Halkun's 3D Arts

halkunhalkun Registered User regular
edited August 2011 in Artist's Corner
capturewch.png

Little model I'm whipping up for a game. She's sitting at around 3800 polys. (Many are still quads, so It's probably closer to the high 5000s when she's triangulated.) I've put in about 40 hours into it. I'm going to be UV mapping tomorrow and then rigging her up. I haven't chosen an engine yet as I'm still in the asset making phase.

Created with blander 2.5 - I may just make the game using blender's own game engine, but the game is *VERY* text heavy and there is next to no text management in Blender's engine.
I like Torque because it has an awesome scripting engine, but Torque export support was removed in Blender 2.5
Source would be better if it had scripting support. Recoding the engine just to tweak it is waaay overkill.
Crystalspace + CELStart? I donno.

dA03mgx.png
halkun on

Posts

  • KochikensKochikens Registered User regular
    Post with the wireframe and some closeups of the face. Do you have concept art for this character? Is this the main character? What is her role? What is she supposed to be giving to us, as far as feeling goes? Is she strong? Is she smart? Is she a generic anime girl for a dating sim?

  • halkunhalkun Registered User regular
    edited August 2011
    Because you asked...
    kaowork.png
    capture2fh.png
    capture3z.png
    capture4mx.png

    This is a project I started in 2007, but quit as I didn't have the skillz at the time. The is actually version 2 of the model. Version 1 has thankfully been lost to time. That one had over 68000 polys and I started with a Makehuman model. I made some pretty bad assumptions about video game models. (What do you mean they draw over the model with a skin? I have to learn how to use Photoshop too?")

    The wireframe was done using a Japanese 3d sketch program called "Hexagreat 3D". Because of it's sketchy nature, the loop cuts are an after thought. I'll tell you right now the wireframe is a mess. It's actually much better than it was.

    The game itself is a game to teach you Japanese. You are stuck on an island with her and she doesn't speak any English. You have to learn to survive. Lots of puzzles. I been throwing back and forth puzzle ideas with the Japanese staff at my old college.

    It might be nothing, but it was sure fun to model her, and I've gotten tons better. I deviated from the original design a little. The bobbed hair made her look top-heavy. also, she was originally wearing a tank top, jacket and long skirt. In my original model, when I went to weight-paint the skirt I realized that modeling animated fabric was a royal pain in the ass. Simple shirt and miniskirt for the win.

    The miniskirt is not merged with the rest of the body yet. (That's why you can see her body though the mesh) Ill be doing that tomorrow.

    halkun on
    dA03mgx.png
  • KochikensKochikens Registered User regular
    Whooo
    that is a really
    really messy model
    Have you ever taken a look at any professional done character models? They're built so that where the polys are they will deform properly for animation. All of those bizarre tris everywhere will make it so that when you go to animate her, shit will go crazy and ugly.

    I mean. Her eye isn't even separate geo. How are you going to make her blink? Have you tried throwing a basic rig into her yet? That will make the problems very apparent. You can spend forever weight painting her to try and make shit look right, or you can re-topo her. I advise re-topoing her because damn. You should stick to quads. Draw lines down and around her, how you'd like her to deform, where you need room to bend. Then when you re-topo, take that into account. I know this is your first model, but if you try to go through an entire game using this model, you will be very cross with yourself at the end of it

  • halkunhalkun Registered User regular
    Last night I showed this off to others. They pretty much agree with you, and I do too. I tried a few test deformations and it was not pretty. Tonight when I merge the skirt I'll be "laddering" up the body and down the arms. Trust me, it looked a lot worse, and I know where I can fix it. The arm and legs have been massively cleaned up. The chest is such a mess because I kept subdividing triangles to get the shape right. Now, I'm in "clean as you go" mode.

    As for the eyes, there is a strange problem when making anime eyes in 3d. Namely, anime eyeballs don't fit into and anime skull. I found this out the very hard way with v1 of my model. If you make the eyes big enough so they are not bulging out of the eye sockets, the edges are poking out at the ears. A technique I picked up is you make the eye flat or slightly curved and you animate the UV under it. It's a little odd, but my tests in the past have worked great.

    I'll update tonight.

    dA03mgx.png
  • GurtPerkGurtPerk Registered User regular
    I wish I had more time to critique, I might edit this comment later, but for now have a gander at this beauty: http://www.hippydrome.com/

  • MagicToasterMagicToaster JapanRegistered User regular
    What ever happened to those Vocaloid boxes you were re-branding?

  • KochikensKochikens Registered User regular
    halkun wrote:
    As for the eyes, there is a strange problem when making anime eyes in 3d. Namely, anime eyeballs don't fit into and anime skull. I found this out the very hard way with v1 of my model. If you make the eyes big enough so they are not bulging out of the eye sockets, the edges are poking out at the ears. A technique I picked up is you make the eye flat or slightly curved and you animate the UV under it. It's a little odd, but my tests in the past have worked great.

