ZBrushCentral

Hulk - done!

Finally got around to finishing this fella’s head, and adding on a body.

I did him to fill out my videogames portfolio, keeping within real-time specs. So no nice subsurface scattering or soft shadowing, I’m afraid - just a 3-4000 triangle model with colour, specular and normals maps and a simple Maya Hardware render.

All his normal maps were generated from a high-poly version modelled in ZBrush. I also relied enormously on The Greatest Plug-In Ever Made - a.k.a. ZAppLink! - for the texturing. :slight_smile:

hulkCU.jpg

Attachments

hulkwide.jpg

really nice low poly

can we see the wire ?

the body has lost a bit of definition compared to the head, but maybe it’s due to low rez texture.

Top notch

Don’t take this the wrong mate, this is spot on modeling. This just brought a smile to my face because it actually looks like a little baby Hulk! :laughing: Maybe because of the proportions and the fact he’s a little soft and rounded.

Probably just my sense of humour mate. Nice work though, I like it :wink:

:+1: :+1: :+1:

Nice ear.

It’s very much a pre-Mignola Hulk, isn’t it? Bit like a Byrne version, I’d say.

That poly count is not next gen, but the tag on the picture says PS3/X-Box360. If you are shooting for a next gen model you should be allowing yourself at least ten thousand polys; twenty would not be excessive.

Cheers,

R

HULK SMASH!!! :wink: Nice work. All of my game models are in my signature. I have to keep them at 4k or less as well. It certainly can be tough to get a lot of detail in the mesh at those numbers, but I believe having limitations like that is great for improving your skills. When you’re allowed to use higher polycounts it’s just that much easier.

Thanks for the feedback, gents. :slight_smile:

Wireframe: ![hulkwire.jpg|407x433](upload://sTp6eMZgaq4twMAbXhn2g3MHNJt.jpeg)

guillom: Yeah, he has a 2048x2048 texture just for his head and hair, and then another the same size for the rest of his body. So the detail in the body does look “soft” compared to the head. (Bigger texture sizes and my graphics card dies on its knees…)

Angelstein. Heh heh… well, my overriding impression of The Hulk comes from the funny books I read as a kid. Which was, I’m embarrassed to admit, back in the '70s. So I wanted a ‘beefy’, old-skool look, kind of like the actors you’d see in the old Tarzan films. I know there have been more ripped, sinewy, veiny Hulks since, but I was on a nostalgia trip for this one.

The Google Image search I did for reference turned up all sorts of weird variations on the character. Check this one out done by the original Hulk artist in the '60s…! :D:

[img]http://perso.orange.fr/encyclopedie.univers.marvel/Rubriques/Etat%20civil/The%20Hulk_3.jpg[/img]

Rory_L: With the greatest respect, I have to disagree with you: there is no “rule” for polygon count levels on next-gen characters. The polygon budget all depends how much else is going on in the scene. For example, I’m certain that the characters in Dead Rising are way, way below the 10,000 mark - some are maybe a third of that.

In my experience, normal map technology means that it’s not even necessary to use a massive amount of polygons anymore. You just need enough to give a relatively smooth silhouette to the figure. The fellas at work have created PS3 characters that look pretty much movie-quality at the size they appear onscreen in the game, well within a 5,000 triangle limit. Remember, the goal with game models is to be as polygon-efficient as possible. If you don’t need 10,000 polys to make a model look good in-game, why use them?

20,000 polygons per character - excessive, in my opinion ;) ! It'd be OK for one-on-one fighting games where there are only two characters in the scene, the camera gets close enough for the extra detail to make a difference and there's not much physics and AI processing and so on. But otherwise, I'd say that's too high to be practical in most types of game.

Also, given that game characters have no proper muscle system, higher poly counts create nasty problems as far as deformation goes. For example, in the crook of the elbow and behind the knees, where the polygons have to ‘crush’.

Kaiju san,

You got me! Yes, with all those provisos, a 5K poly count is fine. I was indeed thinking of models that occupy most of the screen. When Eye of Judgement comes out take a look, won`t you?

I was right about the Hulk style though, wasnt I? You could even do a grey Hulk render with that model. Id love to see that.

Cheers,

R

Yes, This is definately a Kirby-Hulk.
IMHO it was the best version of the Hulk, namely a massive lump of meat,
not particularly a bunch of oversized muscles.

good job 'Kid :wink:

Bimm

Rory_L is actually right. 20.K is a bit too much but ten thousand for a lead character aint too much. Allthough I find the model lovely, it’s definitly not next-gen. It’s Xbox or playstation 2 gen.
I do think you should spent some extra time to create seperate fingers. I don’t want to be rude but since this is for your game modelling real. You should be aming at next-gen models. Take a look at half-life 2 there’s more detail in the average model on screen than there is in yours. You could easily use double the amount of poly’s (only on the places where needed the face looks good allready but the body could use some extra detail.)
Take a look at the unreal engine 3 http://www.unrealtechnology.com/html/technology/ue30.shtml
As you see there is even dynamic realtime softshadows.
Anyway you’ve got plenty of space to spice up your model.

Joost

Er, Joost, I’ve worked on PS3 since last summer - I know first-hand what qualifies as next-gen! :smiley:

It seems people are hung up on the poly count. There’s actually more to it than that. Yes, PS2 can handle 4,000-triangle models. What it can’t handle is large texture sizes and normal maps. My Hulk model has both. You can literally put the camera an inch from his face and count the skin pores. I challenge you to name a PS2 or Xbox character model where you saw that happen!

hulkpores.jpg

I’m repeating myself now, but high poly counts in games are less of an issue since normal mapping came in. You don’t need to add extra geometry to create the impression of 3D detail anymore; normal maps can take care of that instead. So for PS3 I’d say that texture size is as important to creating a detailed model as poly count.

