technical highpoly workflow tutorial and scripts..
polycounter lvl 18
this might not be everyones cup of tea.. but i am a character guy through and through and have been doing highpoly models for 9 years now.. and i HATE doing technical stuff.. so i sat down and developed some scripts and i had lots of help from mojo getting this stuff to really work.. and its not finished yet.. but its at a point i feel i can release it and hopefull people will only make it faster and more intuitive with scripts and maybe new ways of getting the same result.. i made the post over at cg talk casue i was looking for scripting help to get the preocess more streamlined .. but i wanted to share it with everyone.. if you have a great way of modeling technical stuff then by all means share.. i am personally a quad nazi and love super clean wires.. the workflow i was going for was to just basically make a lowpoly representation of what you want the highpoly to look like run a script and BOOM!! highpoly love! with nice smoothable corners and edges.. so here is a link to the post and a big thanks goes out to poop for the woping 4k of webspace he generiously gave me for the scripts... let me know what you guys think...
http://www.cgtalk.com/showthread.php?t=258695
http://www.cgtalk.com/showthread.php?t=258695
Replies
edit: post removed.
its just simple math really why you wouldnt want to do chamfers for every situation.. you will only be able to have an even number of intersections.. so try and make 3 edges intersecting and champfer them.. . its an absolutle mess.. extrude dosent matter.. 3 4 5 6 7 8 9 10 a million edges dosent matter.. any number of edges can be intersected have kwick corner applied to them and it always produces quads.. no matter what.. no exceptions.. if the whole process could be scripted to one operation i cant see how chamfering every edge and going and finding every instance an odd number of edges come together and find some kind of voodo magic to hide the tris or have a wireframe that looks like a busted window... and that would take several clicks..
here is a small example.. made a goofy shape and made it technical and organic and did it all in those 3 or 4 steps.. the hole thing took less than 15 mins.. if i was chamfering all those little notches i would get lost ... but thats me..
I tend to go the same way most of the time, ala modelling a tech object as if it was a 'mediumpoly' one and add the chanfers afterwards. And then EdgeSlide from CSpolytools is a great help for detail control once you are at this step.
One kickass thing to your already nice scriptset would be the ability to give the chanfer different values along it's path. I think there was plugin for Rhino allowing that but cant remember the name and don't know if it's been done already in Max.
Anyways! Can't wait to try these scripts, thanks for sharing
this really is very cool-- this is the type of thing i'd like to see integrated into an application. awesome work man.
From what i'm guessing, chamfers are a max related 'problem', if you're using max, and max somehow forces you to pay special attention to those chamfers (because of meshsmooth maybe), then i'm sorry... otherwise i'd forget all that stuff.
Doesn't seems to be an issue outside max. It seems to be dated and the most troublesome and time consuming approach to hipoly modeling someone could ever come up with, and probably the reason you don't see many models at all following that insane approach ...outside PC anyway.
There's 2 standard hipoly modeling methods you need to be aware of, hard surfaces, and SubDs. In both techniques, quads are nice, quads let you make edge loops... other than that, Ngons are your best friend, no reason to stick to quads if you dont have to. If you're using Lightwave, it will support Ngons in SubD with the next patch, that should allow you to model correctly with it.
-Hard surfaces are faster and a no-brainer for machinery, the downside is that rounded stuff isn't that great to do.
Should be easy for anyone that can do lowpoly.
You usually use that technique for things like:
http://www.strangefate.com/portfolio/server1.jpg
http://www.strangefate.com/portfolio/misc9.jpg
http://www.strangefate.com/portfolio/misc4.jpg
http://www.strangefate.com/portfolio/misc2.jpg
http://www.strangefate.com/portfolio/elewire.jpg - no crazy dirty wires, just rounded edges. Most of that stuff was done in LW still, any app, Ngon support or not should be able to do with that.
