In archvis, our 3dsmax units would be mm, and my display unit scale would be in metres. I'm attempting to model 'a church' typical fare for my portfolio, and Ive adopted mm and metres, (since I'm used to it, and its real world scale) Ive kept my model to the max grid as close as poss, Should I be modelling to different units, or are these units defined by the Unreal editor ?? for example.
The size of units is all relative to a characters size. For example GOW uses a character roughly 96 units high. This means that one foot = 16 units. However I have worked on games where the character is 180 units high. This is the same 'real world' height but in centimeters. (180cm is roughly 6ft)
All your environment assets should match up to the characters height. Benchtops should be waist high and doors should be a bit taller than the character. What you set your units to in max isn't that important as long as the models match the character models in scale.
Standard sizes tend to be dictated by the scale of the characters, the speed of the rate of movement, the type of gameplay mechanics, method of interaction with the world.
Gears has some established heights of gameplay devices, a set height maximum for steps ( though its worth noting that steps look and FEEL better to walk on in Gears when they are wider than they need to be so I build that way rather than to the exact limits, I aim below them so its a graceful movement), set height for cover blocks, door handles etc.
It's not simply about how big a thing is relative to the character, you could waste a lot of time trying to replicate perfect scale in a fast paced game like UT where the world scale drowns the player height in order to compensate for the very fast rate of travel through the world.
Gears on the other hand, makes you feel more manly and implies you are bigger and tougher than the world so you take your time progressing through it, which is great for an artist as it allows those playing the game , the time required to appreciate the work you put in.
Modular presets, I always try to provide a straight piece, short, medium, long , a curved piece in the same style , intersections of L X T shapes. Or something along those lines, you need to know what you are trying to accomplish, its not easy to explain, its better you try doing it and learn from screwing it up a little.
Every modular piece I made for the first couple of years was wrong. I could not see how it was wrong until I took my best guess at what was required, made it, realized I didnt anticipate a lot of things.
EQ is spot on in his points about the workflow I showed being something you can take part of and apply it to your own way. I often do the other method, I only rough the bricks in zbrush and then overlay fine and sharp detail on to them in crazy bump, sometimes I just tile a texture in photoshop and then crazybump it when its something plain, sometimes I mix and match. It all depends.
What does it depend on? Try each method and find out. It's the only way, you can't play pick and choose through guesswork, you need to try and fail and understand why you messed up so you can do it better next time. It's worth doing this because you are working toward building an awareness of different methods of doing the same thing so that each new task you come at, you can sit and consider which to use.
At the end of the day, modelling is about brute force. At some point we all drop any of the semi fancy tricks we know and get our hands dirty by eyeballing verts and trying to massage kinks out of a shape we go too ambitious with, or changed our mind about mid production, or were asked by our leads to change.
It's not finesse that takes us to the finishing line, its just raw determination and bashing away at things until they are in the right shape.
For instance, model a 48 sided sphere,cut one half away and try to thin out the edgeloops by a factor of 2 as you near the center where everything gets compressed. This is a good learning excercise, its useful in order to thin polygon density as forms compress, its useful for takeing sphere shapes into Zbrush and placing the density only where we need it, its good for understanding more about gracefull terminating edgeloops.
Initially its hard to do, eventually you know 20 different ways to do it and a handful of different situations in which one way is better than the other.
Tiling plain BSP walls in concrete materials can be done many different ways, sometimes its easiest to just use crazy bump, when you want something specific you can use projection master and render out of Zbrush direct, or you can stamp concrete alphas on to it with the tiling wrap around functions on a regular plane.
Often that warps the perimter, a quick dirty way around that is to just duplicate part of the mesh at the edges back in max and move it across to hide the seams, so you are sort of doing Filter Offset by hand. I tend only to do this if I was a very particular effect.
TK: You make a good point, I don't know how much of a factor this stuff truly is in the decline of editing work in the community but I always suspected its more to do with texturing no longer being something you can just see when it looks correct in photoshop and modelling a character takes SOOooooo much longer and requires more resources.
The Eat3D thing fell through, contract stuff didnt get resolved is all I know. I'll eventually do more tutorial stuff, but I'm tasked officially with doing that stuff at work atm so I have no incentive to be duplicating it for the community in the evenings. Plus... that stuff takes a serious amount of time to do right, you end up processing your recording, processing it again into the format you want, rewatching it while recording audio, then reprocessing that and so on.
Coffee? Normally 2 sugars and coffee, though this morning I won the King of Dumb award for hitting the buttons on the coffee machine at work, walking to get milk while it goes to work, fiddle in the background while listen to it pour, then when I went to get my coffee it wasn't there and I couldn't understand it.
Eventually I realized the one thing I forgot to do was put the cup there waiting under the faucet part.... and that was only because I heard a splash sound under my foot and realized where my coffee went.... duh...
I remember someone asked about crashes, yeah I get crashes. I used to get them most of the time for the first year while I pulled my hair up, eventually you spot the patterns and work around the things that cause them. It is natural to get them, as natural as it is to get all kinds of normal errors for a good while until you get more ninja like with the workflow.
