problems with next gen games?

[SHEPEIRO]

probbably the hours of work for each asset going through the roof.

if you hope to give the same level of attention to each vertex, and pixel, you did in last gen, then id say 10x the modelling time and 4x textureing time. so a week long current gen model (a day modelling and 4 days texturing) translates into a five week long model for next-gen (at most effecient bu8t with extra complexity....)

but this is using "oldskool" hand drawing methods(textures) i think that you will see alot more, baking of lighting etc to compensate and speed up the texturing proccess, but this will lose some of the artistry, but with the prevalent drive for reality.....

so learning to use and to overcome some of the niggles of these systems will probably be the utmost importance.

[TheNeverman]

Agreed.

The content creation side of games even on current gen engines has gotten significantly more time consuming than in the past few years. Somewhere sacrifices will have to be made unfortunetly to keep the projects cost effective, most likely as allready mentioned, 'automating' some of what was previously done by hand.

[Dark Soldier]

well kinda for school , well just my education ingeneral, i want to be game artist , so witht this whole new generation of game production, i want to keep up to date, learning zbrush, getting better at creating normal maps , trying to undertsand how the the shaders works, theres just not much out there , all very hush hush.

i wanted to know poly counts , texture sizes , problems with game engines like unreal 3 and what kind of textures i can have in the shaders , any thing changed with was kind of rigs can be used, more animation cycles? anything should know about modelling, lighting(hdr) ? FX capabilities/problems?

these things are all coming up in the next game conference and any similar shows , but no help if your not in industry

thanks

[SHEPEIRO]

not working in the industry (yet) i cant give you deffinate info, but i can give you a few pointers as to what i know.

poly counts, anywhere between present gen, to 100k for things like characters and vehicles, but the most important aspect i think is more LODs (more work)

texture maps, i havent heard of anyone using greater than 2048* but thats big enough! most models will probably use a colour(pos wid alpha), specular, normal, and maybe an extra illumination, reflection map mask, or sub surface scattering (skin)map.

i think getting to know how maps effect the look of an object is important. try searching for some free Direct x shaders (there was a particularrly good skin shader on these pages a while back) and have a play!

[Dark Soldier]

thanks alot

that helps a little, any is much appriciated. i wish more people would get involved , its very important topic

[kaylon]

The question of what to use..poly..textures..shaders..etc is impossible to answer, as it changes from project to project...a Xbox 360 boxing game or even a FPS set in confined spaces could have 50k poly characters...use 2048x2048 skins on the characters and 1024x1024 textures on the environments with hundereds of shaders...this is cos you only have a few things on screen..

But..a free roaming city sized game or a large open field based rpg will have to use 256x256 textures...5k poly for characters and lots of creative planning to keep it looking next gen.

Next Gen is not as Next Gen as people think...it's all situational.

Learn how to do things as low as possible...it's easier to add..then to take away.

And the main prob with Next Gen....is all the games are starting to look the same..lets get some style going folks..lets be different.

K.

[WAcky]

very well said.

I think you're right in saying that the 'next gen' isnt just going to happen tomorrow. There are still going to be a multitude of games still using much lower spec art work. The range will be huge and there will still be alot of places around for artists that arent specialised in extreme high poly art work.

[SHEPEIRO]

i agree whole heartedly with Kaylon, the range of object "quality" (specs wise) is going up but there will always be low poly stuff, and not always high so as long as you make your models within the boundrys and have good reasoning behind its use of specs then it should be good for portfolio.

models produced using these new techniques are going to look similar at first (due to companys, or marketing wanting to emulate the latest trend, and artists learning off other artists and the "me too" factor) and it will teake a while for new styles to emerge from them. IMO

there will always be portables aswell, and with nintendos latest trend for not letting them die will keep low poly artists in jobs for years.

[Brownboot]

A rep from EA did a demo here at my college the other day. He said a lot of game design kids are asking what poly count people are looking for in a demo reel these days and he said "It doesn't matter how high you go anymore." But he did say it was important to not have exclusively low poly stuff. He also said don't put your best work at the end, put it at the beginning because odds are it won't get watched all the way to the end during the first cut phase. Unfortunately he was just a PR guy and the animator they sent wasn't too big on texture sizes etc, he only wanted to talk about mo cap.

