Zbrush for REALLY low-poly work
vertex
I'm aiming to create models with relatively low resolution in Zbrush, particularly something along the lines of this:
https://www.youtube.com/watch?v=PU1fHQByrz8
Before I get into the process of doing so for many more models, I was wondering if anyone more experienced with Zbrush wouldn't mind offering some advice.
I'm really trying to avoid the box-modeling route because I'm really fond of the spontaneity one gets from zbrush's sculpting flow. Though, unfortunately the type of model I'm going for doesn't seem to lend itself well to that approach. Can I get the best of both worlds? I'm thinking I can use some tool in zbrush to distribute the edge-loops I want after I establish the basic forms, but I'm unclear what tools/methods would be useful in this approach. A full-on retopo seems like it'd be overkill for something this low-poly, and really, I love the fidelity of the sharp points on the head, mainly ears, nose, and hair and how that fits into the rest of the geometry. The back portion of the hair below the hat on the back of the head seems really awkward to model, high-poly or low, thus leading to my confusion as to whether there's a clear answer to this.
Any suggestions on what the best way to approach this type of model in Zbrush would be?
https://www.youtube.com/watch?v=PU1fHQByrz8Before I get into the process of doing so for many more models, I was wondering if anyone more experienced with Zbrush wouldn't mind offering some advice.
I'm really trying to avoid the box-modeling route because I'm really fond of the spontaneity one gets from zbrush's sculpting flow. Though, unfortunately the type of model I'm going for doesn't seem to lend itself well to that approach. Can I get the best of both worlds? I'm thinking I can use some tool in zbrush to distribute the edge-loops I want after I establish the basic forms, but I'm unclear what tools/methods would be useful in this approach. A full-on retopo seems like it'd be overkill for something this low-poly, and really, I love the fidelity of the sharp points on the head, mainly ears, nose, and hair and how that fits into the rest of the geometry. The back portion of the hair below the hat on the back of the head seems really awkward to model, high-poly or low, thus leading to my confusion as to whether there's a clear answer to this.
Any suggestions on what the best way to approach this type of model in Zbrush would be?
Replies
Regarding retopology for geometry of this sort, it's been a while since I've used Zbrush seriously, but is it capable of decent manual retopology yet, or do I still have to use something like Topogun for that? (Topogun looks really cool for most stuff, but I'd rather stay in Zbrush if possible (and user-friendly-enough for retopo work) and avoid Maya/etc. where I can, but I do use Modo for animation rigging in my workflow, and I can use 3DCoat for my texturing workflow -- if that helps offer some decent workflows for this.) Trying to limit the number of apps I switch between too.
I'm mainly looking to understand the best tool(s) to use to approach a head like this with lots of sharp points (ears/hair) and tight edges like this one. Got lots of tools to approach cylindrical shapes, but very few to handle tight/sharp polygons directly without jumping into Maya/etc. for box-modeling.
Zsphere topo in ZB works great, and is quick. There is also the topology brush but it can be a pain. If you use Modo then the topo pen is a great option.
for the pointy extrudes you can use the zmodeler brush.
But as I mentioned above, creating the basemesh using Zspheres at density 1 is very fast.
Absolutely, I'd love to hear all work relevant to this type of work. I'm looking for ways that fit how I'd like to work more, especially if they're both practical and possible. Though, if you have something else in mind other than what I'm thinking, feel free to share it! I'll definitely take your advice to heart. I'm aiming for it to be done with very little back-and-forth between programs, or integrating programs into that workflow that are used for only one purpose in the entire pipeline, while also being fun to use, since it's really great for flow when you're using a program you love to create in (in my case, I love to create in Zbrush wherever possible.) I, like most artists, tend to want to maximize flow to keep my motivation going as much as possible! I know it probably doesn't make sense, but that's just me.
@musashidan
I guess I don't have a real reason to avoid box-modeling, I just haven't found a program I enjoy doing it in yet as much as I enjoy sculpting in zbrush. I've found form-Z is loads of fun to use for that type of stuff, but I don't yet know it well enough to start a model in it from start to finish, so sticking to what I know for now is probably best, assuming the techniques I use are up to the task of at least comparable edge/vert control and time-sink. It just helps that I also enjoy using the technique too.
Also, thank you for the Zspheres for retopo hint. I had no clue they could be used that way. I'm having trouble hiding my zsphere though while I'm trying to use it for retopo. Using the dog as my test mesh, but not really figuring out how to do it yet. This seems like what I was looking for though. I just need to get it working first.
I also read somewhere that using the "Topology" brush and the "ZModeler" brush together does a good job, and to get those points on the hair extrusions, that does seem like a good plan. Do you think this might be a viable alternative to the Zspheres, or is there a better way in something like Modo? I'm really a fan of Modo for animation, though I don't use much else in the program. If there's a better workflow there instead for retopo, I'm all ears! This is the first time I've ever heard the topo pen mentioned, so I'll have to try it out! Thank you!
Ironically by hoping to use Zbrush for this sort of stuff, you *will* definitely end up with a workflow requiring a ton of back and forth between programs. I understand that this is not the answer that you want to hear, but that's the reality of creating models like the low resolution Link. Why making it more complicated that it needs to be ?