Are those modeled in max and then placed along a spline, or path etc. And then brought in to zbrush? Or do you make those details in zbrush and duplicate around? If so, how did you duplicate them in zbrush so nicely spaced and along the path? Or is this a 'brute force' example where you just tweak the placement of each one to get the overall correct look?
I took the lazier road with the leafs, I had modelled a single leave and zbrushed in some other piece, so I took that, applied a bend modifier, symmetry, tired it, wrapped it around a bud shape and called it done. I then planced that one deco piece on the grid, lined up with the arch behind and then just instanced it up with grid snapping, then centered the pivot on the X axis to 0 and rotated around what I knew from previously to be the central pivot point of the arch.
Then I did the same with one clamp I had not zbrushed, did the above steps, but later I attached a smaller clamp on the inner most arch rim before you meet the metal. I then selected the bigger clamp and attched the smaller one to it, so the changes i made to one were duplicated around again, this time without my needing to place them by hand.
Actually, I think that the only thing I modelled fresh in this scene was the metal frame work around the glass, the rest is all old kitbash files that are instanced. I think I took 2 days on it, possibly 3 including the lows.
if im not mistaken, in GOW2 you had destructable environments to an extent. How did that affect the modelling? Where you still able to bake down objects such as wall and floor tiles to single planes or did anything destructable have to be a 3d shape with depth?
Hey Kevin,Thanks for the brief answers u gave. I just wanted to know about the usage of Zmapper in zbush in your workflow. Do you use em? or you totally rely on Max's processing in generating all the maps?.
Also can you share with us the render settings you use to bake out the maps,i know,you use lighttracer with a sky light,other than this,what render settings do you change, for instance like the AA filters,sample sizes,etc.
Do you use Mentel ray for any of the renderings or it is just the standard scanline renders all the time?
if im not mistaken, in GOW2 you had destructable environments to an extent. How did that affect the modelling? Where you still able to bake down objects such as wall and floor tiles to single planes or did anything destructable have to be a 3d shape with depth?
You still make your asset the way you normally would. A tool called "fracture" in UE3.5 will boolean chunk it up for you.
That's a shame the Eat3D thing fell through Kevin...I was looking forward to that. Maybe tutorials could still be sold directly thru Epic?
if im not mistaken, in GOW2 you had destructable environments to an extent. How did that affect the modelling? Where you still able to bake down objects such as wall and floor tiles to single planes or did anything destructable have to be a 3d shape with depth?
Anything destructable has to have a 3d shape if you want to have chunks like they did in Gears 2. Actually if you look closely at the pillars or other pieces that break apart in game before you shoot them you can see where they are actually cut up.
That's a shame the Eat3D thing fell through Kevin...I was looking forward to that. Maybe tutorials could still be sold directly thru Epic?
Funny how fast things can change. The talks have been resurrected! I can't really say much more than that though, but the DVD now has a good chance of being produced. Stay tuned!
Oh and amazing work Kevin, you are an inspiration to us all!
I haven't forgotten about the thread, I will update more. I'm just busy with vid tutorial documentation of my process at work and I've gotten really obsessive with the theme I'm working on atm. Usually I drop things for a while when like this but I'll get back soon with a big update.
...its not easy to explain, its better you try doing it and learn from screwing it up a little.
I think that is one of the most accurate, and modest, things I've heard any artist say. I really can't ask anything that hasn't already been mentioned, but this thread has just become a .PDF for technique reference.
What does it depend on? Try each method and find out. It's the only way, you can't play pick and choose through guesswork, you need to try and fail and understand why you messed up so you can do it better next time. It's worth doing this because you are working toward building an awareness of different methods of doing the same thing so that each new task you come at, you can sit and consider which to use.
I really wish more game artists were more like you!! It's kindof sad to see so many people not experimenting at all, just waiting for stuff to be shown to them from A to Z.
I think it's mostly because many artists don't get their hands dirty enough with engine exports and tweaks, and just stop at the fancy render stage.
There are soooo many cool approaches available now, this thread being such a great proof of that!!
Pior: Thanks, you just made my day. I worry so much over my word use and picture people imagining me as some puffed up pontificating type that thinks he knows it all.
Its easier for most artists to let the work talk I think.
Pior: Thanks, you just made my day. I worry so much over my word use and picture people imagining me as some puffed up pontificating type that thinks he knows it all.
Its easier for most artists to let the work talk I think.
Although I think the 'let the work talk' is a good point, someone who is articulate and has excellent communications skills PLUS examples of said work makes for a great mentor or teacher. And definitely someone to look up to when it comes to their craft. I'm pretty sure we're all in agreeance that your modesty hasn't made you 'some puffed up pontificating' person.
I realize that sounds like a real kiss-ass thing to say, but I definitely stand by it.
Freakin sweet, I love it. Do you model most of the pieces from concepts that are given, or do they just say "give me an arch" and you dive in? How many variations do you go through before you decide on a final?