    Make a separate plane with the eye under it, at least. Unless you're going to animate eyelids into that UV you are giving yourself a LOT of unnecessary grief.

  • halkunhalkun Registered User regular
    edited August 2011
    What ever happened to those Vocaloid boxes you were re-branding?

    It was a matter of the company being in europe and me being unemployed at the time. I now have a "real job" and doing this as a hobby

    halkun on
    dA03mgx.png
  • Angel_of_BaconAngel_of_Bacon Moderator mod
    What I said about this project 4 years ago to the month still applies just as much now as it did then.

    That being:
    I think the bigger concern over whether it is "anime" or not is secondary to developing an art direction that will actually allow you to actually finish this project at some point. I mean, you're making this game from scratch, all by yourself? Coding, scripting, level design, art, animation, and QA? You're just not going to have the time to do settings and characters at the level of detail you're describing there in that screencap and get further than being able to make maybe a demo, unless your ultimate deadline is around 4-5 years from now.

    So here are your options to actually get this done:
    1) Get more people to work on this project.
    2) Design for what you can realistically do.

    Assuming you don't spring for the first, this is what I'd recommend. Simplify. Look at DS games like Zelda: Phantom Hourglass or stuff like Katamari Damacy or Sims Wii- very few polys, and the assets wouldn't take a lot of time to make, but they still look good because they have a solid, coherent art direction.

    I point out Katamari in particular because it was designed expressly to be able to pump out a ton of assets in a short amount of time from a fairly small team, which is- if you ever want to finish this- what you should be planning to do, what you should be designing for.

    It seems to me that going for (somewhat) realistically proportioned humans with anime faces in 3d- and then, of course, having to make environment assets that match that style- is going to be more work than one person can realistically handle in a decent timeframe and still manage to look good. And if you don't need (somewhat) realistically proportioned humans with anime faces in 3d to look good, why screw yourself over? Just make it fun looking, it'll save you time and energy and probably look better and be more fun to make and to play.

    EDIT: Also? If this is the only concept work you've done so far, you should not be moving on to modeling yet. Before anything you should have some art that conveys the overall feel of the game- the mood, the tone, the style of the thing. Characters, in the environment, doing their thing.

    When you've solidified what you want the player to get from the game, visually, then you'll have something to refer to when designing assets- 'how do I make X object or Y character fit in this world I've made'. Otherwise, you- no matter how solid the idea seems in your head right now- are going to end up hacking away at this thing piecemeal, and end up with stuff that doesn't quite work with other stuff, and will more than likely lead to a game that ultimately is visually and stylistically jumbled, weakening the overall impact of the game. You need to develop the overall idea on paper first before committing to it in polys.

    EDIT2: On a side note, I think it's kind of weird that I also once worked on a Torque engine educational game set on an island. :P Are you planning on doing episodic/individual lesson releases of this, or just one gigantic game? I ask because episode releases are probably a better idea when it comes to actually getting something released and getting people interested in a timely fashion.
    ok..i have to say this. i'll be that guy..

    everyone here is acting like this guy is actually making a game.
    good intentions aside, judging on his drawing skill, and the posts i've seen there's nothing to make me believe this game can get funded, or actually made. No offense but if you can barely draw how are you going to create the whole game, write the code, make appealing cover art....

    i believe you have an interesting idea for a game, but come on, how are you "making this"
    it just seems like you have a few VERY rough sketches, and some brainstorm ideas...

    Well, that's kind of a downer response, but not without its merits.

    I'd say that yes, he can make this game, if he starts planning around what he can actually accomplish rather than what he'd like to accomplish if he had years of experience and years of free time and decent funding and a bunch of people working on it, which is what appears to be the case currently. It's just not realistic and smacks of poor time budget management.

    If people got bonus points for high ambitions that fall short of the mark because of poor planning, we'd all be praising John Romero and Derek Smart and that 3d Realms and that Too Human guy.

    Honestly, if you simplify all the crap he's said down to its most basic, fundamental goals, it's not unthinkable in scope by any means:

    1) Teach Japanese.
    2) Be fun.

    That's it. That's all it has to accomplish to be a successful game (setting aside marketing whatever for the moment, I'm using the 'did it achieve what it set out to do' definition of 'successful'). It doesn't need epic stories or 8K characters or amazing wardrobe designs or cut-scenes or voice acting or whatever else kind of crap he might think of adding. It just needs to do those two simple things well. If he designed and planned around doing those two things well with what time and talent he actually has at his disposal and not what he'd like to have at his disposal, it would be doable.

    Otherwise, it'll probably end up, like no small number of independent movie/video game/book/comic/etc. projects and end up being a long and arduous road to nowhere; which would be a shame because it sounds like could be awesome.

