block out the world, and artists detail it out and make it look pretty. That's actually a fairly common scenario now. So my point is, a designer doesn't necessarily need to know the ins and outs of normal mapping.
I think what you're driving at is that the knowledge and talent required to make content for games is increasing all the time. Whilst I think that's true, dont lose sight of the fact that people are beginning to focus. So you really dont have to 'know it all'.
Btw, I dont think theres anything inherantly wrong with normal mapping at all. Its just been its overly shiny implementation up to now I think. I think you're doing it a disservice dismissing it as a gimmicky engine trick. It's a shader that makes your model look much higher polygon in game than it is. That's all it really is. Why would you not want that?
Im not sure what you mean about the whole 'elite talent' thing. But I think that the ante is constantly being upped yes. Are Epic gonna hire someone to make characters that isn't a highly skilled digital sculptor? Hell no. Is there still a place for other skillsets in the industry or mod scene? Hell yeah.
I think there is a huge majority of people in the industry as artists, who aren't really artists. They are technicians. They have learned a very limited ruleset of this+that=game asset. While it results in ok shippable level work, it's tired, uninspiring, and very blah. Eventually the tools will be available to elliminate this position, and the only time an actual artist will be needed, is when an actual "artist" is needed. When there is a heavy style, a high quality, or a definately look required. Think of it like Poser. At one point if you wanted a human, you had to make it yourself, and there were tons of super horrible human models out. Now if you need a high res model, you can generate one at a click. It won't be anywhere near the quality of a model made by one of polycounts good high res modelers, but it's much better than the average person could produce.
It seems to be headed in this general direction for programmers a bit already, after seeing the UE3 demo. Some of the visual scripting tools looked so easy to use, and the graphics engine was so feature complete, you could probably make a great many games with a large art staff, and a couple scripters.
Eventually there will be the art tools available for this same type of thing. That will result in an overall drop in jobs, but people who can create good art will still be in demand for the companies who want to go with a clear visual style. In fact it might even result in more stylistic games.
I know this was a roundabout answer to your question, so let me address it a bit more specifically and show how the above actually relates in my head. Yeah I think there were a great many more tutorials back in the day, and for a major reason. Most of the people who made them weren't in the industry at the time. They had spare time, it was a hobby, and therefore there was more time, and more philanthropic motivations to make said tutorials. It was also around the time where everyone was just figuring all this stuff out, and not only were tutorials greatly needed, but stuff just wasn't as complicated back then in some ways.
Now most of the people with the learning and the know how are employed, and their time is much better spent on their personal art, or with their families, or on their hobbies. I try to put as much time as I can into tutorials, but it's hard, and I see how many people wouldn't be willing to make that type of time sacrifice for the benefit of others.
Also though, you gotta know that those of us who can do next gen asset creation, didn't have access to some secret tutorials we aren't handin out. We learned off the same general tutorials of old school asset creation and worked our way up the complexity chain. As soon as you learn how to learn, and get a basic understanding of the tools, if you constantly add to your knowledge, there won't be too many radical jumps for you to hit. Those people on the boards who have already made the jumps, those are the people I'm sure you are referring to as the elites, but honestly those are the type of people that will always have jobs, even if/when we get tools that allow us to generate realistic content for our games without an artist present. Because they don't need a tutorial or a hand holding for every tech, they go out and figure it out, or ask around their friends, or find from what current related tutorials there are and reassemble the knowledge from that.
In summary from my long ramblings above, I think there will always be room for artists who have the knowledge and the knowhow to figure out new things, and put down some inspiring work on the paper/wacom/viewport. But I do think the days of the casual artist who doesn't get serious about their craft, being able to find employement in the industry might be coming to a close.
*edit* I forgot to mention why I think the above is a good thing. As soon as making realistic graphics is as easy as a button click, we can hopefully start concentrating again on great gameplay. And even more promising, is that it will pave the way to game developement from your garage again. As soon as game developement is viable to even a team of 3 people working with almost no technical, artistic, or design knowlege, we are gonna see a lot of games that deviate from the norm, a lot of fresh ideas, and out of that pile, hopefully a lot of really fun games.