RDM: We get scene concepts and we identify ourselves often which elements would be the most useful and worthwhile to build to allow the LD's the tools they need to make the scene. Sometimes that means we have to build things that are not in the concept, or rather the things no one notices in the concept or cares about, like plain walls and floors, other times it just means we research similar artifacts from the real world and use those as a guide when deciding how to add more clarity to things.
Concepts are great, they get everyone on the same page, they unify and open up communication between departments, but they are not blueprints.... you always have to research the thing a bit.
I didnt really do variations, I just made a basic cylinder or block, subdivided and took it into zbrush and worked out the design there. I don't really do the variations, think about it , pick the best, finish it out and then call it done 'thing'.
I take my best guess, model it, call it done. When I've done it wrong, some one tells me, but until I get that tap on the shoulder I just crank away without really taking the time for variations because our schedules never really allow for that kind of ponderous measures thought.
The variations you refer to come through my refining the theme as I go forward between assets, each improves and refines and becomes more sure of what it is until I reach my zenith with some kind of show stopper piece near the end of production on each theme that leaves me proud of what the set evolved into and sad that I have to let it go and move on to the next unknown, awkwardly trying to work out what it is again.
The worm pillar and the arches were earlier on in the Locust theme's production, the forms are still a little too soft, not jaggy enough yet, I'm still vague on what I'm doing, trying to do and the way im trying to realize it. The Block pillar and the floors came at the halfway stage, by then I was more used to Zbrush, working with alpha stamps, getting down more on the sharper edges and doing a lot more confident clay tool work by hand to turn the softer and plainer alpha patterns into newer or rougher unique forms in each instance they were used.
Amazing work Kevin! Your posts and the images within them are some of the most helpful tools to have as a student. I look forward to seeing whats next.
Sampson: I think polycruncher does better with surfaces that are detailed roughly all over, like the Locust stuff was well dealt with because all the flat forms had some sort of soft to hard undulation with some sharp deep cracks and gouged concrete like parts.
Super subtle detail on top of a very flat surface tends to fare less well, like if you were doing those fine pock mark indents like you see on limestone or certain treated concrete / plaster walls. Also thin scratches on metal forms do poorly too.
What I've seen happening is that it deals poorly with the flat areas, it concentrates simply on saving the detail areas and stretches very few polys inbetween them which leads to an ugly washed ou stretchy look. Sometimes when its very fine detail you do only, that gets washed out too.
The trick seems to be to balance the forms, try to always have some sort of medium sized soft detail like subtle bending on metal as you get to the corners, the same way you place fall off slopes from a tight pinch on clothing for instance.
With concrete, a few well placed deeper crevices that lead into those long fine cracks works better, its like polycruncher needs the medium forms in order to understand how to link all the little details
maaan cool stuff. seen some of that in pixologic interview before but its still awesome i need to work more with zbrush moooore hehe. Nice that you take the time to answer so many question. thanks
But i have some questions for you,i have read the entire tread but i might have missed if someone asked the same questions.
First i have 3ds max 2009 with pro optimizer,I suppose its the same as Poly cruncher.
Do you use poly cruncher when you striped down the high poly mesh?
I cant work with it because it messes up the mesh very badly,flips the faces and no real control over different areas in the mesh.
places like corners and sides where i want a sharp design (hard surface) tends to get messy after using pro optimizer!It often leave ALOT of unnecessary faces there.
Not sure how you do it but i export a medium res model to 3ds max and run pro optimizer.
Then as i said before i get a lot of flipped faces but i need the geometry.After that i import my high res model and run a render to texture and render out all the maps i need.
The real issue i have is when importing a 10 million poly mesh into 3ds max,Best would be to render out the maps in Mudbox/Z-brush to avoid such highpoly mesh in Max but then i get issues with uv,s and shit.
Anyway i have to read everything one more time i think
Well you should not be importing a 10 million poly mesh, you should be exporting in chunks from Zbrush that are reported as being about 250,000 polys which is half a million Tris.
If your object is more than 250k, then use polygroups to seperate the piece into 250k chunks and export each of them out as individual OBJ's and then once they are all exported to a folder, just run polycrunchers batch optimize function and point to that folder and it will optimize everything there as much as you desire.
Pro optimizer mostly isnt very good for this stuff, I have tried it, some others here have too, we use polycruncher.
When having problems with retaining the detail you want, you can optimize less severly for those sections, polycruncher will do incremental optimizations , like a 25%, 50% and 75 % optimized version of each OBJ. That way you can import all the ones optimized by 75% initially and swap out for 50% optimized versions where you need to.
Sometimes you are going to be driven by this to make 2 or 3 process files because of the polycount. Hope that helps.
This piece is from the Outpost level, probably my favourite level in Gears2.
If you note the subtle banding on the large open door edges, I did that with a clay brush and the square alpha 28 set to a very low 20 in strength. I did this to avoid exactly the problem TheWiruz was wondering about, having some sort of subtle undulation underneath whatever fine detail you put on top like scratches is a good way to help reinforce the importance of the whole mesh to polycruncher
For someone who is trying to get into doing more high poly assets for baking purposes do you recommend learning how to do HP work in max first and then learning zbrush or should you try and learn each hand in hand?