All that said, what you really need to do now is to go ahead and get started with it using any tool that you currently have access to. That way you will quickly develop a much clearer understanding of the issues at play here and will be able to make an informed and practical choice as to what to do going forward. At this stage speed is not really important - it's more about developing a good understanding of the workflow requirements. In a sense attempting something like that in Zbrush could be a good place to start ... because it will show you how badly you're going to need a proper way to merge vertices, align edges, snap to elements, and so on
I'm an advocate of learning as many tools/methods/workflows(and often applications) as I need to keep my workflow fresh and efficient. So it gives you a great headstart when starting a project as you know upfront what tools you will use.
Also, don't be afraid of multiple tools from different packages in your workflow. The way things are these days knowing about 5 packages very well is commonplace.
As for the ZB zsphere topo. Here's a vid tut from my YT channel that you might find useful:
I wrote a little macro a while back that will Freeze your subdivision surfaces to a "Level 0" polygon mesh when you're ready, preserving high level geometry and getting rid of your "subdivisions". This should get you to your low-poly model: http://community.thefoundry.co.uk/discussion/topic.aspx?f=119&t=126178
If you need to do further tuning of the low-poly mesh from there, you could get familiar with the Model > Deform > Move Tools, many of which involve operating on a soft selection.
It is hidden behind the Tube Primitive button on the basic tab. Hold down mouse and then select the "Solid Sketch" tool. When we mix that with autoretopo, mesh fusion...ect, Modo really starts to show some really cool workflows that work along side its sculpting feature set.
Despite how I feel about the lifeless creative-approach in Max/Maya from an artist's standpoint, I've accepted that box-modeling really is the best approach for what I initially wanted to achieve. It's faster all around for super-low-poly forms, and I do have Modo if I end up needing that approach.
That being said, I've got concerns as to what the sculpted detail will look like performance-wise on a model with such a tiny polycount anyway and as to whether it's even worth using a normal/displacement-map for it. Any thoughts on this? Looking at the forms I'd like to create, I'm thinking that these models would look better with a slightly higher polycount anyways due to the types of details I'm looking to represent, as my characters would have lots of hair protrusions and tendril types of attachments with larger heads (with less emphasis on the body, sort of like Link there.) In that case, I probably could use the sculpting approach to make model-creation faster and more fluid for this type of model. The biggest factor is performance, and ultimately I'm not sure whether going the route of displacement/normal vs. actual-polycount+diffuse in this case is truly better. If it helps, it'll be rendered using the Unity engine's tesselation features.
Any insight on this from you 3d veterans would be greatly appreciated.
The issue in your reasoning is that your targets are not clearly defined. For instance the link model shown in your original example (on which all the recommendations made in this thread are based on) is completely built to work with non-sculpted detailing.
But now if you say that you do want to convey sculpted details then that's a whole other story altogether - because there is definitely a limit to how low a model can be if it needs to support baking (for instance, a smoothly sculpted arm will not bake convincingly to a 6-sided lowpoly limb).
So my advice would be for you to spend the time clearly defining your visual target by simply looking at existing games. I get the feeling that the example you showed (N64/NDS spec) is actually not quite what you are actually aiming for here (my gut tells me that you may be thinking more about something like Skylanders/Yooka Laylee). If that's the case, all the recommendations so far would become irrelevant ...
The end result I was looking for (in my mind) was a moderately-higher polycount than the LoZ: A Link Between Worlds models on the 3DS. Not quite Skylanders, not quite N64/NDS, but somewhere in-between. I admit that I was only looking at a (very) limited number of screenshots for inspiration though (particularly the first and the last shots below were my main references), so it skewed the way I was thinking about the models after I looked at the polycount of the Link model.
I had assumed it used a displacement map for some of the character heads, such as Zelda in the first image, and Gannon-dork in the 4th:
After studying these a lot more in-depth, I now see these don't use any sculpted detail at all, but they sure did convince me otherwise -- even after looking very hard at them originally!
Despite looking at the construction of the Link model geometry, I was totally convinced the other model heads used displacement. The faces (particularly the noses and the jaw-lines, and especially Link's hair and chin), seemed to be lit using a combination of painted detail and 'true' 3d lighting (especially in the close-ups), but much of the more subtle 'light' everywhere else in my mind was clearly hand-painted on (in my mind at least). However, after looking more closely, they used the normals to do the actual 'painting' of the darkest shadows instead.
Even now, looking at the close-up of Zelda with that knowledge, it's still hard to visually tell if the lighting on the jawline/nose is painted or lit. It looks that convincing.
Looking at Link, especially in the bottom-most image above, it appears as if maybe the normals on his jaw/chin were edited, making it appear darker from the angle the light is actually being cast from. So I figured maybe that's the secret to the lighting on the jaw in the close-up shots of Zelda and Gannon-dork. So.. I went digging for more clarity....
Above is the diffuse texture from the Link in those screenshots showing the paint. I've also got the Link model below imported straight from the game into Modo:
Looking at the lighting both in-game and on the model itself, it clearly seems a bit wonky -- like it's lit directly from overhead. After bringing it into Modo, I've found it definitely has edits to its normals.
My suspicions that the normals were edited stemmed from the lighting direction being perfectly straight from the top, pooling light in all the right places, ignoring any placement/rotation of the scene light source. I've been trying to understand the lighting method used in this game for some time now, but up until now, I did not know the secret was in the normals until -- thanks to you guys! -- I decided to study it up more-closely!
If you guys did nothing else for me, you've totally given me a new technique to work with besides pure sculpting! This will totally help when I need super-low-res geometry!
Thank you to everyone who took the time to help me out here!! You rock!