I know, Zbrush (and sculpting in general) is taking over the game dev world.
But for a low budget indie title, I want to keep my toolset simple, using Maya LT for both environment modelling and character modelling. I had some success with it before, though only with very low-poly characters.
Because of the prevalence of sculpting, I can't seem to find a good character modelling tutorial that doesn't require either Zbrush or 3D Coat. While I think they are very fine tools, I would like to get away not using them, if possible.
What I am aiming for: 80% good. I am ok if using Maya alone can only go so far, I just want to see a tutorial that try to take it as far as possible.
(btw, Japanese dev often do character modelling without Zbrush and 3D coat, but they are quite secretive with their approaches using Maya/3ds Max. If you have any resource on that, I much appreciate!)
(fyi, I have tried searching 3dmotive to no avail, but I do have a subscription there.)
Replies
Here is a very old tutorial that goes through old school poly modeling a character:
https://www.3dtotal.com/ffa/tutorials/max/joanofarc/joanmenu.php
@kwyjiboI checked out your suggestion as well, it's decent and I will keep it as a reference.
I am not sure planar approach is my thing though. It's a valid approach, but I am happier redirecting edge flows and block modelling than figuring out where my next quad go.
-----
Thx I will check back when I am not on the phone. (is there a polycount forum app?)
To answer your question: not very low, I am thinking in the 2000-4000 range, as I intend to use her mainly in a strategic camera view. I may also make a higher poly (20k) version for closer view.
But I need to double check how much my reference game characters actually has.
kwyjibo said:
Basically, the tutorials I am looking for:
- Don't heavily rely on Zbrush or 3D Coat: it can make use of them, but I should be able to create 80% of the final model in Maya.
- Give me some tricks and tips specific to Maya: there are so many good reference out there, my problem is applying them in Maya.
(I am fine paying for video download, online subscription is fine too.)
- familiarise your self with the topology of all the basic primitives and experiment with their creation parameters.
- create polygon tool
- combining and separating meshes as well as detach component, and merge vertices tool
- booleans
- insert edge loop tool, slide edge loop tool
- extruding vertices, faces, edges
- splitting tools
- bridge and append
- bevel
- mirroring
- transform component
These tools along with basic move, rotate, scale transforms are all you will need 90% of the time for basic character modelling. Be sure to familiarise yourself with all the options for the above tools as well as having a good understanding of the transform tools (setting axis, symmetry, altering the pivot point etc).
Some other useful things:
- soft selection
- deformers (lattices, bend etc)
- sculpt geometry tool
- generating surfaces from curves (extrude, loft, revolve etc)
there are many super cheep and/or free ways to do sculpting for characters.
I understand your situation, but there is really nothing that beats sculpting for character work, even if your happy with 80%.
Please correct me if im wrong, but I feel this is more a case of you not really want to learn sculpting rather then practicality.
box modeling characters is a really outdated approach, and thats probably why you wont find any good modern tutorials on it. even a lot of mobile companies I know and have worked for use sculpting for their characters, its not just about the model, it also helps a ton with the textures, since you can bake down the details and lighting information to your textures, having 80% of the work already done for you.
on a separate note, which are these secret Japanese companies that dosnt use sculpting for their characters? Im genuinely curious.
Honestly it sounds like you would be much better off in Blender than in Maya. You can do the whole high to low pipeline (sculpt, retopo, UVW, bake) in it and you can paint handpainted textures in it just like 3D coat. Nobody would need to care about the budget because its free to use for commercial purposes.
I am fine learning sculpting, but I try hard to stay in Maya LT, for my own convenience. (Maya does have some sculpting tools, but for obvious reasons, most tutorials stick with Zbrush or 3D Coat when sculpting.)
Pricing is a factor, but not the ultimate factor. I am fine with 3D Coat's price actually.
I am just beginning to grasp the 3d asset pipeline (from Maya to PS or Substance), I fear throwing another tool into the arena is going to slow me down again.
(As for the Japanese companies, I think Tamsoft and Gust Co. Ltd etc. might not be using Zbrush/3D Coat in their workflow. Their games often have to work on 3DS or Phones, so creating a low-poly version from scratch might be a decent approach. Similarly in the west, Riot Games' LoL models are often created without Zbrush, right?)
(But obviously I am talking out of my ass by looking at their character models; IF I know their workflow I am surely going to try them )
as for the Japanese companies you might be right, I honestly have no idea, so Im not gonna say anything.
as for Riot Im pretty sure they sculpt all their characters and then retopologize it down to a lowpoly.