[Dark Soldier]

very good point

refreshing games like okami or unlimited saga going for the different look, definetly was developers who try to do more. but doesn't this next generation allow this even more so with shaders instead of just texture mapping.

i understand your points about poly count, i didn't want to know the llimits so i could be inefficent and just aim to stay just under them like i've seem alot of work, my aim is always to go as low as possible but still keeping to my designs, and then add, i always do , even with high poly work its good practice.

any one know much about shaders?

[kaylon]

Shaders are usually specific to the project also...Our DX9 shaders are custom written by our chaps here and they then tell us how to use them (on a tech point of view) ...we use lots of multi-pass custom stuff...but also normal maps and spec...so play with that sorta stuff...again all you have to use are Max based things...but it's all worth a go.

[makkE]

As a hobbyist, I can only say that the next gen is both fascinating and a little scary.
I was glad the engine I started modeling and skinning for was very simple. Only colour maps, not even specular maps, and low polycounts, low resolutions.
If I had started on current gen or would even imagine starting on next gen, I believe things would have been pretty overwhelming.
I think it will get ever harder to get into the whole thing with the years, especially for spare time modders.
In the mod scene, the audience is starting to expect modders to keep up with the developements, so I guess there´ll be less mods/hobby projects getting finished in the future, because ever less people will be up to it, or simply lack the spare time needed.

[Dark Soldier]

thanks kaylon

u've been very helpful today. about the multipass stuff do u mean just in the shaders on with camera tricks too, with shadow of colossus they used multipass effects on the camera, or did u mean both?

this maybe stupid question , but what file format at textures usually, jpeg, tga or does it depend on game engine, is there custom types?



thanks

[GLandolina]

thanks kaylon

u've been very helpful today. about the multipass stuff do u mean just in the shaders on with camera tricks too, with shadow of colossus they used multipass effects on the camera, or did u mean both?

this maybe stupid question , but what file format at textures usually, jpeg, tga or does it depend on game engine, is there custom types?



thanks


a couple of engines i've used took TIFFs and spat out a dx format on compile,
but generally for portfolio stuff i'd reccomend something like png which is lossless and supports alpha.

[ts-falcon]

hey guys this is an interseting thread, just thought id share some thoughts. 1st of all people who are saying next gen can push out 50k-100k poly characters, maybe if its a single character onscreen doing a simple rotation. the specs given to me be modelers from EA and blackbox are : 16k polys for cars such as the ones in project gotham 3, and 9-12 k for characters, lower poly counts if normal mapping is being used as it is a major system hog. next time you play a next gen game look at the outline of a character and you'll see some rough edges. normal mapping is awesome but doesnt, best example ive ever seen is doom 3 (not next gen i know) those characters are only about 2.5k polys, if that in some cases but look awesome head on due to normal mapping and texture work. but up against a light their heads are boxy as hell. anyways hope this info is useful guys!

[Headless]

If you're starting out and you're looking into what to learn to keep up with next gen art, worry alot less about technological limits and just learn to be a good artist. Sure having good topology and efficient texture usage is important, but the best game artists are always the ones who have the core art skills down. Learn good observational skills, learn to be creative, learn to be patient with your work, take time paying attention to detail, and learn good use of color.

If you can get that stuff down and just keep practicing your art skills in general, you shouldn't have any problems, as those are the skills that differentiate the good game artists from the bad ones, regardless of technological advances.

[Onikudaki]

As far as what I have been reading the nex gen games are not going to be much more in the modeling department than before even less work. The models themselvs are not that high poly. Its the normal maps that have the HIGH poly count, but when used with like zbrush creating an extremely detailed mesh from a really low poly model takes a 10th of the time as appose to modeling lets say in max a high poly model.

I think yet, using this technique will extend to creating all of the statice meshes aswell and if you want a lot of stuff your going to have to model it.

