next gen woes?
polycount lvl 666
How are you all finding the next gen stuff, normal mapping and the like.
Part of me says, yeah this is fun , another part says I miss making mid range poly models and making a nice simple textures for them
there seem to be so many processes, plugins and other various annoyances before you can call a piece of work finished these days.
Part of me says, yeah this is fun , another part says I miss making mid range poly models and making a nice simple textures for them
there seem to be so many processes, plugins and other various annoyances before you can call a piece of work finished these days.
Replies
I like the freedom that modelling in high-poly gives, and I haven't run into any major pitfalls (yet :P) with normal-mapping and baking textures, it's all working pretty smoothly for me.
There's still plenty of time for "traditional" painted textures though, I'm doing some right now...
I had a way of putting my own style on my previous texture work, but it seems harder to do that now.
And jesus, painting at 1024x res as opposed to 256 is a whole different ball game.
KDR_11k - I totally agree with you there, Trying to keep quads is a nightmare at times
As Stahlberg and StrangeFate (both seasoned pros with high-poly modelling) say, you can get away with n-gons in a lot of areas.
A triangle or a pentagon here or there is no harm if it smooths correctly, and if it doesn't get in the way of a good ring/loop selection.
Keeping perfect quads everywhere is overrated IMHO ... I pretty much entirely work in triangles for low-poly models, and quads/tris/n-gons (whichever works best in a particular area) for high-poly models. After all, for game art at least, it's just gonna be smoothed for a render-to-texture, so it really doesn't matter how "perfect" or "clean" the geometry is, as long as it's easy to work with, and looks fine after subdivision.
I guess if your modelling app is restrictive towards n-gons, that would be a bit of an annoyance...
I used to hate painting textures at 1024x1024, but after doing a fair bit of texture painting for work this year at that resolution, I've warmed to it a little. I still prefer lower-res, since it's faster and easier, but 1024 is nowhere near as daunting to me now as it used to be.
Ruz: If you're normal mapping, usually just having a normal-map and diffuse texture is gonna look pretty weird, you really need a good quality specular map (probably should work on it at the same time as the diffuse) in order for stuff to look right ... also a nice lighting setup is always helpful too. Don't forget the dirtmap/ambient occlusion/GI (whatever you want to call it) to add to your diffuse texture too, they go a long way towards making stuff look really cool.
MoP
I suppose what bugs me most is that normal maps at least for heads and orgainc stuff have a tendency to look scruffy
Though, as for the development techniques, i am interested in learning how they work, since it will be taking up a big part of the industry, probably.
Personal-taste-wise i dont like it much, but i think it's a cool development which can be used to make some very cool games.
The nicest thing about quads is that they keep a mesh looking less cluttered.
It's been said many times - work smarter, not harder. I know work won't care when I say that I've written a few scripts to automate things. One sets up the character folder, creates master and TGA texture folders, copies base checked textures into the TGA folder and sets up the max materials to be named and pointing at the TGA textures. All the characters artists need to do is save their textures over the copied base TGAs and the file naes are always correct. This helps the entire pipeline - artists have a little less boring work to do, and the coders have a naming convention.
Ambient occlusion bakes? Yup a script that dupes the model, splits into seperate models based on material Ids, applys a gray material and loads a light rig, then bakes out AO maps.
Once you get your head down and a few models built, it's easy enough.
he he,I was just thinking that today after being excited about the new stuff initially.
I think once we get in good habbits and software develops a bit more everything will be fine. Granted normal mapping does have it's quirks at times and can be finnicky but it'll get better and become more streamlined (there are already better programs out there than others).
It also bugs me when people say the use of normal maps and complicated shaders produces realism and leaves style in the dust. I think this is a misconception. You can do plenty of awesome stuff thats stylised and even do more than you could before with the new tech.
All in all i dont think ppl should be worried about change
but i like modeling both ways low and high poly,makes you think a little more, atm im working on this,been close to 5 days
edit:i know it aint pimping or previews but noone ever offers C&C on my work when i post in the what are you working on thread
i like being able to practice in both forms
Too fussy, too convoluted, too confusing.
