hey guys quick question, I did a test bake today with just a cube. the highpoly had supporting edge loops and all in one SG for the first bake. the second bake the hp was turbosmoothed based off the same first HP. My bake gave the exact same result.
It got me wondernig, do you not always need to turbosmooth the mesh?
Sorry this may seem like a random stupid question >_>
hey guys quick question, I did a test bake today with just a cube. the highpoly had supporting edge loops and all in one SG for the first bake. the second bake the hp was turbosmoothed based off the same first HP. My bake gave the exact same result.
It got me wondernig, do you not always need to turbosmooth the mesh?
Sorry this may seem like a random stupid question >_>
This is sort of a strange question and really depends on exactly what your meshes look like really.
I mean, if you've got enough detail in your cage, or your texture is low resolution enough.... Really, if you're looking for someone to tell you that you don't need to learn how to use sub-d, I wouldn't expect that response. Sub-d is for a lot more than just beveling a cube.
sorry that's not exactly what i was implying, i just never knew in SOME circumstances, you can get the same results. i guess i was more trying to figure out why it gets the same result.
You can model by hand the same mesh that sub-d will create, the baker doesn't care how you modeled it, just what it looks like in the end.
Its really hard to say anything other than that without seeing how you created the mesh in the first place.
Here is a very simple example:
Up close it is obvious which one is a sub-d mesh, but if the object is very small, or the texture resolution is very low, the difference might not be noticeable at all.
I guess more then anything I'm just amazed I only just found out about this haha. Thanks for the explanation.
Its hard to say without seeing a smoothed shot without wires, but from the looks of it, your edges are just so hard, that there wouldn't be much difference either way.
EarthQuake, thanks for the help. I'll keep going as separate chunks and see how it turns out.
Because it is so flat and blocky, it would be extremely boring to see this asset in FPV in video game. Creating or choosing a good concept is such a huge part of making a good model.
Yeah, I'll finish this one and then find something more interesting to do. I start and never finish a lot of things lately, so bad concept or not, I'm going to keep going with this one. I don't usually go off other people's concepts either, so I thought I'd give that a try as well.
Yeah its certainly not a bad idea to finish the asset for the sake of finishing assets, lots of people have trouble with that especially with personal work, so I totally understand wanting to break the habit.
As far as using your own concepts go, IMO its more about picking a good one, than doing one yourself. I'm a crap concept artist myself, so I don't really ever model stuff that I design. If the end goal is to make a nice asset, for your portfolio or whatever, its gotta start with a great concept.
Yeah its certainly not a bad idea to finish the asset for the sake of finishing assets, lots of people have trouble with that especially with personal work, so I totally understand wanting to break the habit.
As far as using your own concepts go, IMO its more about picking a good one, than doing one yourself. I'm a crap concept artist myself, so I don't really ever model stuff that I design. If the end goal is to make a nice asset, for your portfolio or whatever, its gotta start with a great concept.
Thanks for tihe thoughts. I find it interesting that you don't model your own designs. For a long time I had no motivation to model anything that wasn't my own design. Luckily, I'm getting over that now.
I'll probably start a thread over in the previews section as I get a bit further with the current gun.
4 quick steps:
A. Make 6 sided cylinder
B. Bevel/inset Top/Bottom faces
C. Select and Bevel 6 edge loops(at once, of course)
D Loop slice, with 2 symmetric loops
And another one for fun, turn a cylinder into a hexagon.
A. Create 18 sided cylinder with 3 segments, scale to desired proportions
B. Select every other third edge, rotate +15 degrees
C. Select every other other third edge, rotate -15 degrees
D. Bevel/inset the top/bottom faces
E. Loop slice
^^
Same here, the highlight on the concept look really weird.
And it would be better if you could show is the smoothed version, since this topic is about sub-d modeling.
How about this one:
I am trying to cut out a door into a slightly curved surface.
This error is not very prominent but noticeable. There will be other details
added to the surface. I tried it already and it will not become better.
Most of the time I read that such behavior might be overcome with more geometry.
I smoothed/ subdivided the mesh twice, but as far as I see this just reduces the effect but not the problem itself.
It would be the perfect job for NURBS, but I am not really used to NURBS
and struggle at the most basic.
Honestly, the fact that you had to point out the error, means its minor enough not to worry about. It would be good to see your attempts to add further detail, without that its hard to give any additional advice.
Thanks for your input. I did an other run and interestingly it seems that positioning the vertices by
sliding along the edges minimizes the ugly shading no matter how bad the topology is.
I thought I had a rendered image of the error, seems that I dumped it. However I was so picky
about it because the surface is very reflective and the error is there pretty visible.
For the 3ds users, why don't you use a combinaison of smoothing groups and 2 turbosmooths (the one underneath being used with Smoothing Groups turned on) ?
Well if your using maya I say you could play around with extruding vertex on a sphere, kinda like how some the guys on here make diamonds for gun handles.