-SubDs are better for curved and organic shapes obviously and require a bit more thinking, but let you do complex/weird stuff you can't do properly with hard surfaces.
http://www.strangefate.com/portfolio/train2.jpg
http://www.strangefate.com/portfolio/trainwire.jpg - there's probably more Ngons than quads there.
http://www.strangefate.com/portfolio/beebugbot-rad.jpg
The box in your screenshot has all those unnecessary loops running around the box, if you wanted to add some detail anywhere around there afterwards, it would be quite a lot of work to clean that up somehow. If there's no reason to have edges somewhere, they shouldn't be there. You'll model faster and cleaner and you will be able to edit or add more details at any time later.
Basically, you need to try to restrict the 'area of effect' of geometry as much as possible, so you can do whatever else you want with the rest of the mesh, as close and complex as you want.
Also, limiting the influence will let you do complex stuff like cuts, booleans, seams etc on any rounded surface without breaking it's flow.
Sorry if i seem to bash a bit that max-chamfering approach, but i get headaches everytime i try to find the reason behind that method.
Just another way to divide edges.
The mesh you show would work the same with Max's subdivision.
The mesh you show would work the same with Max's subdivision.
[/ QUOTE ]
Yes exactly, Max, Maya, Modo, Mambo will derive the same exact mesh when treated the right way. There is no turn-around they all behave the sameway.
to StrangeFate:
thanks god for the drill tool ;P
afaik SubD should behave similar enough and coped with some clean hard surface beveling knowledge you should be laughing at anything.
SubD is incredibly more flexible than hard surfaces so i'd make sure to master that for sure. If you can do it with hard surfaces, you can do it with SubD, but not necessarily the other way around.
SubD is the way to go in the future imo.
Thx Sinister! chamfers, 'micro-chamfers'... got me all pretty confused as of what the mystic behind that stuff would be.
Now, i'm guessing max supports Ngons, and has a proper bevel tool, and as you said, max has SubD too, wasn't even sure about that.
Why is there still a need (or at least debate over it) to have meshes unnecessarily sliced even from the start on, making it a pain to detail them in any way.
Even in Arshlevon's CgTalk treath he starts with an already subdivided cube shape ( http://img115.imageshack.us/img115/5810/16lz.jpg ), rather than just beveling all the edges he wants rounded in 1 step.
With Ngons it should even work just fine in SubD mode, and i'm guessing that meshsmooth is pretty much the same as SubD, just that it phisically SubDs the mesh, so it should work fine too.
I also remember Poops wireframe image he posted in the hipoly thread in the 2D/3D Forum, an overly cluttered wireframe which would have been usually unnecessary. Afaik, he had to keep quads but you could see that he started the same way as arshlevon, with a 1 time subdivided mesh.
What's the point or reason behind that proceding and keeping stricktly quads over a clean and editable wireframe, that outside max only has advantages ?
to StrangeFate:
thanks god for the drill tool ;P
[/ QUOTE ]
indeed! The drill tool is the secret weapon for anything techy!
stf: I often let loops roam all around my model, simply because its faster to select a full ring than a limited range, or to clean up a full ring. And it rarely gets in the way of anything, because I dont add edges to sharpen corners until after I've modeled all the shapes of a mesh
[/ QUOTE ]
Absolutely, if you can let loops flow (ie, they dont break the curvature) and are not in the way of anything, let them flow.
However, in my box example, you would have to remove loops to create the circle, and you wouldn't be able to select the full loop because you need part of it for the insets.
I didn't have to remove the loops and do any clean up, because they weren't there in the first place. The shapes around the insets are Ngons, so the loops won't flow out of the insets into the Ngons outside anyway.
The same goes for all the posted wireframes, i didn't have to clean up anything, because i didn't start with a mess in the first place.
If you work correctly (clean), you shouldn't be doing much clean up at all and your final mesh should be pretty clean and still easy to work with or edit if necessary.
This way I can quickly use CSslide as a more efficient way of quickly moving my edgeloops nearer or further apart as I would manually with the gizmo and edge constraints applied to control the tightness or looseness of my edge curves.
Even when it comes to hard surface modelling I maintain this approach as giving everything a slight curve is far superior to the off/on equation that pure hard surface modelling enforces when it comes to the normal map being affected by specularity ingame.