Is it better to learn hard surface modeling and then learning how to zbrush in even more detail later or.
hey kevin,
being at epic your going to have concepts rushing in for you to make through the doors, do you have any advice for having a big case of modellers block and nothing really to work off.. (heh) are there any artists you look up to?
Autocon: try both, try different things in each package, do your organics in zbrush like rocks and more freeform designs, try more panel based designs in max and just gradually start working out how to take a manageable collection of chunks into zbrush and bring em back to zbrush.
Just dont try to do everything at once, pick your battles, marshal your forces like Sauron for a while
Sampson: They don't really rush in like you probably imagine, theres a great deal of research gathering and inspiration building required. We get scene concepts that are quite loose and most of the time they don't quit fit the way a level develops due to gameplay changes often months or weeks after the concepts are done. I mean its always down to you as an artist to find your own hook and then clarify it.
But theres a whole world history of art and inspiration out there and perhaps its my awareness of that, that prevents me from ever really getting artists block. I guess I look up to everyone who ever built something cool in the real world or made a sexy looking movie or wrote an inspiring story, these are the things I tend to look up to.
I'm a fan of weird things, like Hawksmoor who was responsible for a lot of cool buildings in London and in general its the dead artists long before my time that invented or pioneered a lot of the building plans and art movements I routinely steal from. The people that worked all this out for us to rip from are were very clever and I like reading about their ideas, the way they subverted the goals of the people with the money and transplanted them with their own dreams without them knowing.
I just love the story behind many of the pieces of art out there, just understanding how much planning and thought went into real world constructions is a huge inspiration to me.
Love the vault door thing, I think that's one of my favourites in this thread (although the dome a few pages back is great too). Really chunky and solid, I also like how the concrete is mirrored without being hugely obvious about it (maybe we need to redefine "the rorshach effect"?)
The rock stuff is cool too. Very natural and a nice blend of soft and sharp shapes.
I think I prefer the earlier stuff to some of this locust architecture - mainly because some of the locust pieces seem a bit "soft" to me in general.
Out of interest, for these more organic pieces do you tend to use ZBrush "Layers" at all, or do you just sculpt everything in one pass and build it up that way?
Cheers for all the information and, of course, the delicious eye candy.
Yeah I always felt my Locust stuff was too soft, I was more reliant on Mike Buck texturing them right to save them. I felt as I went forward I got better at sharpening things, like with the craggy spiky trock thing.
I started off controling things more with layered passes and then found as I subtooled more it wouldnt let me use the layers so I just did it all in one. I'd just do it in stages though, moving up in res after blocking things at lower res.
I still feel the mirroring point is blatant in the door, makes me feel bad to look at it but i had to fit everything on one texture and I knew the shadowy corridor and monitors in that area would hide how obvious it looked ingame!
I just had to say kevin, your work is really inspirational to a new enviroment artist like me its nice to have a high bar set to reach, your work really is amazing i was going to ask for some modelling tips, but i noticed you already gave some to autocon lol very helpful. Keep up the good work! and i hope i can get some feedback from you when i start posting my own enviroment pieces
Hey I have to say this stuff is really inspiring me to try out more hipoly work, I'm always afraid it will suck up my time . I really like how you work very modular on these pieces, creating massive detail out of a few simple shapes basically. Most of this stuff looks like it really wouldnt take that long to make, you just need to think cleverly on how to build up the piece! I wonder if it's possible for you to post some of the final assets aswel? I don't have gears2 and dont really plan on buying it since its really not my type of game.
Awesome stuff Kevin, all the meshes have really rich detail. Congrats on Gears 2, loved playing it through, I remember taking notice of all the cool details on the vault in game as well.. nice to know where it came from. *high five
Hi Kevin, I am working on my dominance war model at the moment but as soon as I am done I am going to mash out some enviro bit and bobs using tips and ideas that you have posted here. My work flow for my DW was weaksauce since I didnt do enough research and I just sort of winged it... OOPS. Next time I will put countless days into preparing for my model. I will call it preproduction. HURRAY! thanks again cool dude.!
I love the feeling of wear on that vault door. The metal really looks like someone has been pounding on it to get into it for years.
What kind of texture space were you given for a piece like that? I'll admit I'm not too familiar with GoW2 so I don't know how big that is in game (I'd imagine pretty big).
And also if I may as a followup grid-style question. The locust arches on page 5 don't seem to really fit flush with anything. The mesh seems to be bubbly all over the place. how do you plan for lining pieces like that up with the straight walls or other pieces? As in do you know from the top there will have to be some overlap? Or that the player will never see the X face of an object so you don't bother with details.