    If you can get a hold of an actual video game producer, try to get them to look over your design doc and -probably the more important of the two- your complete schedule broken down, by month or week, including every task and asset needed, including the time needed for concept art, writing docs, QA testing, asset integration, rigging, etc, etc, etc. If you don't have a schedule like this, you should make one, immediately. You will have a much clearer idea of what you're actually up against when you've got a list of hundreds of required assets and thousands of tasks staring you in the face.

    If you show something like that to a producer, they'll be able to help you to figure out how to actually accomplish this stuff: you've got 2 characters: can you reuse the same rig? You need a bunch of trees- can you get away with modifying one tree model into the others rather than modeling from scratch? This one mission has a lot of fluff and not a lot of meat- can we put that content into one of the other missions? Having a realistic tiled texture go on forever looks bad- if it's more stylized, it'll look better and you don't have to do a ton of new textures, etc. etc. Find out what is necessary and what isn't- and under such tight constraints here, you've got to be prepared to be brutal in axing stuff that you thought up and just adore, but it doesn't work for whatever reason.

    tl;dr: Proper Planning Prevents Piss Poor Production

  • TheKoolEagleTheKoolEagle Registered User regular
    edited August 2011
    I thought I'd chime in here, my understanding is that you imported a front view and side view picture into Hexagreat 3D to generate the wireframe? I can only imagine that always ending up as a nightmare for polys, not only for rigging but also for texturing, the UV map must make no sense at all.

    I am by no means a great artist but I would recommend maybe grabbing the free versions of either 3ds max or maya, and picking up a character creator book to give you some good habits when creating a character.
    http://www.amazon.com/Model-Rig-Animate-3ds-max/dp/0321321782/ref=sr_1_3?ie=UTF8&qid=1313620351&sr=8-3 is a link to a book I had used several years ago, which gave some decent fundamentals since it took you step by step through the process of creating the character, rigging, and animating him. This might be a good idea for you to try first and then move onto your first original character.

    this is what happened to the first character I tried to rig and animate:
    zomgpic-1.jpg

    TheKoolEagle on
    uNMAGLm.png Mon-Fri 8:30 PM CST - 11:30 PM CST
  • MagicToasterMagicToaster JapanRegistered User regular
    Halkun, Ive seen you start lots of projects, but Ive never seen you finish one. I think you need to work hard and get this done so you can move on to other things. Seriously, Id be bored to death if I were working on something such a long time.

  • halkunhalkun Registered User regular
    Halkun, Ive seen you start lots of projects, but Ive never seen you finish one. I think you need to work hard and get this done so you can move on to other things. Seriously, Id be bored to death if I were working on something such a long time.

    Nonono, don't get me wrong. I picked it up and put it down 4 years ago. I recently picked up the new Blender and I've completely fallen in love it. After devouring a whole bunch of tutorials and books, I decided to start modeling again. When I was looking for a subject to model on my own, I looked into my archives and found this old thing and got hit by a spark in inspiration. My current goal is to model it, UV map it, and then make a walk cycle. That's the end of it, but it will also put me in to a position to import it into a game engine if I feel like it. Digging about I found my old notes on the game and thinking "what the hell".

    May be nothing, may be something. I'm just having a blast. It's looking sooo good and so much better than my last run years ago. Maybe I got a little more wise, I don't know. Man, If I could show you the original models, you would think it's a massive improvement.

    Anyway, I'm listening to your critiques, I decided to hold off on UV mapping and work on doing a re-topo of the model. Kochikens gave me the idea to look as some professional game models to get some ideas. When I get to the head I'll look into the eye. I'll pick apart the model for Miku Miku Dance and see how that's rigged up and steal study the rigging (especially for the eyes) to see what can be useful.

    captureswb.png

    Miniskirt has been merged. The feet suck, but they are just shaped so weird, I can't help but to get triangles in there. Most of it will be hidden by shoe anyway. I put a lot of loops in the knee joints and that helped the deformations massively, the ankles can use a few loops though and I can probably consolidate a few triangles in the process . I might give that a go. Also I might be subdeviding the skirt and top too much, but she does need to bend at the waist and rotate her back, so maybe not. I haven't gotten the arm yet bit that will be fixed much like the legs.

    Better?

    dA03mgx.png
  • KochikensKochikens Registered User regular
    halkun wrote:
    The feet suck, but they are just shaped so weird, I can't help but to get triangles in there.

    Better?

    no

    that shoulder, that chest, they are Not good. Neither is the foot. Don't make excuses. Fix them! You still aren't thinking with deformations. Throwing in a crapton of edgeloops does not mean you're going to have a good time animating.