While there are sections of my model I build purely as hard surface models generated from splines - bevel mofifier and what i call 'lazy' or 'generous' low poly editing and use of the smoothing groups , mostly i stick to the soft curves because they make better normal maps and that makes for better lighting effects ingame as theres more than an on / off option for those surfaces.
It's also helpful to let the edgeloops run around the model so that when you take a mesh or section of mesh into zbrush it will subdivide easier and you will not be restrained to working up more detail in certain sections only.
I think almost everyone at Epic is working similar to how I'm describing, I think it is only Josh Jay and myself that rely also on things like just modelling a lot of stuff raw and using smoothing groups for certain hard surface sections or then using the smoothing groups as the turbosmooth constraint instead of extra edgeloops.
r.
Personally I think it's whatever an artist prefers, and can work with quickly. There isn't a "right" or "better" way that's universally true for everyone. Each software package, and person, works differently. All that matters is the end result and the time it took to get it.
Or, pick and mix
r.
Also about edge loops, I tend to not have them "roam" around the model, but rather follow the curves I'm trying to achieve (I'm sure thats what you guys meant:)) just like with organics, u try and make your edges do what you need them to do, and create loops wherever possible, makes it easier to select stuff, then those edges can be chamfered to make them hard... or not
Also about edge loops, I tend to not have them "roam" around the model, but rather follow the curves I'm trying to achieve (I'm sure thats what you guys meant:)) just like with organics, u try and make your edges do what you need them to do, and create loops wherever possible, makes it easier to select stuff, then those edges can be chamfered to make them hard... or not
[/ QUOTE ]
i think that sums it up quite nicely.
As with modeling musculature, if you do it properly the edgeloops and topology will look pretty decent automatically. Of course, you can do it messy and still get a similar final look, but i wouldn't be happy with it personally.
I'm also guessing that you work mainly with hard surfaces and physical Subdividing (which afaik meshsmooth is) while i work mainly with SubDs from the start.
If you physically subdivide your models you end up with a relatively dense mesh on which it doesn't matter if you have loops around or not, with SubD, loops will often break something if you had them roam around.
I'm really curious about your wireframes, would your hard surface models look like the first model in http://www.strangefate.com/webby/modeling-plain2.jpg with a ton of unnecessary loops roaming aroud or more like http://www.strangefate.com/portfolio/elewire.jpg ?
You can't have loops roaming in SubD, there's too much tension going on and almost every edge has a reason to be there. Then again, you only really need loops to sharpen up edges, so there's no reason why they ever would flow into the rest of the model.
How would you model something with rounded surfaces and drills in those rounded surfaces. Would you physically subD the rough shape and then just drill the hard surface ?
http://www.strangefate.com/webby/modeling-SubD2.jpg
I don't think there's really any difference between "SubD" and "Meshsmooth", they're just different names for the same process, is that right? I'm not entirely sure...
*edit: ten it wouldn't matter having loops around, since you're not going back to SubD mode
To me it looks exactly like what you'd get if you made an Editable Poly object and used Meshsmooth/Turbosmooth on it... you can go back and forward between the low-poly and smoothed version (as seems to be the case with your train here), and nowhere in the process do you have to "commit" to smoothing... I would do hard surface detailing on the control mesh here, with edges close together for "hard" edges.
I'm not quite sure what you mean by hard surface modelling, is this a different process to SubD?
You could physically subdivide a SubD model and then just hack in additional geometry that wouldn't convert properly to SubD, but you wouldn't use anymore anyway from that point on.
So, meshsmooth doesn't physically subdivide your model, thanks for clearing that up
What i do is pretty much the same then, i work with the lowpoly cage and press TAB to switch to SubD and see if it all looks good and then just tab back and work on. Can work in SubD too, but that's rather painfull.
Then again, Per says he sometimes doesn't meshsmooth until he's done with all the modeling, which makes it sound like a physical definitive process.