Thanks again for sharing the candy and knowledge!
edit: Is that vault door concept available on the net somewhere? Seems to me I've seen it in the WYAWO thread from someone else a while back (an unzbrushed version by eraserhead I think).
toron: Just keep logging the hours , thats all thats involved, if you want to learn to be good at anything just start by being slow and awkward and make your way toward good.
Wiruz: Well, why would you need zbrush to model a lowpoly bumpy brick? It's easy to do again by hand and unwrap, or its easier still to think ahead and make a duplicate of your base bricks before subdividing and use those later with some modifications. Bricks aren't tough to model again and often you find more efficient ways to build something the 2nd and 3rd times.
Using optimizers to do it for you will always result in you needing to get your hands dirty and clean things up yourself. Especially when you are using so many unneeded polygons to match the impression of the hipoly.
We do better if we think of the lowpoly as soft putty, it needs only to be an impression of the hipoly and holds up when you walk around or rotate it. It's the processing of the HI to the LOW that hardens the lowpoly now, its not like before where we need to be exhaustively accurate to the boundaries of the design, think of it as something soft and pliable, it only needs to carry a single smoothing group well enough to catch all the normals off the HI during process.
Peris: Exactly, I'm not doing anything that complex here, the Necris stuff on UT3 is complex, it does take a lot of skill to make but the locust stuff in particular, blocks and cylinders that are equally subdivided is all you need to be capable of in 3d. After that , its good old reliable doodling skills in Zbrush3 where theres no rules and restrictions really
I think in general I took 1-2 days for each of the locust HIpoly meshs.
Kessler: haha, dont spend too long planning though, or your will be tired out halfway through the fun stuff
Saidin: Thanks, I put a lot into that door, there was no concept or guideline beyond ' we have a door we need to blow up, it'll be in the cinematic too' so I just grabbed one of my favourite doors from bank heist movies and made it more utilitarian. it was a single 1024 texture.
Well spotted on the Locust stuff, they are not gracefull modular pieces that fit flush like the manmade stuff, which was a concern to me initially. We worked around this by giving the designs a slight twist on the X axis so that when you put 2 together, they do fit semi flush with an organic feeling join that looked more naturalistc and sculpted artistically rather than like something we would make thats neat and ordered and clearly built by hand.
We also made sure all the designs were self sealing elements with no open edges or empty planes and less mirroring than normal so it would further communicate the notion that they'd been gradually worn away and made chiselled at haphazardly or stuff had grown on them.
Replies
The size of units is all relative to a characters size. For example GOW uses a character roughly 96 units high. This means that one foot = 16 units. However I have worked on games where the character is 180 units high. This is the same 'real world' height but in centimeters. (180cm is roughly 6ft)
All your environment assets should match up to the characters height. Benchtops should be waist high and doors should be a bit taller than the character. What you set your units to in max isn't that important as long as the models match the character models in scale.
Gears has some established heights of gameplay devices, a set height maximum for steps ( though its worth noting that steps look and FEEL better to walk on in Gears when they are wider than they need to be so I build that way rather than to the exact limits, I aim below them so its a graceful movement), set height for cover blocks, door handles etc.
It's not simply about how big a thing is relative to the character, you could waste a lot of time trying to replicate perfect scale in a fast paced game like UT where the world scale drowns the player height in order to compensate for the very fast rate of travel through the world.
Gears on the other hand, makes you feel more manly and implies you are bigger and tougher than the world so you take your time progressing through it, which is great for an artist as it allows those playing the game , the time required to appreciate the work you put in.
Modular presets, I always try to provide a straight piece, short, medium, long , a curved piece in the same style , intersections of L X T shapes. Or something along those lines, you need to know what you are trying to accomplish, its not easy to explain, its better you try doing it and learn from screwing it up a little.
Every modular piece I made for the first couple of years was wrong. I could not see how it was wrong until I took my best guess at what was required, made it, realized I didnt anticipate a lot of things.
EQ is spot on in his points about the workflow I showed being something you can take part of and apply it to your own way. I often do the other method, I only rough the bricks in zbrush and then overlay fine and sharp detail on to them in crazy bump, sometimes I just tile a texture in photoshop and then crazybump it when its something plain, sometimes I mix and match. It all depends.
What does it depend on? Try each method and find out. It's the only way, you can't play pick and choose through guesswork, you need to try and fail and understand why you messed up so you can do it better next time. It's worth doing this because you are working toward building an awareness of different methods of doing the same thing so that each new task you come at, you can sit and consider which to use.
At the end of the day, modelling is about brute force. At some point we all drop any of the semi fancy tricks we know and get our hands dirty by eyeballing verts and trying to massage kinks out of a shape we go too ambitious with, or changed our mind about mid production, or were asked by our leads to change.
It's not finesse that takes us to the finishing line, its just raw determination and bashing away at things until they are in the right shape.
For instance, model a 48 sided sphere,cut one half away and try to thin out the edgeloops by a factor of 2 as you near the center where everything gets compressed. This is a good learning excercise, its useful in order to thin polygon density as forms compress, its useful for takeing sphere shapes into Zbrush and placing the density only where we need it, its good for understanding more about gracefull terminating edgeloops.