I think in general the industry is too caught up at looking for that technological edge to increase visual quality. 2d was beaten to death... justifying the jump to 3d. But now... we're simply pushing through what can be done with 3d just because it generates buzz (doom3/UE3/etc). There's still so much that can be done with basic 3d.
and more. But I agree on being just too much work
for a single char. For an ex. It can take up to two
bloody weeks, modelling and texturing.
Looks like making chars these days are becoming
more of a team effort.
I agree with Soul.
I think hf2 was a great example of a healthy mix of both technologies.
I hated normalmapping when appeared. I kinda cant stand to make a model all the day adding details over details then spend alot of time doing the lowpoly model then bake and bla bla.
But time but time, I gonna used to it and I must admit: Its different but can be fun anyway. Especially with apps like Zbrush which gives you back the artist feel you always used to put in your models ages ago.
Personally I have to learn a lot with hipoly modelling but i think at the end it looks better. I mean hey! Making games that look like some prerendered CG movies cant hurt anyone.
As I told many others. Learn hipoly modelling with the normalmap stuff. What you personally use and preffer is up to you. ther is always a market for cellshaded games, or painted textures...hmm or PSP
My 2 cents
http://www.xcloud.net/images/2004/f_general_n.jpg
[Edited to make image into text link]
I completly agree pogonip.
If I had a job in the gaming buiesness(Not time for me yet...) I would go with handheld 3d games since I like making painted diffuse freehand texture and they look beter then any normal mapped stuff.
How am I finding the next gen stuff?
Its fine and all...but how will it be in the futre?
Damn I feel smart now.
But seriously, I think the biggest setback, is the impact that it will have on the developers.
It far more demanding making games with normalmapps etc, studios will either have to hire allot more artist or make smaller games.
Im just speculating now, but maybe it will become even harder for smaller studios to stay alive. I dont think the publishers want to pay more to the developers and I have a hard time thinking the games will become more expensive in the stores.
One thing I'd like to possibly see is normal mapping combined with silhouette clipping as detailed here:
http://people.deas.harvard.edu/~pvs/research/silclip/
Most of the FPS are turning into variations on the first Half-Life, imo. I don't really bother with them any more.
Due to the fact that it takes more time (money) to make models, i think we won't see as much new and innovative games any more, since those are a risk for the publisher. We'll probably end up with Doom 23, Half-Life 42, Quake 15, SSX 51, etc, etc.
But once again, my main argument:
Why add more detail? It will increase the chance that the player might miss something important, because it doesn't stick out any more in the whole picture. And look at the Red Cloud pic up there. I think that is the level of detail that is perfect for games.
Normalmap useage does not mean ultradetailled assets, to me it's just a great tool to get better surface definition which is a bit different. One could use Nmaps to smooth things or give some areas more definition without going stellar. Right?
Not every button/bolt/piston/wire has to be put in that...
I *think* the huge advantage of currentgen/oldgen engines is the way they forced a strong separation between 'largescale' shapes (in this case the 3d model) and 'smallscale' details (the painted texture).
Keeping these 2 levels separated is something important in traditionnal art and it has translated very well in games like the first MGS (ultralow texture resolution on strongly optimized models giving the game a kind of abstract look) or the Wipeout series (solid shapes against perfectly balanced color themes) which were both severly limited by the 95 playstation hardware.
On the other side of the spectrum some characters in UT2004 started to look a tad overdetailled and were a bit confusing to read.
I think it's easy to forget about these simple rules when working highrez or normalmappy, since one is often working on tiny areas without looking at the whole picture. Hence a solid planning is required beforehand (as always ...
I believe the work Hawkprey did for the UT2007 reddish character, and the resulting game asset produced by the team, realy sells the nextgen tech. It's awesome to see how the concept is soo strong even if almost monochromatic. It gave birth to a beautiful game asset, instantly recognizable from a distance, not THAT highly detailled, but getting the most of what nmaps can offer.
Plus hybid assets are possible too, like, nmapped armor against a traditionnally painted 256 head texture. I think that can look great. I guess it's close to what has been described above with the HL2 example.
I think it's very exciting!
