Here is my last wip. This method has been a revelation for me and coupled with dynamesh the workflow makes total sense. Thanks Amsterdam Hilton Hotel for sharing this!: Cheerio
Instead of zbrush for the chamfers, couldn't opensubdiv be used to the same effect? An edit poly on top of the stack and a few edge weights could handle a lot.
Instead of zbrush for the chamfers, couldn't opensubdiv be used to the same effect? An edit poly on top of the stack and a few edge weights could handle a lot.
Not to the same effect. See my old tut on this workflow a few pages back. It uses quadify+turbosmooth. It works, but falls short on very complex meshes (such as Kanga's great model above) open subdiv would be a nightmare trying to set up the creaseset with shitty boolean topo.
After converting to Dynamesh in ZBrush half of my mesh is gone ... it just disappears? This happens only with very complex meshes. Most of the time it just works fine. Any ideas?
Hey guys. I did a complete prop using primordially this workflow, and i really like it. Thanks so much because now i am much more secure to go for Hard Surface objects!!! So here is a picture from the final low poly mesh, i really want to share with you all. And again, thank you for this thread. So much helpful!
Original concept from Leading Light Entertainment.
After converting to Dynamesh in ZBrush half of my mesh is gone ... it just disappears? This happens only with very complex meshes. Most of the time it just works fine. Any ideas?
Been playing around with this method of modeling recently, I've noticed sometimes I would encounter problems where after going into proboolean to subtract my mesh it would just dissapear. Can't seem to figure out what is causing this, no ngons in mesh, reset xform, still same issue.
Been playing around with this method of modeling recently, I've noticed sometimes I would encounter problems where after going into proboolean to subtract my mesh it would just dissapear. Can't seem to figure out what is causing this, no ngons in mesh, reset xform, still same issue.
Has anyone ecountered this?
Open edges is the first thing that comes to mind, a quick border select will catch those. I've seen high poly ngons as well, but you've already controlled for that. A stl check modifier can also find stuff that isn't easily visible.
After reading this thread, i've made a script in maya to enhance boolean
workflow. I will probably post it on gumroad if anyone is interested. Here's a video
showing how it works https://www.youtube.com/watch?v=ZdRhp6ETA_o
After reading this thread, i've made a script in maya to enhance boolean
workflow. I will probably post it on gumroad if anyone is interested. Here's a video
showing how it works https://www.youtube.com/watch?v=ZdRhp6ETA_o
I've never tried this workflow before as I only ever really model using SubD but damn...It looks like cheating, seeing it in action. Might give this a try and great work on that plugin by the way
Been playing around with this method of modeling recently, I've noticed sometimes I would encounter problems where after going into proboolean to subtract my mesh it would just dissapear. Can't seem to figure out what is causing this, no ngons in mesh, reset xform, still same issue.
Has anyone ecountered this?
Open edges is the first thing that comes to mind, a quick border select will catch those. I've seen high poly ngons as well, but you've already controlled for that. A stl check modifier can also find stuff that isn't easily visible.
Hey Kary, you were spot on with this.. seems that symmetry created some open edges that I wasn't aware of. Cleaned it up and proboolean worked like charm. thanks for the tip!
Been playing around with this method of modeling recently, I've noticed sometimes I would encounter problems where after going into proboolean to subtract my mesh it would just dissapear. Can't seem to figure out what is causing this, no ngons in mesh, reset xform, still same issue.
Has anyone ecountered this?
Open edges is the first thing that comes to mind, a quick border select will catch those. I've seen high poly ngons as well, but you've already controlled for that. A stl check modifier can also find stuff that isn't easily visible.
Hey Kary, you were spot on with this.. seems that symmetry created some open edges that I wasn't aware of. Cleaned it up and proboolean worked like charm. thanks for the tip!
That's why booleans are also referred to as solids.
After reading this thread, i've made a script in maya to enhance boolean
workflow. I will probably post it on gumroad if anyone is interested. Here's a video
showing how it works https://www.youtube.com/watch?v=ZdRhp6ETA_o
After reading this thread, i've made a script in maya to enhance boolean
workflow. I will probably post it on gumroad if anyone is interested. Here's a video
showing how it works https://www.youtube.com/watch?v=ZdRhp6ETA_o
The main issue is that a lot of people think this is a way to avoid having to learn proper modeling, totally in the spirit of unambitious, lazy computer artists. They may not be able to tell that the only reason an alternate workflow seems to give them good results is that their sub-d skill levels are atrocious.
+1 for this.
I only do the boolean-zbrush to get faster result for complex shapes. This workflow suffers from hardware limitation if you use it too much. I'm still combining regular subdivision modeling with boolean workflow.