Initially its hard to do, eventually you know 20 different ways to do it and a handful of different situations in which one way is better than the other.
Tiling plain BSP walls in concrete materials can be done many different ways, sometimes its easiest to just use crazy bump, when you want something specific you can use projection master and render out of Zbrush direct, or you can stamp concrete alphas on to it with the tiling wrap around functions on a regular plane.
Often that warps the perimter, a quick dirty way around that is to just duplicate part of the mesh at the edges back in max and move it across to hide the seams, so you are sort of doing Filter Offset by hand. I tend only to do this if I was a very particular effect.
TK: You make a good point, I don't know how much of a factor this stuff truly is in the decline of editing work in the community but I always suspected its more to do with texturing no longer being something you can just see when it looks correct in photoshop and modelling a character takes SOOooooo much longer and requires more resources.
The Eat3D thing fell through, contract stuff didnt get resolved is all I know. I'll eventually do more tutorial stuff, but I'm tasked officially with doing that stuff at work atm so I have no incentive to be duplicating it for the community in the evenings. Plus... that stuff takes a serious amount of time to do right, you end up processing your recording, processing it again into the format you want, rewatching it while recording audio, then reprocessing that and so on.
Coffee? Normally 2 sugars and coffee, though this morning I won the King of Dumb award for hitting the buttons on the coffee machine at work, walking to get milk while it goes to work, fiddle in the background while listen to it pour, then when I went to get my coffee it wasn't there and I couldn't understand it.
Eventually I realized the one thing I forgot to do was put the cup there waiting under the faucet part.... and that was only because I heard a splash sound under my foot and realized where my coffee went.... duh...
I remember someone asked about crashes, yeah I get crashes. I used to get them most of the time for the first year while I pulled my hair up, eventually you spot the patterns and work around the things that cause them. It is natural to get them, as natural as it is to get all kinds of normal errors for a good while until you get more ninja like with the workflow.
A quick question about instancing. Less about the bricks and more about the small trim details. Say the little leaf bits that go all around the inside of this archway.
http://www.kevinjohnstone.com//Images/Gears2/COG_Hospital/COG_Hospital_Arch.JPG
Are those modeled in max and then placed along a spline, or path etc. And then brought in to zbrush? Or do you make those details in zbrush and duplicate around? If so, how did you duplicate them in zbrush so nicely spaced and along the path? Or is this a 'brute force' example where you just tweak the placement of each one to get the overall correct look?
Then I did the same with one clamp I had not zbrushed, did the above steps, but later I attached a smaller clamp on the inner most arch rim before you meet the metal. I then selected the bigger clamp and attched the smaller one to it, so the changes i made to one were duplicated around again, this time without my needing to place them by hand.
Actually, I think that the only thing I modelled fresh in this scene was the metal frame work around the glass, the rest is all old kitbash files that are instanced. I think I took 2 days on it, possibly 3 including the lows.
how do you find polycruncher for hard surfaces??
i made a base concrete jungle kinda building in max, roughed it up a whole lot in zbrush and polycruncher gave me a pretty lousy result....
Also can you share with us the render settings you use to bake out the maps,i know,you use lighttracer with a sky light,other than this,what render settings do you change, for instance like the AA filters,sample sizes,etc.
Do you use Mentel ray for any of the renderings or it is just the standard scanline renders all the time?
thanks Kevin
Vj
You still make your asset the way you normally would. A tool called "fracture" in UE3.5 will boolean chunk it up for you.
That's a shame the Eat3D thing fell through Kevin...I was looking forward to that. Maybe tutorials could still be sold directly thru Epic?
Anything destructable has to have a 3d shape if you want to have chunks like they did in Gears 2. Actually if you look closely at the pillars or other pieces that break apart in game before you shoot them you can see where they are actually cut up.
Funny how fast things can change. The talks have been resurrected! I can't really say much more than that though, but the DVD now has a good chance of being produced. Stay tuned!
Oh and amazing work Kevin, you are an inspiration to us all!
- Riki
Nice! Great post man, thanks for taking the time to do the write ups and tutorial!
Yeah, he saw you were browsing and decided not to post anything just to hurt your feelings, dude.
Kevin, this workflow insight is awesome. Saved some of your whole posts to my reference folder.
Looking forward to, Kevin!
Really great stuff, I would love to look forward for a not so detailed, more designed, minimalized piece of yours.
Nice job
I think that is one of the most accurate, and modest, things I've heard any artist say. I really can't ask anything that hasn't already been mentioned, but this thread has just become a .PDF for technique reference.
Thanks, a lot
I think it's mostly because many artists don't get their hands dirty enough with engine exports and tweaks, and just stop at the fancy render stage.
There are soooo many cool approaches available now, this thread being such a great proof of that!!
Thanks for all the shared information Ror!
Its easier for most artists to let the work talk I think.