And the hopefully good news for those who wish to remain employed as artists in the industry, is that we can probably start to focus again on a strong visual style, and not just have "ultimate realism" as our goal.
It seems to me, that normal mapping is being over-used, or overly relied on in many cases to generate detail. When I see some of the normal mapped models broken down into their texture flats, I see some very bland texture work, with most of the detail coming from the normal map.
I think if more effort were put into the base texture, and the normal map used to "enhance" that rather then rely on it to create so much of the detail, then you'd have better models out there.
Doom 3 had a very specific look to it and you know what, I LIKED IT! The reason it doesn't look as good as HL2 (or at least doesn't appear to look as good) is because it IS the first generation of this sort of tech, with unreal 3 out you will see normal maps, per pixel lighting and real time shadows REALLY be put to good use, whith this sort of tech you NEED different surface shaders to achieve different looks, you can't just paint skin anymore and say OK THIS IS SKIN, you have to actually have a shader that MAKES it look like real skin realtime, doom 3 couldn't handle alot of the more advanced shader effects, like sub surface scattering, and thats why everything in it basically looked like it was made out of the same plastic material. And like poop said pretty soon you'll have all that available to you, you'll have all the shaders and all the little technical things in the engine and all you'll have to do is probably just activate it and maybe make an alpha mask texture for it.
Truth is, normal maps aren't going anywhere and neither is highpoly modeling, if you want a job you should stop whining and go learn it, find tutorials, and practice, theres tons of tutorials for highpoly modeling, which has been around only ... forever? in movies and cg. Its not as difficult as you imagine.
And as much as I hate things like facegen I believe they do have their place for games I'm not working on.
And I have a job that pays me quite well as a programmer. Not to sound elitist or anything. I don't care or aspire to become a professional game artist.
Sonic said something that caught my eye, because I had been thinking the same thing about the "blandness". That's why I posted. I don't give two shits about "where things are headed" within the industry.
I know how normal maps work. You don't need to know how they work to see that something looks "bland". Just your eyes. When I say "bland" texture work, it was more in regard to color than detail. There is more to "detail" than just elevation changes.
The UT3 engine videos Ive seen, only look slightly less bland. Maybe it's just the direction of the artwork these days. Or maybe it's because normal maps don't take into consideration the way colors are affected by shadows and lights.
Don't get me wrong, I think normal maps are pretty damn cool, and look great and all, but there is *something* that is lost in it.
On the artist/tech point - what art does not take some techical knowhow? Look at painting- that takes years and years and only a handful can do it well.
Anywho, back to Zbrush...
Daz: generally I was addressing character artists. I think I was expecting every company to demand artists of Epic quality nowadays
Poop: very good post, but do you believe that companies will be using character generators? Even if it's focused on amazing gameplay, won't the assembly line figures ruin the game? Well... now that I think about it, look at how the war games sell, and all their characters look like they were generated...
Irritant: it seems that when a lot of time is put into an amazing texture, the quality is actually degraded when using a normal map to "enhance" it. I think you have to paint textures for normal map usage when you use that technology, otherwise it won't look right at all.
rockstar: I see what you're saying. U3 is an amazing example of normal maps! But although I keep up with upgrading my graphics card, my Radeon 9800 pro would probably crap itself trying to do some of the engines tricks. So when these are disabled, will there still be artwork left behind to make the game look decent? On the other hand, the newer consoles can apparently render it fine and they are gonna be around $300. And don't get me wrong, I know the technology. I did high poly modelling back in the day when I also did 1000 poly models for Half-Life. I think normal mapping is a great new tech, I'm just questioning the focus and direction that game art seems to be heading in.
Am I the only one who thinks normal-mapping looks like ass?