Instead, tricks and new methods will make graphics look better.
the dynamic ambient occlusion demo by nvidia and the stuff about spherical harmonics was already quite nice to see.
as for the topic, I am currently more on the "tech side" and its fascinating what you can do with the shader stuff, indeed most prominent might be normal mapping, but if you take for instance the specular shine of asphalt in many race games these days, which is just a simple pixel shader, but makes the surface just a lot more cooler.
also blooming, glow, and other shaders will add to a much greater variety in the looks.
why we dont see this variety so much yet, is simply that there is still a lot of older hardware around, but I think that especially the new consoles ahead we will see a lot more stylized stuff.
however I also think that with modelling every nut and bolt, a bit too much is done on the art side, while the game might suffer. I just wonder if the money the art costs is really worth it, I remember playing mohaa some years ago, and while not nextgen or so, they had those really hi detail dds textures, and there was one mission where you'd blow up a submarine. What surprised me is the amount of detail that went into the sub and its textures, considering that you only went in planted the bomb, ran out, which lasts like a few seconds...
also about the hi poly detail stuff, someone here had actually posted a sort of blend technique of mixing painted bump maps with generated normal maps, so that you mostly use the normal maps for smoothing detail, but not model every bit in the hi poly model, which looked to me to a quick and easy method of generating normal maps... however I am not really into production using next-gen stuff so...
in general I think the development around graphics is definetely cool, as progress is unstoppable anyway. however as others mentioned, ultra tech graphics arent everything, the game and the general mood is what counts, and I can still enjoy playing SNES games or other lowtech stuff, however the design/game is so well and "round" in itself, that its fun to play and look at.
i think valve pulled it off nicely by using subtle normal maps on their characters/creatures ... and yet they still painted their textures traditionally, more or less ..
games are becoming technology demonstrations rather than actual fun
I mean Unreal 3.0 surely looks spectacular, but i've become kind of immune to it now, i expected Epic to push gameplay/story/plot of their next-gen game as far as they pushed the technology and frankly i am quite
infact, i think Prey does that somewhat well ... At least they are introducing some interesting elements into the experience and having their own take on things, even if it is powered by next-gen technology
and yet they still painted their textures traditionally, more or less ..
[/ QUOTE ]
weren't much paintin' goin' on there actually. Largely photoslappin' ;-)
World Of Warcraft to me is a perfect example of a video game experience that is utterly beautiful to behold, and an incredibly immersive, compelling game experience to boot, but with pretty modest use of technology.
The immediate environment that surrounds you is pretty low poly, but all painted with very nice cohesive textures, and the world is so vast, the whole experience just feels incredibly rich.
My point being, you don't have to use all the graphical bells and whistles available to us to make a compelling game.
That said, I am so jealous of you lot that are, because I for one, am ready to throw the PS2 devkit out the window. I am so utterly sick to death of how that machine holds me back in what I want to achieve.
Bring it.
As far as gameplay goes, i dont understand how this gets brought up every time someone mentions better graphics, like improving the art automatically degrades the gameplay.
[Edit] Does anyone actually believe that if games werent spending a lot of time working on higher quality art assests that there would be a surge in new and unique gameplay? Fuck no there wouldnt, you would see the same games you're seeing now just churned out faster and for less money. Sure you will always find a couple cool games that offer something new but this will remain the same with next gen tech. Publishers like games that they know will sell, not that they think are unique.
That said, does anyone have a faster way for baking GI lightmaps in max? i've used standard radiosity and light tracer and they are both very slow when it comes to 1024s, 2048s and god forbid a 4096.
[Edit again] Pogin could you link to that image instead of posting it directly? It breaks the entire thread unless you're at some very high widescreen resolution.
I mean, why spend all that time and money into a generic game when you can try new things and try to be original and innovative
[/ QUOTE ]
I think this has been discussed often enough, Publishers DON'T WANT original and innovative stuff. Same old stuff with 1-2 original/improved keywords to stand out from the croud, and you're all set. Make something original and you won't get a deal, too much risk to flop to them.
That's how it is, period.
Prey doesn't seem original or innovative to me, it's a straight shooter with some puzzles. Along those lines you could say all the scrypted monsters spawning in Doom3 was innovative too, since it's something you usually don't see in other games.