Example, this is the highpoly and this is the lowpoly for Xayah.
agreed that sculpting is so valuable for the textures. in fact, no point in doing them purely by hand in my opinion. baking shading down and painting on top of that is a superior method with a lot more consistent results.
besides, sculpting is fun, managing vertices on the other hand.... blergh.
if you want to go really low poly, you might want to have a look at old books by paul steed. they are written for ancient versions of 3ds max but the modeling functionality used is generic and can be applied anywhere and it's really step by step explanations of the how and why.
he was pretty good at maximizing silhouette and all in a time when 3000 polys was considered detailed. i used his tricks a lot in the old days (how to build deformable joints with very little geometry, cheat vertex normals shading with floating elements, etc).
nevertheless, these books are probably 10 - 15 years old now and i'm not sure if they can be found anywhere still (ebooks perhaps?), so you have been warned.
it would certainly make for more detailed retopologised models at lower polycounts that can hopefully deform better than what you can usually find in tutorials aimed at largely quad-based modelling. so perhaps not so far off from the 'secret sauce' of japanese mobile developers.
Regarding the OP : I think the approach you are taking is quite healthy, and from what I remember the texture painting tools in Maya are pretty decent if you are sticking to simple painted textures and low resolution (used it in production for exactly that about 10 years-ish ago and it was fine).
As for japanese devs : it doesn't matter one bit which app they use, but one can certainly learn a lot from their best models (very clean meshes, tight UVs, and so on). Have a look at models extracted from games like the FF and Kingdom Hearts series, they're great.
All I am saying is, I am nowhere as good as being able to stare at great models and get to 80% good. Thus my quest to find some tutorials
(PS: I know some japanese so I have been looking at this tutorial series from Autodesk Japan, unfortunately it's not as detailed as I want, they don't speak in the demo video neither.)
Here's an example. When I first started modeling lowpoly models I noticed that even rows of polygons where more pleasing than loosely placed edges for multiple reasons : it looked cleaner, was easier to skin weight, lended itself to very clean UV layouts, and so on. This kind of stuff almost never comes from tutorials but just from observing good models.
Put differently : there is absolutely no need for you to look for tutorials at this time, especially if you have not been looking closely at extracted models and UVs yet. Now once you do that you might have questions about a specific aspect of a model, but until then I don't think there is any point asking broad questions. It really is just about moving, extruding and merging geometry. if you already know how to do that you already have all the knowledge you need. That's a good thing, and precisely why I believe that your approach of sticking to Maya-only is quite healthy.
- What if I want someone to tell me? Veteran artists know how to observe, compare and improve a model, that's what I am trying to learn from tutorials. If I am on my own, it would take me a much longer time to start appreciating the crafts in great models.
- Shouldn't I take every help I can get? While I am prepared to take time to learn this craft on my own. I am bad at observing great models, as in I don't know the right questions to ask yet. Tutorials have been my catalyst for asking more specific questions, so why not?
- I don't have a game company background. I worked at Microsoft on enterprise software development. I haven't gone through any mentorship that artist might have in a company or institution. Which is why I am trying to catch up on it through tutorials.
All is all, I find the "shut up and go back to look at models" approach isn't really working out for me.
Or, after doing the research : out of all the models you've carefully observed, which strike you as the best / most pleasing / and most fitting to what you want to achieve ? It may be an artistic field, but these sort of attributes can be identified quite objectively.
i firmly believe that once you can point this out you'll be much closer to your goals than after watching tutorials.
1. Find some orthographic images of a character to model
2. import them as image planes in maya
3. Add primitives that approximate the main volumes of the character (torso, head, limbs weapons etc) and set an appropriate amount of starting divisions and position them to match the image planes. manipulate the edge loops to further match the reference.
4. Add more geometry where necessary (insert edge loop, split polygon tool, extrude) and manipulate the new geometry to further match the reference.
5. If you made parts separately (head, torso, limbs etc) and you want to combine them, use combine mesh and make use of bridge, append etc.
6. constantly look at your model from different angle to check the shape and make adjustments.
This is basically all there is too it. Understanding how to add/delete geometry, combine/separate meshes, and mastering of the basic move, scale, rotate transforms is basically all you need to know technically. The rest comes from trial and error and experience. Other little problems (like fixing up bad geometry etc) can be researched later as they come up.
It's very difficult to cover every little bit of knowledge you might need in a nice linear tutorial. At some point you just need to make stuff and research problems as they come up.
- The model: Elf Girl on sketchfab (she fits my target aesthetics and I can download the original model legally, plus it contains both the posed model and the T-pose model.)
- The game: Senran Kagura Estival Versus from Tamsoft (Ignore the sexualization, their animation is quite well done, and their clothing are done in modern-day aesthetics. It also has an in-game viewer for characters. Obviously I still want to extract the models for a proper look in Maya, but I haven't figured out how to use those tools yet.)