[/ QUOTE ]
I think the highlights and shadows created by normalmaps look aweful on certain assets that aren't made to resemble metal, think shiny plastic trees. But highlights and shadow are how you show detail right? I used ORB to create normal maps for a FarCry mod once. I wasn't impressed, and thought the textured versions were more visually appealing to the eye. Not as much contrast. I think better lighting with gradually higher polycounts would look better and require less dev time, considering the amount of power systems such as the PS3 are boasting. I'm creating 2500+ polygon models for Q3 right now, which is more than the original UT2k3 models I believe. I think the normalmaps in HL2 were used intelligently. Unreal3 tech is amazing for showing the ability of normalmaps, but the Unreal universe is unique, and I do remember seeing a shiny tree somewhere.
Another concern that bothers me is, how difficult is it going to be for future game artists starting out right now, to catch up with technology. I'm optimistic the tools release will help bridge the gap as they are now. I personally had trouble finding the right tools years ago thanks to Autodesk, but am happy to see companies like Softimage give the little guys a good headstart.
A designers bottom line is framerate. On at least 3 of the dx9 ppl games I've made the designers took on the responsibilty of lifghting their own scenes with an artists creative input.
This is the first game I'm on where we artists do the lighting, abnd to be honest I doubt we;'ll keep this pipeline for long (designers will do it soon I'm guessing).
PPL lighting is very different from old sk00l lightmaps cuz every surface a ppl light catches multiplies the passes by 2-4x's (depending on the number of materials on that shader/surface) They have new tools in their arsenel now, they come with benifits and some serious limitations too.
This is an example of where a LD's job is different. Their goal is the same, a target framerate, but now there are more variables, lighting is just the most obvious example I can think of.
-R
This gets back to what i said in my post about specialization. Before long, I can assure you that level designers wont be doing lighting. Dedicated lighting artists will. This is allready happening in a lot of studios.
Agreed with Daz on the "specialisation" of art/design tasks in the future.
-R
UT2k3 had 2-2.5k poly models.
Bad use of normalmaps makes normalmaps look ass. Like sugar instead of salt in your meal, will make your meal taste horrible, which doesn't mean that sugar tastes horrible.
Highlights are not created by normalmaps, therefore their highlights can't look awful. The highlights are defined with the specular and gloss maps. Depending on how much control the engine gives you over these properties and how good your artist is, will tell how good or bad they look. Quit blaming the poor normalmaps, it's not their fault!!?!111!!
FarCry had some horrible misuse of glossiness (worser than Doom3), i wonder if there was a technical limitation behind such blatant sacrilege.
A common problem with the highlights is that they tend to ignore the underlying color of the texture and give you a bland colored highlight layer, partially covering the original colors, again, has nothing to do with the normalmap itself. You can often work around this a bit by overexposing the materials so the highlights have more the effect of color dodge in PS, rather than painting with transparent white over a surface. For skin you'll usually have special skin shaders that will create warm speculars and highlights.
Can't do that much work for every material tho, and it's not needed anyway. Whoever does glossy bricks and trees should have his hands chopped off with a lightsaber. I do think that some people using the new tech get so excited to finally have dynamic crap that they overuse, kinda like CG and other things in movie. Lot's of geeks playing around with that stuff afterall. Still, if it looks bad, blame the artist or if applicable, the engine limitations.
Another thing with Doom3 is that their color maps are pretty bland by themselfes, desatured colors and low contrast between colors. I'm _guessing_ they wanted that realistic desatured look (like in Killzone and Co.) but uhm, i guess it didn't always work with the limited spec/gloss control they had. Good news is, technology can only get better from there!
Daz is right about specialization. Most 'next-gen'ish studios won't have Level designers doing the lighting, or putting the level together at all.
Their job now is to concentrate on the gameplay of the level, get the AI and puzzles working and block it all out roughly, with boxes if it has to be.
We make all the final models, hi and lowpoly, textures and necessary maps and the lighting. It's 'our' goal now to keep the framerates, woo-hoo.