I honestly dont think it will be nearly as difficult as people are making it out to be.

[urgaffel]

It's not difficult, just time consuming.

[AndrewRaZ]

Learn how to do things as low as possible...it's easier to add..then to take away.
K.

when i went to school, professors told us exactly the opposite. i majored in game development, but took as many film courses as i could get in. professors in both departments always pushed us to take our work past "game art" because it's easier to ramp down the detail later than it is to increase it. every studio i talked to said the same thing, and all but one told us they didn't even want to see poly edges any more; it should be high poly or subdivided at render time, with a preference toward the former.

game studios are pulling from film more and more, not just personnel, but also techniques and technology. they want to see film-quality work more than "gamey" work.

one thing they did specify, however, is that there are stylistic differences between film and games and how things are portrayed. they still want to see that stylistic differentiation, but at film quality.

that's my east-coast experience.

[braincell84]

The main problem for me right now is getting an engine with a sort of unified material property editor. I mean you need to simulate different materials with different combinations of shaders, normal maps, specular maps, refraction, reflection, attenuation etc. The hard part is that different shaders do this and it takes effort to combine shaders to have a combination of different properties. I mean one shader will usually combine a normal map with reflections and add specular map, but you cant tone down reflections and add diffusion so easily. Its all "hardcoded", and often games today still look plastic because of too much reliance on specularity. I find this the hardest, at the moment at least. I would like it if we had a more optimised way of making shader-specific materials as easily as we can make materials in a 3d package such as 3ds max and not worry about the actual shader code (ie HLSL) and how it combines. Thats one problem on my mind right now :P but there are bigger ones but they're mostly specific to the engine you're using, or other limitations of your project. Lots of very specific things.

[whooosh]

Here's my take since I'm working on a Next Gen title..
1 Victorian SF style house would take:
1000-2500 poly all textures are 2048x2048

1 day modeling
1.5 Days Unwrapping 2 UV channels (1 for color/diffuse, 2 for lightmap/dirt/Ambient Occlusion)
1.5 days to texture..this includes; 1 x Color/Diffuse, 1x Spec, and 4 x alphas (Shader purposes)
2-3 days of NormalMapping. Weather it be modeling high detail or using NVtools for photoshop.

In the end, we have about 7x2048x2048 unique textures.

to put it simply, it all just take too long. Art director always seem to think we can do it faster....now that we're in crunch, he'll have to get his hands dirty. We'll see! muwahahaha.

"Feel the pain!" Zim

[neonneo]

Im not in the industry so this would be a student question.
If people want next gen, and companies want us to create next gen games shouldn't that equate to higher pay or atleast longer production time for games? I can't imagine crunch times on this titles, must be intense.

Though the only 'next gen' games i've seen are
Gears of War
Huxley
Project offset
Fight night
new Unreal

[SPIDER2544]

biggest thing im seeing, is that the game industry is finaly afforded the oportunity to be basicaly using the same pipelines that a lot of film companys use. from 3d scans, to Zbrush, to using photograhs to gain photo real textures.

it is going to be tuff for people to be doing just the old school hand painted textures the whole time, because it just takes WAY to long to do. why bother when you can take a photograph and just remove a lot of the lighting information and start to tweak it to fit your needs. for me the camera is just another tool in that respect.

[SHEPEIRO]

its getting very interesting in here,

i m quite intrigued by the idea of braincell84s material editor. with the engines ive worked on in MODs its been a case of make the model , make the maps send it to the coders to get it in game, and get the results, as it is i have very little control over what the shaders look like, so it would be excellent to get the control something like the MAX material editor affords you.

as to the question of learing downwards or upwards surely that can be personnel choice and the result, working in different parts of the industy. as long as you have those base art skills

[urgaffel]

When it comes to shaders, if you have a good graphics coder he/she can give you either an .fx file (or several for different materials) that you can use inside max (or other apps that can load shaders) or write a custom material plugin. Just look at the most excellent JISkin shader avaliable for free. It's a .fx file that you just load into the dx9 material in the material editor and you get to see the effects in realtime, no hassle. Regarding specularity, just make sure the coder includes controls for adjusting specularity :p

Lastly, you don't need a material editor but it sure is nice having a material editor that can preview realtime shaders. But it takes time implementing something like Rendermonkey into an engine/tool pipeline so for those that are short on time, using .fx files in max is a faster way of doing it. You lose some control, but you can get the same instant feedback.