I used to be really into LW now I get really confused after working with Maya and Max. Maya from LW was a pretty smooth transition but Max has ruined me, it's soo different
I still love the material editor in LW and Qemloss2 still destroys any native polygon reduction tool I've seen. Max Optimize and Multires is a joke.
Practice is better than scripts.
Okkun check Polygon Cruncher for Max and Lw.
imho.
[/ QUOTE ]
exactly. you haven't done or seen a lot of proper SubD modeling lately i guess, my fellow spanish brother ?.
haven't been there for a long time
[ QUOTE ]
Does LW still mess up the UV's with tab? Max used to have the same problem but now the default compensates and there's a check box for "old style mapping"
[/ QUOTE ]
I'm guessing that you mean the UV distorsion when switching from normal polygons to SubD ?
If so, it was fixed in one of the latest patches. You can now have subD UV maps with all the curves and stuff of the original subD model, and switch back and forth between normal and subD UVs.
LW is also getting Ngons and edge selection in the next patch or so, so if you liked it, it should do pretty well now. Or you can use modo like i do, Keys and tools are very similar, just better.
The cross-apps terminology has always been driving me nuts indeed. I think with your post and the bits of info before i can finally put it all together, and the truth is that... it's everywhere the same, max is just missing a quick tab function for polygon-SubD view... and i guess doesnt allow editing the cage in SubD mode.
Wow, i was confused for years i think.
I assume that a modifier is something like a morph target (?)
I've never used them much so i dunno if they would support that many operations to them as yuo describe.
[ QUOTE ]
ahhh now I think I understand your terms. what you mean by "hard surface" and "physical" is permanently applying subd and then continuing work, right? I never do this.. thought in some cases I guess I SHOULD do it..
[/ QUOTE ]
Yes and yes!
I sometimes almost feel like doing that too, just for speed when everything else is done.
[ QUOTE ]
I think pretty much 100% of what I do is subdivided twice, no more, no less.
[/ QUOTE ]
same here.
[ QUOTE ]
the lowermost shot has like a "cockpit" are in the upper right which is a curved surface with two hard lines cut into it. I always do stuff like that IN the sub-d... but on the condition that I get to alter the shape according to what is POSSIBLE to do in subd, cause a lot of the time it just won't be possible, you'll screw up the curvature
[/ QUOTE ]
Agreed. That ship was my 3th hipoly model ever, and done in LW. I didn't even try but i don't think i could have done all those details there with it's lack of Ngons and my crappy SubD modeling back then.
Nowadays i do everything without having to freeze the model ( = physically SubD the model in LW until it looks the same as in SubD) unless it has to be done yesterday.
[ QUOTE ]
The aesthetical cleanliness, it's a pointless masturbatory practice
[/ QUOTE ]
Agreed, but i dont have a gf.
Basically you have your lowpoly control mesh, and a togglable "meshsmooth" as it's sometimes referred to (there's a Max modifier called Meshsmooth which does the same thing, but NURMS smoothing is built into the EPoly functions). I have "S" as the hotkey for toggling between the smoothed version and the control mesh.
I guess similar to how you press TAB to toggle between the two, Strangefate.
http://www.spearmint.pwp.blueyonder.co.uk/NURMS1.jpg
on the left, the Editable Poly control mesh ... on the right, the NURMS subdivision toggled on, with 3 smoothing iterations, and Isoline display (basically only shows the original edges).
So I think when you talk about LW or Maya's "SubD" stuff, it's exactly the same as Max's EPoly and NURMS smoothing (aka Meshsmooth), but with different names.
I still don't quite understand the "hard surfaces" ... Per, any ideas?
I likewise have a toggle for smoothing on NURMS and a toggle for have the selected moddle display edges or not having them.
I actually try more these days to stick to 1 iteration of smoothing though when time is pressing and im not working much past 2 million polys i know i can afford the 2 iterations.
I'm glad all this cross program definition stuff got sorted out, it had never made much sense to me
r.
That looks pretty much like normal SubD Mop yes... from what was said tho, you can't continue to edit the mesh in max-SubD tho, it's just a look-but-dont-touch-thing ?