Some alternatives:
- Hyperdimension Neptunia U also from Tamsoft (It's not exactly the aesthetics I am going for, but their animation and models are equally well done, plus I can find extracted models to inspect.)
- Megadimension Neptunia VII from the same series (ditto.)
There are other Japanese games I like: say Atelier Sophie from Gust, but I can't find any properly extracted models, only ripped ones, so I am less inclined to use them as a reference.
Some backups:
- The Sims 4 (I am always amazed the amount of animation their characters can perform. There are base models I can inspect, and it would be nice to figure out how they approach custom clothing.).
- MMO like Archeblade, MOBA like LoL. (The cool thing is they have many models available to inspect, the sad thing is they are not quite the body type or aesthetics I am aiming for.)
But I bet my end result is nowhere as good as yours. Problem is, there are known issues and there are unknown issues.
- Known issues: I know my character isn't great, say, her arms are too thin. Whether I can fix it is down to my ability.
- Unknown issues: I don't know my character's topology is broken until I show it to someone else. Whether I can spot it is down to my experience.
Tutorials help because I will see how authors critique and explain their own work, I might not learn them straight away but at least I discover my inability and inexperience faster...
You saw these right?
No. It lies down to you measuring the diameter of a good reference model, and thickening yours appropriately. You can do that. Forget the myth of the artist feeling/improvising things. Observation lies in accuracy.
"I don't know my character's topology is broken until I show it to someone else. Whether I can spot it is down to my experience"
Again, no. You will know by comparing your model directly to references coming from professional sources, and rebuilding it that way. You may think that it is "dumb repetition", but trust me, this is about a thousand times better than recreating what a random Mr Youtuber does on a tutorial channel. And anyways, the only thing one could ever say about bad topology boils down to "Well, it doesn't shade smoothly, doesn't deform properly, and leads itself to uneven UVs - which themselves won't play well with lowres textures." There's nothing else to it, and I just saved you hours of scouting for tutorials
Be proactive ! If you "don't know that your topology is broken until you show you to someone else", that simply means that you have not taken the time to do the homework of locating a solid, professional model, loading the mesh and textures into your 3d scene next to your model, and actively scrutinized both.
There are a lot of people out there extracting and posing models from games, the material is definitely out there. And for every Workshop content contest that pops up the studio making the game always provides reference models. Granted these are usually not of the cute anime kind but that's just another example of reference material being easily accessible.
Lastly, here's an extract from an interview of Emiko Iwasaki, the director of Battle Fantasia by Arc System Works - the company that ended up building GGXrd and DBFighterZ.
- - - - -
It’s also the first game from Arc System Works to use 3D characters in a 2D battle system, a style many fighting games even Guilty Gear Xrd uses today. That was a big shift from the 2D games you worked on, can you tell us about this too?
In those days, it was the turning point when 2D fighting games were trying to shift to 3D, but no one was able to make that marriage perfect. It was said that incorporating 3D in a 2D versus fighting game was impossible.
We were a team with zero 3D experience, it was difficult to assign to me men who have experience and who entered the company earlier and are older than me. There were gender role issues deeply rooted in Japan. It is considered a shame for a man to work under woman because many people think "men first, ladies second".
I only had 2D art experience, so I had to buy a 3D textbook to study and to find a way to make a 2.5D basic game format work with a main programmer and a newly-graduate junior artist I found. We treated 3D as just an art style to give expression to a 2D fighting game, and we combined it with 2D techniques from pixel art ages. I have experience as a pixel artist and I’m free from common 3D wisdom.
I discovered that the way we did this was very unorthodox, at least for most 3D graphic experts. Most experts focused too much on consistency and realism.
- - - - -What that tells you is that the issues you are facing are probably not related to 3d modeling at all (how to build 3d models), but more related to knowledge of the human figure and its artistic interpretation. The best thing you can do at this point is take life drawing classes.
Good luck !
It would be like giving someone a blueprint to a spaceship and go "its all there, just build that thing".
I mean, your not technically wrong. but I dont think its especially good advice.
@bitinn - He is right though, Listen to Pior.
OP, Focus on understanding the three dimensional form of what you're making. If the shape isn't clear from your concept art, search for more reference or observe real life. Look at your hand as you model a hand etc. Look at your model from different angles and check if it is the correct shape. Focus on correct form and correct (or good enough) topology should follow. I mean it should be clear topological what to do if you truly grasp the shape of the thing you're trying to make. Then if you get stuck on a particular shape, you can google or ask here how to make it. "how to model character?" is too broad, "how would you start modeling this sword hilt?" is more specific and people can give you advice.