In the third last of the video, where Michael imported cad data and made the polygroups manually, yielded the best result. it's very time consuming but the result is much smoother. I found that it's faster if I created autogroups with UV. Just give every hard-edge their own uv island, no need to unfold or pack them.
How do you get the dynamesh to play properly once the booleaned mesh is all faceted out when dealing with round objects? When the dynamesh smooths out the facets don't go away.
The Polish step after Dynamesh will smooth them out as long as the underlying curves have enough segments. Use high segment counts in the boolean operands.
The Polish step after Dynamesh will smooth them out as long as the underlying curves have enough segments. Use high segment counts in the boolean operands.
I just re read the tutorial and found that part. Thanks!
Great Tutorial @Amsterdam Hilton Hotel ! One question guys - I'm making a lot of these sheet metal car parts and this method works really well with them, with one small exception. After making the "sheet", applying Shell modifier in MAX , sending the model to Zbrush and smoothing it, the edges of the sheet get all rounded and not very "metal"-y (upper left on the pic). Do you know a way to prevent this and keep only the edges sharp?
@jazznazz I think this is one of the main caveats of the method, but I believe there are some workarounds for this.
One would be to make the shell strip a different UV shell from the top and bottom then use polygroups by UV, then hide the shell border, apply a polish to the main areas you want rounded, inverse polygroup, polish edge slightly to have sharper result.
OP examples are totally convincing. I think it's a great workflow for "booleanable" objects, which can almost be totally automated. It's close from how CAD softwares work (and mechanicals parts are made). Better than MR round corner trick, better than double subd trick, exponentially faster than chamfering subd. You can sculpt easily on it (weldings and stuff) as the mesh is homogeneous. It's non destructive, let you re-use your booleans setup for the low poly... If you assign a multi/sub material to operands, mat IDs will be preserved (you can use the material modifier to assign an id without converting to poly). Could be used later to polish by polygroups, etc...
For those that use 3DCoat instead of ZBrush for this workflow, is the best way to get a high poly model from a voxel just to use the autotopo? Is there a way to control how much the smooth all option smooths the mesh?
That looks great. Not having to jump to ZB would be nice. I suppose the cleanup in advance adds a little work although it's probably deductible from the LP creation time.
Cheers Ben. Yes, there is a little cleanup, but once you plan your boolean and know what to look out for before and after running the script it works well. Plus, we still have all the options from the chamfer mod available.
I saw that Blender can do this step too. I have found this maxscript and it seems the C++ source code is available. So if anyone can make a modifier out of it, and then combine it with Turbosmooth and pro-optimizer modifier after that you would not need to export to zbrush. I understand that the Polyherz quad chamfer workflow works well in most cases, but this workflow could also add some flexibility.
Dynamesh at a higher resolution, if you're getting issues like that. @Dan Powell
Will try this on the next object, cheers
That's the reason Dynamesh Master is recommended. Instead of guessing the Dynamesh resolution you just type the number of polys you aim for and it will try its best to deliver
Hey guys, blown away by this thread but a question from an amateur: I think the only doubts I have with this method is that, is this a good method I should learn (as a noob in modeling)?? On one hand, I feel like this is amazing and believe many things can be created (some even seem impossible to do in sub-d w/o giving you a migraine). Then, on the other hand, I'm wondering if it's bad to learn this because people will think it's lazy and bad because of the topology (like the "quads quads quads!") but I've heard somewhere that this method was used to make The Divisions weapons??! unless.... it's actually really easy but manual to get nice topology afterwards... please shine some light to my ignorance xD (side note: ... i dont even understand where the low poly comes from when making high poly ;~;)
Actually the OP gun has smooth rounded edges, not crisp sharp ones. It's common practice to exaggerate the smoothness/width of hipoly edges used to bake down to realtime assets.
Yeah, I get the whole exagerrating smoothness part. Bakes edges generally come out better when you slightly over-soften edges.
This is the image I'm referencing when I talk about crisp edges after dynamesh:
Notice how crisp and perfect all of the edges are on this model after the Dynamesh?
Unless I'm misunderstanding... This model at that point in the opening post has been dynameshed, hasn't it? D=
@Dan Powell are you referring to the visible faceting? If so, that's caused by not having enough segments on the cylinder primitives used as boolean operands. You have to set them pretty high to relieve this.
@Musashidian, no - I'm referring to the ugly results on my own test with Dynamesh (see a few posts above.)