This should make eveyone happy since artists get to be more artsy, rather than doing random models for LDs that end up in an average looking map featuring art mismatch and lack of functionality of the environment.
And, LDs get to spend about the same amount of time per level they always did, but only have to worry about the layout and gameplay now, which should give us all a much more enjoyable gaming experience(tm).
I don't believe in a 1-click solution for creating human heads (as example). If that's the best an art team can do, it's gonna be a pretty mediocre looking game. With budgets for next game titles being anywhere around $ 10M you'll need to impress people with visuals if you want to sell enough units to stay over water at all. There will be sure the one or other Studio making use of that tho.
Basically tho, with EA, Ubi, MS and others buying every Studio that looks interesting, and with the interest of those publishers in pretty visuals it shouldn't be a problem to shift decent character/concept/enviro artists around. After the first project, or after the first months of one project, you'll have enough base heads created that you can change relatively quick in ZBrush. Same for all kinds of other assets. And hey, i remember EA had that awesome 'facescan' technology that made 007 look so realistic!, i'm sure Daz is still all hype over it.
Also, the better handheld consoles get, the more interesting they will be for smaller Dev studios.
I do hope that there will be some filtering in the industry, making artists more valuable and le$$ exchangeable. Right now if you quit, there's 10 others fresh out of school waiting to do your job for free, and they probably even have all the skills required to make perfectly mediocre models, skins and textures.
On that note, we're looking for several extraterrestrially talented artists.
**edit: I seem to have way too much time since i quit playing WoW.
It's certainly a lot more intimidating to try to get into it as a hobby(money, thanks gmax) and a profession(can I even do it).
[/ QUOTE ]
It damn well should be extremely hard to get into the industry. It's a highly skilled job, and I've worked with way too many idiots that have made my life harder over the years.
I've just worked with some real idiots in my time that have no business making games is all I meant. It was a general comment.
Daz: Those ads for Westwood College probably aren't gonna help future prospective (idiot) employees much, then
[/ QUOTE ]
What are you talking about? My mom said I'd never get anywhere with these games! Now my boss needs me to design another game, but first I need to tighten up the graphics a little bit (after I'm done with the QA, of course)! Man, can you believe we got jobs playing videogames all day?!? I know, it's AWESOME!
The impressiveness of the production values in the Incredibles and the Day after tomorrow have not prevented the continued success of Woody Allens films.
r.
I think we can also look forward to more artist friendly UI's for these new features. The first roll out of a new technology always seems to be programer friendly (since they make it up) where you need a masters in C++ to get models in game. Then someone puts out something that makes life slightly more understandible for us artist types.
Also make sure you have worked with the new tech enough to give it a fair shake. 1-3 half finished models does not an understanding bring. Most people in the industry that have to use this new tech learn by doing it a lot.
Some company has to be pretty close to putting out a 3D software package that "does it all", but you can count on that prog being out dated at some point. Right now you have to cobble together 14 plug-ins, 3-5 apps and sacrifce a chicken to get things working. But even in the last year things have gotten easier to work with. If you are a max user I can see where your since of fustration comes from, it seems that AutoDesk is resting on its good looks and charm while Alias swoops in and takes over. I am not saying Maya is the answer to your woes but it is alot simplier to use and better laid out. It also feels like Alias isn't stuck looking back at the glory days, and hasn't given up on its free version of its app, G-max who?
Basically, when you can do highpoly you'll be in the match for a long time to come. No matter how much trickery we'll see, most of it will attempt to replicate hipoly models. Unfortunately I can't do highpoly, I'm always confused when there's that many verts in play.
[/ QUOTE ]
That's it exactly, with highpoly people will just re-use a lot more and save time that way.
Big companies like EA will probably have a libary with meshes/parts of meshes to choose from, and there might even be companys doing only stuff for a libary the can rent to other companies (isn't Liquid already doing something close to this?).