The second one one the left is kinda easy too(at least in max)
1.Make geosphere icosaeder with 2 segments
2.Select all verts and save selection
3.Tesselate once
4.load vert selection and invert it
5.Turn every invisible Edge
(6.Chamfer all edges a little bit if needed)
And the first one on the right is kinda easy too:
1. Pentagon Sphere(dodecahedron)
2. Turn face selection into vert selection and save the selection
3. Scale them in
4. Turn on Face selection again and inset again
5. Load saved selection and invert
6. Collapse and scale them out
(7.Chamfer if needed)
This is what comes out of it, if done right:
Edit:
And the last one on the right is just like a soccerball with more segments.
The verts in the middle of the 5-sided parts turned outwards and the verts of the 6-sided parts turned inwards
Alright chaps. Wondering if you kind people could help me out with this problem I've ran into.
It may be fairly simple to solve but as I'm fairly new to hard surface modelling its stumped me.
Ok Images.
So what Im trying to do is get a curved shape for the outline, I've highlighted this in green.
This isnt normally a problem, but the edges I have running along the side (Highlighted in red) are making the curved edge hard in places, giving it a stepped look when smoothed.
As the edges running along the side need edge loops for definition, there is no way I can lose them at the end.
Im stupmed, and have no idea of a way around this. Your help would be awesome.
Align the relevant support corners to the green line to make the curve. From what I see I think you still have them in straight zags from the side view.
And more geo for the long faces to support the curve.
Align the relevant support corners to the green line to make the curve. From what I see I think you still have them in straight zags from the side view.
And more geo for the long faces to support the curve.
(I think)
Cheers man. adding more geo worked. Its strange how adding more geo is the solution to most of the problems I'm having. Im so used to keeping things low poly its difficult to get used to.
One thing that adding more geo does get in the way of is the finer details.
On the model im working on at the minute there are screw holes. Adding more edge loops to help one part, often intersects the screw portion, and then affects how circular the screw is...obv.
Is there a way of getting round this? I've seen people just delete the edge loop before it meets the "screw".
Is this good practice? I suppose as long as the HP looks good, then it doesnt matter how neat your edge flow is??!
Hehe glad you like it. Nah, I started out with a hedra(Namely Dodec/Ico hedra in Max), from that
point I extruded inwards the triangular sections and extruded outwards the pentagons, after
adding the cuts on the extruded pentagons I chamfered the entire thing.
I need to model a tree for the scene below. The teacher wants me to to use animation deformers to get the curves and such... but I'm wondering how I should go about doing the rest. Like should I make a cylinder and extrude the faces to get the branches?
In the end I want to take it into zbrush but the teacher wants me to do as much as I can in maya.
A damn shame, that. Zspheres and dynamesh would make quick work out of a tree like that.
I know right! I even made a full tree in zbrush! I guess I just wish the teacher was more of the type that would say like "here is a tool, but you don't have to know it, you just need to know it is there when you need it" instead of forcing us to use it when it is easier for me to do it another way.
But he is the boss! So I need to figure what the best way to do this is. Anybody?
I suppose a better question would be how would you model triangular shapes using sub-d. Its not going to be imported to ZBrush so It doesn't matter that much at the minute. But for future refernce, how would you go about doing this?
Not like the shape in question is "that" triangular, it kinda just tapers back into the grip.
For such shapes and for further tips I would recommend you Grant Warwick's hard surface essentials (http://vimeo.com/10941211) if you haven't seen it yet.
Replies
It got me wondernig, do you not always need to turbosmooth the mesh?
Sorry this may seem like a random stupid question >_>
This is sort of a strange question and really depends on exactly what your meshes look like really.
I mean, if you've got enough detail in your cage, or your texture is low resolution enough.... Really, if you're looking for someone to tell you that you don't need to learn how to use sub-d, I wouldn't expect that response. Sub-d is for a lot more than just beveling a cube.
You can model by hand the same mesh that sub-d will create, the baker doesn't care how you modeled it, just what it looks like in the end.
Its really hard to say anything other than that without seeing how you created the mesh in the first place.
Here is a very simple example:
Up close it is obvious which one is a sub-d mesh, but if the object is very small, or the texture resolution is very low, the difference might not be noticeable at all.
here's mine that I took for you:
I guess more then anything I'm just amazed I only just found out about this haha. Thanks for the explanation.
Found what? I don't get what you are saying
Its hard to say without seeing a smoothed shot without wires, but from the looks of it, your edges are just so hard, that there wouldn't be much difference either way.
Yeah, I'll finish this one and then find something more interesting to do. I start and never finish a lot of things lately, so bad concept or not, I'm going to keep going with this one. I don't usually go off other people's concepts either, so I thought I'd give that a try as well.
Thanks again.
As far as using your own concepts go, IMO its more about picking a good one, than doing one yourself. I'm a crap concept artist myself, so I don't really ever model stuff that I design. If the end goal is to make a nice asset, for your portfolio or whatever, its gotta start with a great concept.
Thanks for tihe thoughts. I find it interesting that you don't model your own designs. For a long time I had no motivation to model anything that wasn't my own design. Luckily, I'm getting over that now.
I'll probably start a thread over in the previews section as I get a bit further with the current gun.
Thanks again.
BOLTS! how do you make a hexagon bolt with smooth edges!?
we'll i did a lot of faffin around and made this
it took far to long and involved loop, extrudes and deleting edges. i was going to use this on a recent model but decided to make a new one!