  • McDMcD Registered User regular
    As far as 3D work goes, I tend to focus more on animation than modelling, but I'll offer some advice where I can. I don't think I'm putting forward a new opinion when I say that the topology here still needs a lot of work... The amount of triangles is probably the biggest issue. When I first started learning Maya, one of the first things about modelling we got told is NEVER have triangles on a character model you need to rig and animate.

    That page on articulation GurtPerk posted looks like a really great resource, I'd take a close look at that, especially for the shoulder area. I can see that some of the tris have come about because you're trying to consolidate edge loops, but that's not the best way to do that. I'll post up some sketches/draw-overs as soon as I can to show that you can pull a couple of loops together and still wind up with quads. You can see some examples of this most obviously on the hands shown on that same hippydrome page.

    I'd suggest that you start over for the upper body. This might seem extreme, but I'm not sure whether trying to tidy up what you've already got is the most time efficient way to fix problem areas like the shoulder and the chest. If you're going to be keeping it low poly, then I think a fresh start would be the fastest way and you'd probably learn more solid modelling techniques by doing that then you would trying to rescue what you've got. The legs and feet aren't as much of an issue, you could fix those up relatively easily from the looks of things here. I'd need to post a wee draw-over to show you how I'd go about it, which I'll try to do as soon as!

    Keep going though! Modelling, especially organic modelling, can be really tricky and it takes time to learn. The important thing is not to get too discouraged at the start of a project like this, you'll learn little tricks and techniques that you can apply to future projects, so it's all valuable experience.

  • halkunhalkun Registered User regular
    edited August 2011
    Kochikens wrote:
    halkun wrote:
    The feet suck, but they are just shaped so weird, I can't help but to get triangles in there.

    Better?

    no

    that shoulder, that chest, they are Not good. Neither is the foot. Don't make excuses. Fix them! You still aren't thinking with deformations. Throwing in a crapton of edgeloops does not mean you're going to have a good time animating.

    I forgot to mention I was only done from ankle to waist. I didn't do the chest, shoulders, arms or head yet. I ran out of time and had to go to bed. Also I found that I can re-loop the model pretty easy. The upper body isn't a total wash, but I can work with what I got.

    Tonight I'm going to start with some good loops and re-topo in these areas. The shoulder loop is icky, but workable, same with the ankle. Those will get fixed.
    capturerkc.png

    halkun on
    dA03mgx.png
  • halkunhalkun Registered User regular
    edited August 2011
    Update...
    capturebqa.png

    Fixed ankle in the feet, and made good progress in re-doing the chest. The shoulder needs to be fixed, possibly re-extruded from the body. Head is still untouched. I can collapse the polygons if I'm making too many. They are fun to fidget with so I can get shapes right.

    May re-do the breast.

    Is it progress? I feel like I'm learning a lot!

    halkun on
    dA03mgx.png
  • halkunhalkun Registered User regular
    Torso done, I'm re-looping the hands and fixing the arms now. Moving on to the head last.

    capturepkx.png

    dA03mgx.png
  • KochikensKochikens Registered User regular
    You've 'cleaned it up', but there is still obviously no thought given to where things deform other than by adding edgeloops. Adding just a crapton of edgeloops everywhere isn't how you make a good model. This thing is ridiculously high poly. Go look at a real in-game-model.

  • halkunhalkun Registered User regular
    I've been looking at a reference model and *everything*.
    RT-2.jpg
    RT-4.jpg

    After all my loops are done I can "pull" one loop into another and drop the count and then "pull" the other loops to take up the gap where old one used to be. It's waging in a 3000 faces right now. I've budgeted for 6 so after the mirror apply, I should be hitting that. When looking at blender tutorials they pull edge loops at the joints. It's true that the sims models have really only one or two loops at the joints, but that's also sims 2, an older game, and they look like straight tubes. for arms in kind of a "T" pose. I'm more going for shape and I guess can pull the loops together when I'm done. Is that a good plan?

    dA03mgx.png
  • sainsain Registered User regular
    The neck seems very thin, and the head very small.

  • halkunhalkun Registered User regular
    I think the small head is because there's no hair...

    Here's pretty much my final draft...

    capturewnl.png

    I almost completely re-did the hands, and cleaned up the face. I tossed out a whole bunch of loops on the body. In the end, including the hair and shoes, the model weighs in at ~8300 triangles. About the same as Alex Vance from HL2. I am very happy with this, and worlds better than what I started with. It's not perfect, but as it's only the 4th human model I've ever made and even then after a 4 year hiatus, I'm ecstatic.

    I'll do the final object merge tomorrow. The eyes will be held off on until I can find a good rigging solution for them. Then comes the UV seaming and the start of the UV maps.

    dA03mgx.png
  • KochikensKochikens Registered User regular
    Your body still looks nothing like that example you posted. You can delude yourself into thinking that it does, if you like. The only way to learn why what you have done is So Bad is by trying to use it. So, good luck.

Sign In or Register to comment.