Hard surfaces... someone used that term in the other hipoly thread in 2D/3D so i just picked it up there. It's doing hipoly stuff without SubD. You just round everything with the bevel tool for the hipoly rounded look and that's it.
It's a lot faster when doing panels and other simple surfaces and dont have to care much about any loops and flow.
I'm glad too it got all sorted out, it's like everyone speaks a different language...
StrangeFate: The way this works in Max is, you can still edit the control mesh, even if you're viewing the smoothed (SubD) version... however this is usually pretty slow since there's a lot of polys in the scene on bigger objects.
Basically it displays the solid, smooth-shaded version as usual, but overlays an orange wireframe cage of your control mesh, you can see and edit the original control points while seeing the smoothed SubD version update in realtime in the viewport. So it's a "look AND touch" thing... but like I say, with more dense meshes it just runs really slowly even on a good computer, so I usually just edit the control mesh and toggle the smoothing on every now and then with a key-press, to see how it's looking in SubD.
Edit: Added image for more clarity:
http://www.spearmint.pwp.blueyonder.co.uk/NURMS2.jpg
Ideally I'd like to be able to just select edges and verts as they are located on the "Isoline display", rather than having the big orange cage which intersects the smoothed mesh and is generally nasty to work with.
Does SubD in Modo or LW allow you to edit the smoothed model without looking at the control mesh? Like add/remove edges or verts interactively on the smoothed surface?
I guess "Hard Surfaces" would just be, in Max, an Editable Poly with no smoothing applied ... you can do all the beveling and nGons just fine too, just don't bother to toggle the meshsmooth. Seems that way anyway.
I've never used them for much at all, maybe i should look deeper into them, i can imagine they're very usefull for hipoly, keeping a simple base cage. Never thought of using them that way.
[ QUOTE ]
StrangeFate: The way this works in Max is, you can still edit the control mesh, even if you're viewing the smoothed (SubD) version... however this is usually pretty slow since there's a lot of polys in the scene on bigger objects.
[/ QUOTE ]
Oh its the same as in LW/modo etc then. It does get slower too when switch to SubD. It's good to tweak curves and such while the mesh isn't too dense yet but later you dont really want to work in SubD.
I just turn off the orange wireframe of the cage so you never see that in my screenshots.
[ QUOTE ]
I guess "Hard Surfaces" would just be, in Max, an Editable Poly with no smoothing applied
[/ QUOTE ]
Yes, unless the person that used that term was referring to something completely different and i mixed it up as usual.
As tip, don't know if you know this new function of max..
Work on the edit poly modifier, just that
This is what i do to work like i were using modo. Not so powerful than Modo but very useful when i work with one iteration (i have not a workstation). I'm now working with Max and Modo toguether, trying to capt the Modo's Philosophy. I'm very customed to Max or expressed in other way my methodology of modelling fits perfectly Max's philosophy
hehe 3 iterations much?? people using Zbrush puts more than 4 hehe. I also use as base 2 iterations and 1 for modelling, but you people with workstations should not be worried of that. How many of you use a workstation card at work??
edit: Wait ... why do you need the extra Edit Poly modifier there? You can do just the same with a basic Editable Poly model with a Turbosmooth modifier, and Show End Result turned on...
here's a screen capture showing a model, i work only with the isoline mode because i have hidden the cage.
Per, with your backround you should look into scripting modo, you can do so much. Seneca from id keeps cranking out incredibly usefull scripts that are real time savers.
...omg, just saw that that new perfect circle script for all my stenciling love!
No idea if they will go the LW path, it seems more tho that everything would be integrated into the main app, then you'd just make your own UI with the elements you want or pick a defined one, like a modeling or UV UI now. I hope they go that path anyway.
I wouldn't want to have to communicate between several apps like in LW.
Anyways insta mesh smooth + messing with origional control mesh is very possible in XSI too just to let you guys know its damn simialar. Plus you get all the goodies like hard edges etc in real-time