I personally like next gen, i wouldn't want to go back to the old stuff again. I enjoy hipoly too much.
You're all a bit too quick to judge the effect of next gen costs on gameplay. The part of the industry going next gen is in a transition time, and those have always been bloody.
Studios are still discovering outsourcing, finding proper partners for that, trying to find and hire skilled people that can efficiently do the job and a lot of people already working with the tech on a product are still learning the tech and hipoly themselves. That wastes a lot of money and time.
There's gonna be some next gen crap and studios in trouble in the first round of next gen games, but once everybody is used to the new things and found proper ways to deal with the costs/time needed, and the higher costs become standard it will be all like it is now again imo.
I think this still possible to to stylistic work once you get over that shiny normal mapped look.
I just manged to get the hang of z brush a bit more this weekend also.
I can finally paint fine detail with out making 'dotted' lines
i am still finding that doing this high poly stuff is making me more tired generally
Its just sesm much more contrrollable that way
I guess it's more of a 'Arcade vs Simulation' situation, which i'm talking about. I'd hate to see the 'arcade' games go, since those are the games i have most fun with.
On the other hand, i'm also getting the hang of Zbrush and i must say it's a pretty nice workflow
And stuff like unreal3 is still heavily stylized and far from reality, or can you carry a giant armor and a big f***in gun, same goes with the super contrasty colors in farcry or other games. we basically talk movie graphics imo and you will find that movies have their own depicition of reality and play a lot with lighting... as well, and looking at movies like "Big Fish" you can do a lot with "reality"
I would also say that the arcade stuff is still out there and will be there, definetely would say a lot of nintendos games are like that.
Anyways, I did some thinking and it might be cool to have WoW in-game look more like the intro CG movie. I'll just stop whining and see how it turns out first.
I just hope that graphics wont get a higher priority than gameplay, in the time to come.
Our company is making the switch to producing next generation graphics. The biggest hits I have seen are learning curves, and time. Learning curves are temporary.
The examples are assuming 8 -10 hours a day or production at a company, not hobbyist rates of production:
Quake2 Style Asset took about 2 days to create the mesh (you needed a nights rest to look at the model with fresh eyes in the morning to make sure it didn't suck, or that you may have overlooked a construction error.), two or three days to create the texture map. and a couple of days to animate them. Some people took more time, some less. but in general you had an in game asset in a week.
Cut to Quake 3. The asset creation took a little more time, 2 -3 days for the geometry, maybe 3 or four days to do the various texture maps. A day to fix the Tags, and troubleshoot import issues, and the animation was a little more complicated, as it all had to be in a row, so it might take a little more than a week, so the development time would take 2 and a half to three weeks.
Unreal Tournament took less time, provided you used default animation, but animating them from scratch, to work well took a lot more time, so assets may take around a month.
On our project, that I cannot go into details, with our old pipeline, it took about a month to create a creature,from start t to finish, provided there weren't any export issues.
No making the switch, asset creation may balloon into 2 months per creature. The "in game Asset's geometry isn't much more complicated in terms of polygon count or rigging, but the High polygon asset seems to be a drain, as well as the higher number of maps for the various shaders. Now this may make things good for the level designers, as they don't have to account for new creatures to populate levels until the old ones are tested and tuned, and Environmental Art is going to blossom like we have never seen before, but for the Character guys, this may be the first time that programmers and designers will have to wait around for us, rather than the other way around where we finish the model, and surf while the others crunch.
I am trying to learn techniques that are above all fast, because even if the budgets for Next Gen Titles grow, to Hollywood proportions, It is doubtful that the schedules will follow suit.
On the other hand, there are advantages for a guy like me, who loves geometry, and loves animating, but thinks that UVW Mapping and texturing is the equivalent of eating the contents of a baby's diaper with a spoon. Because the diffuse maps are now medium saturation, medium tones that look like swatches of house paint, filled into the Photo shop Lasso Tool, and the follow on maps are either generated from the geometry (Baked in GI maps, Normal Maps), or are just swoops and swipes of various brushes and filters following the rules of anatomy and material science (Skin is oily especially at the forehead and the bridge of the nose, Metal is shiny, unless rusty.) So for me it makes some parts easier.