Although I think the 'let the work talk' is a good point, someone who is articulate and has excellent communications skills PLUS examples of said work makes for a great mentor or teacher. And definitely someone to look up to when it comes to their craft. I'm pretty sure we're all in agreeance that your modesty hasn't made you 'some puffed up pontificating' person.
I realize that sounds like a real kiss-ass thing to say, but I definitely stand by it.
RDM: We get scene concepts and we identify ourselves often which elements would be the most useful and worthwhile to build to allow the LD's the tools they need to make the scene. Sometimes that means we have to build things that are not in the concept, or rather the things no one notices in the concept or cares about, like plain walls and floors, other times it just means we research similar artifacts from the real world and use those as a guide when deciding how to add more clarity to things.
Concepts are great, they get everyone on the same page, they unify and open up communication between departments, but they are not blueprints.... you always have to research the thing a bit.
I didnt really do variations, I just made a basic cylinder or block, subdivided and took it into zbrush and worked out the design there. I don't really do the variations, think about it , pick the best, finish it out and then call it done 'thing'.
I take my best guess, model it, call it done. When I've done it wrong, some one tells me, but until I get that tap on the shoulder I just crank away without really taking the time for variations because our schedules never really allow for that kind of ponderous measures thought.
The variations you refer to come through my refining the theme as I go forward between assets, each improves and refines and becomes more sure of what it is until I reach my zenith with some kind of show stopper piece near the end of production on each theme that leaves me proud of what the set evolved into and sad that I have to let it go and move on to the next unknown, awkwardly trying to work out what it is again.
The worm pillar and the arches were earlier on in the Locust theme's production, the forms are still a little too soft, not jaggy enough yet, I'm still vague on what I'm doing, trying to do and the way im trying to realize it. The Block pillar and the floors came at the halfway stage, by then I was more used to Zbrush, working with alpha stamps, getting down more on the sharper edges and doing a lot more confident clay tool work by hand to turn the softer and plainer alpha patterns into newer or rougher unique forms in each instance they were used.
thanks,
looking marvelous as always
Sampson: I think polycruncher does better with surfaces that are detailed roughly all over, like the Locust stuff was well dealt with because all the flat forms had some sort of soft to hard undulation with some sharp deep cracks and gouged concrete like parts.
Super subtle detail on top of a very flat surface tends to fare less well, like if you were doing those fine pock mark indents like you see on limestone or certain treated concrete / plaster walls. Also thin scratches on metal forms do poorly too.
What I've seen happening is that it deals poorly with the flat areas, it concentrates simply on saving the detail areas and stretches very few polys inbetween them which leads to an ugly washed ou stretchy look. Sometimes when its very fine detail you do only, that gets washed out too.
The trick seems to be to balance the forms, try to always have some sort of medium sized soft detail like subtle bending on metal as you get to the corners, the same way you place fall off slopes from a tight pinch on clothing for instance.
With concrete, a few well placed deeper crevices that lead into those long fine cracks works better, its like polycruncher needs the medium forms in order to understand how to link all the little details
But i have some questions for you,i have read the entire tread but i might have missed if someone asked the same questions.
First i have 3ds max 2009 with pro optimizer,I suppose its the same as Poly cruncher.
Do you use poly cruncher when you striped down the high poly mesh?
I cant work with it because it messes up the mesh very badly,flips the faces and no real control over different areas in the mesh.
places like corners and sides where i want a sharp design (hard surface) tends to get messy after using pro optimizer!It often leave ALOT of unnecessary faces there.
Not sure how you do it but i export a medium res model to 3ds max and run pro optimizer.
Then as i said before i get a lot of flipped faces but i need the geometry.After that i import my high res model and run a render to texture and render out all the maps i need.
The real issue i have is when importing a 10 million poly mesh into 3ds max,Best would be to render out the maps in Mudbox/Z-brush to avoid such highpoly mesh in Max but then i get issues with uv,s and shit.
Anyway i have to read everything one more time i think
If your object is more than 250k, then use polygroups to seperate the piece into 250k chunks and export each of them out as individual OBJ's and then once they are all exported to a folder, just run polycrunchers batch optimize function and point to that folder and it will optimize everything there as much as you desire.
Pro optimizer mostly isnt very good for this stuff, I have tried it, some others here have too, we use polycruncher.
When having problems with retaining the detail you want, you can optimize less severly for those sections, polycruncher will do incremental optimizations , like a 25%, 50% and 75 % optimized version of each OBJ. That way you can import all the ones optimized by 75% initially and swap out for 50% optimized versions where you need to.
Sometimes you are going to be driven by this to make 2 or 3 process files because of the polycount. Hope that helps.
If you note the subtle banding on the large open door edges, I did that with a clay brush and the square alpha 28 set to a very low 20 in strength. I did this to avoid exactly the problem TheWiruz was wondering about, having some sort of subtle undulation underneath whatever fine detail you put on top like scratches is a good way to help reinforce the importance of the whole mesh to polycruncher
For someone who is trying to get into doing more high poly assets for baking purposes do you recommend learning how to do HP work in max first and then learning zbrush or should you try and learn each hand in hand?