So all this talk of, “a game model isn’t to next-gen specs unless the poly count is over 10,000”… sorry gents, that’s just not true, and anyone working on PS3 or Xbox 360 will back me up on that. If somebody creates a normal-mapped character that looks great at 3,000 polygons… why worry? Nobody in the videogames industry is going to shout at you for using fewer polygons! :wink:

Well I won’t argue with you on the polycount but I do think that you have the resources available to get some extra detail in there. I know normalmapping does a great job in faking geometry. But you can’t make seperate fingers with it.
That said… why did you use tri’s instead of quads?

Joost

Joost, I didn’t model separate fingers because I knew from the outset that I wanted the character to have clenched fists.

The trouble is, if you model a game-quality hand with outstretched fingers and then close it up into a fist, normally the fist looks less than great. Look at this from Dead Rising! :smiley: :

DRfist.jpg

So I thought, “the hell with it, I’ll just model him with fists right from the start!” Other PS3 models I’ve done have the full complement of ten fingers, so I thought I could get away with it just this once. Or not. :wink:

Re: triangles vs. quads: I model in quads, then convert the mesh to triangles for binding. That’s because Maya doesn’t fix the triangulation on quads - the triangulation can actually flip the opposite way as the character animates, producing an odd jumping, rippling effect with the textures. Triangulate the model and that doesn’t happen. :slight_smile:

Great work kid. What a great thread. I have never worked in the gaming industry and have no plan too, but I have learned some interesting things reading this thread. :slight_smile:

People! He is limiting his poly count on purpose as an exercise. It is a good idea. It is good to have some kick ass low poly stuff in a portfolio to show you know what you are doing.
I think this looks really amazing. It totally goes to show that it is not poly count that makes great looking models. Doom 3 also really brought this point home when it was first released. Those models were so blocky when you looked at the silhouette but when you looked at them head-on they were amazing! These days it is really more about the maps you stick on there.
I agree about your assessment of Dead Rising. Those zombies are pretty low-poly. I wish they had put some better maps on them though. They get a little pixelated when you zoom in on some of their clothes sometimes.
Anyway, I think this is totally kick ass. It really harkens back to the old days of Marvel with a little of the new thrown in.
great job!

The rule of thumb I use, is that you only need as many polys as it takes to make an acceptable curve. When possible I leave flat areas as a single poly. The polycount of any model should be appropriate for it’s use and complexity of shape. Mostly unclothed models such as the Hulk require fewer polys to look good than ones wearing many accessories. As far as the situation of the head using the same size texture as the whole body uses, as long as the amount of pixel detail spent on the skin of the face is the same as that spent on the body, it shouldn’t stand out. Poly efficiency FTW. :p:+1:

OK that’s a really nasty looking hand. Since you’re an expert… Is PS3 really as powerfull as they suggest or is it just like PS2 lot’s of power but that power aint accessable…

Joost

Gr8 character… NiceSilhouettes… To me the character is fine enough for a nex gen game… but if it was a lead character I would hav wanted to push it to some 7 to 8k… but there are not many hard edges in the character… I guess it works out pretty well. Nex gen can basically be passed of as term … with the present stuff that is going around making games… u might find characters having poly limit around 8k to 10k and told next gen… I wont be surprised when in sometime a company has a tag line for their game called “True nex gen” only because they hav used some more poly’s used some new types of shaders (maybe a Super Normal offset map) or textures to get that photorealistic quality or cinematic quality more closer in games… As it was mentions in several articles too… bringing out the maximum outta a Ps3 or a 360 console is not many publishers will be able to achieve at the very moment… they involve high costs
(ppl dont want to take risk as already development cost hav increased a lot) and lots of learning of the new hardware. yeah we might be seeing the consoles at full power in sometime… but lets see… So basically not a hard and fast rule that 10k or 15k makes a next gen character… it depends on wht is required where… again this is wht I think… ppl may differ.

Thanks for the kind comments, folks.

Kaibosh: I think the low-res zombies in Dead Rising are forgivable under the circumstances - there are crowds of the damn things :D! What’s more disappointing to me is the way the makers seemed to have “Westernised” their approach. The visuals look like they owe more to Grand Theft Auto than to Capcom’s previous titles. Give me the beautiful Japanese-style character models of Resident Evil 4 and Devil May Cry 3 over GTA any day!

Trybal: It’s not uncommon to use more overall texture space for the head and eyes, just because that’s the area most likely to receive a closeup in cutscenes and so on. I’ve also found that hair with transparent alpha’d areas needs comparatively more texture space to look decent.

Joost: Re: the power of PS3 - a programmer is probably better qualified to answer that question. :slight_smile: However, the company I work for is developing a cross-format title, and I will say that the PS3 doesn’t get special treatment from the character art department: the models for PS3, XBox 360 and PC are identical.

The skin uniformity of your Hulk is fine. My game models use two 1024x1024 maps; one for the head, one for the body. What resolution do you paint/render your textures at? I do mine at 4096x4096.

Texture sizes for one of my character models: 2048 for the body, 2048 for the head (with the hands on the head map if they separate off the body cleanly). I prefer to work on the textures at their normal size. That way I know that the detail I put on them will make it onto the final model. But it’s all down to personal preference.