It also allows me to "play" a little more in the design, and see if I can make convincing form, to put all that anatomy knowledge to work, to impress.
For those that like " heavy stylization or abstraction" in their games, well unfortunately the bulk of the market is against you. The conceptual end goal of the gaming market is either the "Holodeck" or "Virtual Reality" where things are defined by "Dramatic Realism, and a high degree of Verisimilitude, where you can screw, or look like supermodels, Cleave out empires with a sword in your hand or magic in your staff, or command minions to conquer the stars in your name. Games these days not only have to be fun, but they have to be immersive, escapist, slightly addictive, and entertaining. Above all Gaming is a business, and we are here (most of us) to create compelling assets.
The "toon Shaded" version of Legend of Zelda has now been acknowledged by it's creator to have been in error. The latest incarnation seems to have an art style much closer to that of "Shrek", except with Manga proportions and character designs, appropriate to the cultural expectations and desires of it's home market. And as Cool as Psychonauts may be, or Grim Fandango, were those A level profit centers?
..and it's looking at the business side of things, especially the schedules, that makes me worry. Games with these sorts of time restrictions are not going to have a lot of Variation in Creature. types, except maybe map variations. They may have extensive and beautiful environments, but closer inspection may reveal a commonality of architectural elements, as well as the ubiquitous crates and barrels. But this time it won't be a case of the machine limiting the scope of the assets, but the time needed to create them. Unless the schedules expand out to the luxuriant levels of primary Console Developers in Japan of 3 years plus, I don't see a lot of breadth or scope in the next gen just yet. Plenty of opportunity for a unique visual experience, just not a deep one.
The other thing I am worried about is specialization. The era of the "Character guy may be passing, where the asset is the result of a group of specialists, such as Concept artist, Modeler, High Poly Modeler, TD/Rigger, Dynamics Rigger, Diffuse texture mapper, Shader texture mapper, Shader writer,animator or Motion capture editor, Asset Wrangler, and finally AI implementer and tuner. You know what that sounds like? the production pipeline of a Hollywood feature Special Effects house. Not that such is a bad thing... but... it effects employability and job security directly. Few artists have good heads for math or programming. Artist Programmers are rare, and tend to get the best jobs, so the more common "art Trained" production person who does Modeling or Texturing becomes of less value and more common, TD's who understand animation and performance, will rise to the top, as will those that tinker with the shader makers. Now I do like to tinker a bit, in them, but command lines are the tools of the devil. Once upon a time they tried to make a synergy of games and movies and turned out crap like "night Trap", but now, the budgets, the pipelines, and the corporate risk aversion due to those budgets are resembling Hollywood down to the print ads in Variety, and the exotic cars in the executive parking lots...
Is that a good thing or something to be worried about?
Scott
Actually they said they wanted to show the world of the young Link in a different tone than the world of the old Link. Whether that's ass-covering I don't know but the explaination does make sense.
I'm pretty sure what we're witnessing is the slow evolution into Hollywood. We're the old style art critics that gets more and more pushed aside by the mindless mass-media consumer. Because we don't bring in the money. Capitalism is wonderful, isn't it?
As far as the lack of wild design and style, I think most of you are wrong in that regard. Sure, for the next few years or even decade, realism will definitely be practically everywhere. However, I think that as technology evolves, more and more stylized games will be developed. Hell, look at hollywood. Movies have the best gfx of any media if they want it, yet pretty much none of the cg movies that have come out have been trying for realism.
Regardless, we'll wait and see
I'm pretty sure what we're witnessing is the slow evolution into Hollywood. We're the old style art critics that gets more and more pushed aside by the mindless mass-media consumer. Because we don't bring in the money. Capitalism is wonderful, isn't it?
[/ QUOTE ]
Well I love it as it keeps me employed and entertained and able to buy what I like.
Scott
http://xbox360.ign.com/articles/636/636132p1.html
for anyone working on nextgen titles, is this reasonable/possible polycount or just PR bs to hype the public? i'm thinking its the latter...
Scott
Nabiev Omar