Is it better to learn hard surface modeling and then learning how to zbrush in even more detail later or.
being at epic your going to have concepts rushing in for you to make through the doors, do you have any advice for having a big case of modellers block and nothing really to work off.. (heh) are there any artists you look up to?
Just dont try to do everything at once, pick your battles, marshal your forces like Sauron for a while
Sampson: They don't really rush in like you probably imagine, theres a great deal of research gathering and inspiration building required. We get scene concepts that are quite loose and most of the time they don't quit fit the way a level develops due to gameplay changes often months or weeks after the concepts are done. I mean its always down to you as an artist to find your own hook and then clarify it.
But theres a whole world history of art and inspiration out there and perhaps its my awareness of that, that prevents me from ever really getting artists block. I guess I look up to everyone who ever built something cool in the real world or made a sexy looking movie or wrote an inspiring story, these are the things I tend to look up to.
I'm a fan of weird things, like Hawksmoor who was responsible for a lot of cool buildings in London and in general its the dead artists long before my time that invented or pioneered a lot of the building plans and art movements I routinely steal from. The people that worked all this out for us to rip from are were very clever and I like reading about their ideas, the way they subverted the goals of the people with the money and transplanted them with their own dreams without them knowing.
I just love the story behind many of the pieces of art out there, just understanding how much planning and thought went into real world constructions is a huge inspiration to me.
The rock stuff is cool too. Very natural and a nice blend of soft and sharp shapes.
I think I prefer the earlier stuff to some of this locust architecture - mainly because some of the locust pieces seem a bit "soft" to me in general.
Out of interest, for these more organic pieces do you tend to use ZBrush "Layers" at all, or do you just sculpt everything in one pass and build it up that way?
Cheers for all the information and, of course, the delicious eye candy.
I started off controling things more with layered passes and then found as I subtooled more it wouldnt let me use the layers so I just did it all in one. I'd just do it in stages though, moving up in res after blocking things at lower res.
I still feel the mirroring point is blatant in the door, makes me feel bad to look at it but i had to fit everything on one texture and I knew the shadowy corridor and monitors in that area would hide how obvious it looked ingame!
I have this problem though when i run both polycruncher or pro optimizer
What kind of texture space were you given for a piece like that? I'll admit I'm not too familiar with GoW2 so I don't know how big that is in game (I'd imagine pretty big).
And also if I may as a followup grid-style question. The locust arches on page 5 don't seem to really fit flush with anything. The mesh seems to be bubbly all over the place. how do you plan for lining pieces like that up with the straight walls or other pieces? As in do you know from the top there will have to be some overlap? Or that the player will never see the X face of an object so you don't bother with details.
Thanks again for sharing the candy and knowledge!
edit: Is that vault door concept available on the net somewhere? Seems to me I've seen it in the WYAWO thread from someone else a while back (an unzbrushed version by eraserhead I think).
Wiruz: Well, why would you need zbrush to model a lowpoly bumpy brick? It's easy to do again by hand and unwrap, or its easier still to think ahead and make a duplicate of your base bricks before subdividing and use those later with some modifications. Bricks aren't tough to model again and often you find more efficient ways to build something the 2nd and 3rd times.
Using optimizers to do it for you will always result in you needing to get your hands dirty and clean things up yourself. Especially when you are using so many unneeded polygons to match the impression of the hipoly.
We do better if we think of the lowpoly as soft putty, it needs only to be an impression of the hipoly and holds up when you walk around or rotate it. It's the processing of the HI to the LOW that hardens the lowpoly now, its not like before where we need to be exhaustively accurate to the boundaries of the design, think of it as something soft and pliable, it only needs to carry a single smoothing group well enough to catch all the normals off the HI during process.
Peris: Exactly, I'm not doing anything that complex here, the Necris stuff on UT3 is complex, it does take a lot of skill to make but the locust stuff in particular, blocks and cylinders that are equally subdivided is all you need to be capable of in 3d. After that , its good old reliable doodling skills in Zbrush3 where theres no rules and restrictions really
I think in general I took 1-2 days for each of the locust HIpoly meshs.
Kessler: haha, dont spend too long planning though, or your will be tired out halfway through the fun stuff
Saidin: Thanks, I put a lot into that door, there was no concept or guideline beyond ' we have a door we need to blow up, it'll be in the cinematic too' so I just grabbed one of my favourite doors from bank heist movies and made it more utilitarian. it was a single 1024 texture.
Well spotted on the Locust stuff, they are not gracefull modular pieces that fit flush like the manmade stuff, which was a concern to me initially. We worked around this by giving the designs a slight twist on the X axis so that when you put 2 together, they do fit semi flush with an organic feeling join that looked more naturalistc and sculpted artistically rather than like something we would make thats neat and ordered and clearly built by hand.
We also made sure all the designs were self sealing elements with no open edges or empty planes and less mirroring than normal so it would further communicate the notion that they'd been gradually worn away and made chiselled at haphazardly or stuff had grown on them.
Thanks Austin, how you doing these days?