As for the whole what to put on a demoreel, if you can model a highpoly character/prop AND export the normalmaps and make a cool, good lookig lowpoly model (preferably with textures) you have shown that you can do both which will make you pretty dang interesting to companies. Also, there will be a lot of room for environment/prop positions since someone has to model all those houses/guns/artillery pieces/vehicles/whatever that the characters run around in/with. If you can do both characteres and props and environments, you're pretty much set heh

[RmachucaA]

The biggest problems i have seen are all in EA games, and that is interpenetrating poly's on character's clothes.... looks HORRIBLE imho. Look at any fight night vid, look at the trunks, or the nba game.

[NikLG]

Im not in the industry so this would be a student question.
If people want next gen, and companies want us to create next gen games shouldn't that equate to higher pay or atleast longer production time for games? I can't imagine crunch times on this titles, must be intense.

Though the only 'next gen' games i've seen are
Gears of War
Huxley
Project offset
Fight night
new Unreal

In my experience ( 15 years...lots of 'next gen' moments ) that would be a no.
Dev times may extend initially but give it 18 months and publishers / producers will be trying to claw back the time.
As far as pay goes, you are still 'an artist' ( or whatever ) in the eyes of the people who pay your wages ( or, if you're really unlucky, a 'seat' ) so you get the same deal.

Nik

[Tweek777]

The biggest problems i have seen are all in EA games, and that is interpenetrating poly's on character's clothes.... looks HORRIBLE imho. Look at any fight night vid, look at the trunks, or the nba game.

thats is really subjective, personally, i dont like them either. i also really hate animations where there isnt enough geometry, resulting in strange deformations.

however when i talk to friends about things like that, just regular gamers, not artists, they tell me that those are things that they pay no attention at all to, since theyre too busy playing the game.

[SHEPEIRO]

i find alot of the most inspiring to look at games have those glitches, its normally because they are trying to do alot with the geometry they have, the animated clothing adds such an extra layer of visual richness that it is better than say rigid clothing that always looks good in screen shots. but technically it would look better without, them but as player, the efeect of everything moving more naturally is much better than the (usually) small or momentary glitch of clothing intersecting bodys.

I think animation is goingt to become interesting when devs start to use some of the extra power for simulating effects like these rather than adding some more polys to a model. looking at current selection of "next gen" titles they all look amazing still, but when they move they are let down by those old "gamey" jolts from one anim loop to another.

dont think ive seen one title thats impressed me as much in motion as it has still!

[orko60]

Very interesting thread. I appreciate what you veterans are saying, as I am working on a reel to send to companies later this year. I'd like to do environment/character modelling, but I have a question. For these next-gen games, my texturing isn't as good as my modelling. Should I be sticking with non-textured objects most of the time? I mean, they're "okay," but not as professional as I'd like it to be. Any thoughts?

-Zack

[urgaffel]

Well the old advice of showing only your best work still stands so... Unless you can texture, don't or if you feel the need, put it at the end of the reel (best works first ;)). You can put it in your coverletter or bring it up during the interview that you know some texturing (and maybe you want to learn more?) and could pitch in in a pinch. If you're not interested in texturing, don't bring it up at all :) If you are interested in texturing, work on it as much as possible so your work gets up to pro or at least near-pro level.

[whooosh]

Work harder on your texturing skills =)

Most, if not all, game companies do not have specific positions just for modelers. Especially not for environments. If you're an environment artist, you're expected to go from beginning to finish. Model(hi and low), UV, Texture, sometimes, you would even have to produce the concepts. I would recommend you go and ask companies for tests. Do those tests and put it in your reel instead. I just recently hired someone because of the test his submitted to Blizzard South.

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.