I figured out a faster way. its all quads!
A. Make 6 sided cylinder
B. Bevel/inset Top/Bottom faces
C. Select and Bevel 6 edge loops(at once, of course)
D Loop slice, with 2 symmetric loops
And another one for fun, turn a cylinder into a hexagon.
A. Create 18 sided cylinder with 3 segments, scale to desired proportions
B. Select every other third edge, rotate +15 degrees
C. Select every other other third edge, rotate -15 degrees
D. Bevel/inset the top/bottom faces
E. Loop slice
Heres a nutty bake too:
Thats amazing! the part where you used the objects centre to rotate the edges into place :thumbup:
Thanks
and this is the ugly result I am getting T-T
While waiting for a reply I shall continue trying to work it out ^-^ if I get It right I shall post how I did so.
Same here, the highlight on the concept look really weird.
And it would be better if you could show is the smoothed version, since this topic is about sub-d modeling.
I am trying to cut out a door into a slightly curved surface.
This error is not very prominent but noticeable. There will be other details
added to the surface. I tried it already and it will not become better.
Most of the time I read that such behavior might be overcome with more geometry.
I smoothed/ subdivided the mesh twice, but as far as I see this just reduces the effect but not the problem itself.
It would be the perfect job for NURBS, but I am not really used to NURBS
and struggle at the most basic.
sliding along the edges minimizes the ugly shading no matter how bad the topology is.
I thought I had a rendered image of the error, seems that I dumped it. However I was so picky
about it because the surface is very reflective and the error is there pretty visible.
1.Make geosphere icosaeder with 2 segments
2.Select all verts and save selection
3.Tesselate once
4.load vert selection and invert it
5.Turn every invisible Edge
(6.Chamfer all edges a little bit if needed)
And the first one on the right is kinda easy too:
1. Pentagon Sphere(dodecahedron)
2. Turn face selection into vert selection and save the selection
3. Scale them in
4. Turn on Face selection again and inset again
5. Load saved selection and invert
6. Collapse and scale them out
(7.Chamfer if needed)
This is what comes out of it, if done right:
Edit:
And the last one on the right is just like a soccerball with more segments.
The verts in the middle of the 5-sided parts turned outwards and the verts of the 6-sided parts turned inwards
i thought he was joking when he posted the sphere images. i impressed
It may be fairly simple to solve but as I'm fairly new to hard surface modelling its stumped me.
Ok Images.
So what Im trying to do is get a curved shape for the outline, I've highlighted this in green.
This isnt normally a problem, but the edges I have running along the side (Highlighted in red) are making the curved edge hard in places, giving it a stepped look when smoothed.
As the edges running along the side need edge loops for definition, there is no way I can lose them at the end.
Im stupmed, and have no idea of a way around this. Your help would be awesome.
Thanks guys. :poly127:
And more geo for the long faces to support the curve.
(I think)
Cheers man. adding more geo worked. Its strange how adding more geo is the solution to most of the problems I'm having. Im so used to keeping things low poly its difficult to get used to.
One thing that adding more geo does get in the way of is the finer details.
On the model im working on at the minute there are screw holes. Adding more edge loops to help one part, often intersects the screw portion, and then affects how circular the screw is...obv.
Is there a way of getting round this? I've seen people just delete the edge loop before it meets the "screw".
Is this good practice? I suppose as long as the HP looks good, then it doesnt matter how neat your edge flow is??!
Thanks again man.
Here's mine:
point I extruded inwards the triangular sections and extruded outwards the pentagons, after
adding the cuts on the extruded pentagons I chamfered the entire thing.
I need to model a tree for the scene below. The teacher wants me to to use animation deformers to get the curves and such... but I'm wondering how I should go about doing the rest. Like should I make a cylinder and extrude the faces to get the branches?
In the end I want to take it into zbrush but the teacher wants me to do as much as I can in maya.
Thanks for your help
I know right! I even made a full tree in zbrush! I guess I just wish the teacher was more of the type that would say like "here is a tool, but you don't have to know it, you just need to know it is there when you need it" instead of forcing us to use it when it is easier for me to do it another way.
But he is the boss! So I need to figure what the best way to do this is. Anybody?
Its ruining me. Thanks for ur help.
Yeah, the only thing thats throwing me off atm is adding the supporting edges.
The gun is pretty chiseled at these points, and adding supporting edges to the tris is kinda throwing me off. (Still a little new to Sub-D).
No matter where I put them they kinda skew and warp when smoothed. :poly127:
Here it is mate.
I suppose a better question would be how would you model triangular shapes using sub-d. Its not going to be imported to ZBrush so It doesn't matter that much at the minute. But for future refernce, how would you go about doing this?
Not like the shape in question is "that" triangular, it kinda just tapers back into the grip.
Thanks guys.
Triangles are becoming the bain of my life, and I used to like them....as a shape.
here is how i did this piece I hope this helps. Let me know if not and I can change to help your needs better
but basically I made the triangular corner that a loop all to its own
hope this helps
I would probably go the floater route if its not a huge part of the model