@Perna, fair enough yeah - I'll probably give this a go with Dynamesh Master instead to see if that yields a better result for me; just a bit confused as why generally people's meshes are coming out super clean with Dynamesh, like the one of the gun in the opening post in comparison to mine. Cheers though!
If you want crisper edges, just Dynamesh with a high resolution.
Basically the edges will still look like the edges you were getting, but the geometry will be so densely packed that each face becomes small enough to the point where those imperfections wont even be visible unless you zoom way in.
Notice how crisp and perfect all of the edges are on this model after the Dynamesh?
Unless I'm misunderstanding... This model at that point in the opening post has been dynameshed, hasn't it? D=
You can see how to use dynamesh with boolean objects at the end of this video. Maybe that can help. The secret is to use polygroups before you dynamesh. After that, you can use the polish by features to make your bevels.
Yeah, I get the whole exagerrating smoothness part. Bakes edges generally come out better when you slightly over-soften edges.
This is the image I'm referencing when I talk about crisp edges after dynamesh:
Notice how crisp and perfect all of the edges are on this model after the Dynamesh?
Unless I'm misunderstanding... This model at that point in the opening post has been dynameshed, hasn't it? D=
Bear in mind that changing the size of a mesh effectively changes the Dynamesh resolution. Compare the size of the mesh to the ground plane. If you kept the 256 resolution but decreased the size you'd end up with a mesh with less geometry and softer edges.
Notice how crisp and perfect all of the edges are on this model after the Dynamesh?
Unless I'm misunderstanding... This model at that point in the opening post has been dynameshed, hasn't it? D=
Yeah that was post-dynamesh. The issue in the image you posted is insufficient polygon density for the object. You can solve it by increasing either the Dynamesh resolution or the absolute size of the object
Hey guys, blown away by this thread but a question from an amateur: I think the only doubts I have with this method is that, is this a good method I should learn (as a noob in modeling)?? On one hand, I feel like this is amazing and believe many things can be created (some even seem impossible to do in sub-d w/o giving you a migraine). Then, on the other hand, I'm wondering if it's bad to learn this because people will think it's lazy and bad because of the topology (like the "quads quads quads!") but I've heard somewhere that this method was used to make The Divisions weapons??! unless.... it's actually really easy but manual to get nice topology afterwards... please shine some light to my ignorance xD (side note: ... i dont even understand where the low poly comes from when making high poly ;~;)
I think you should learn it. Then you can decide whether to keep using it based on direct experience. This also goes for subdivision modeling.
The techniques that other people call lazy are the ones you should try and learn immediately. Spending less effort for the same result is a quality that's easily lost on people who spent time mastering less efficient techniques that are now obsolete.
They are no 'lazy-techniques' when you get the end result you want and you can easily make adjustments afterwards. However, you should learn the basics of Poly-modeling/subdiv since some shapes are still easier to create that way. You have better control over large complex forms with poly modeling. Smaller forms and details can be done using faster techniques.
10 years ago I knew Polymodeling would be outdated in the way it was used at the time. I made some predictions for myself and worked based on that; Polymodeling would be replaced in part with faster techniques; Auto-retopo would become availible (Like Z-Remesher and Dynamesh); Auto UVW would become availible; Auto-packing of UVS's;. Everything has come true in the last 10 years. I lost quite a bit of time learning stuff I knew would be less useful later on. Therefore I focused on learning the basics and never went overboard learning Poly-modeling like it was used 8-10 years ago. When a new technique presents itself, or if I think of such a technique myself, I start using it.
Hi guys, how are you doing retopo for low poly mesh? I use a quad drow in Maya, but there is a problem due to non-flat surfaces. This leads to small gradients on the normal map. When I do the traditional method, I have remained low poly (first level subdiv) with a perfectly flat surface.
forgive me if I missed it but is there any nice way to achieve similar fancy booleans in maya? the tools dont seem great in maya - crashes a lot and doesnt seem very non destructive.
Replies
Cheerio
Original concept from Leading Light Entertainment.
Has anyone ecountered this?
https://www.youtube.com/watch?v=ZdRhp6ETA_o
Hey Kary, you were spot on with this.. seems that symmetry created some open edges that I wasn't aware of. Cleaned it up and proboolean worked like charm. thanks for the tip!
http://polycount.com/discussion/172741/dcboolmanager-free-maya-script-for-boolean-modeling#latest
I only do the boolean-zbrush to get faster result for complex shapes. This workflow suffers from hardware limitation if you use it too much. I'm still combining regular subdivision modeling with boolean workflow.
In the third last of the video, where Michael imported cad data and made the polygroups manually, yielded the best result. it's very time consuming but the result is much smoother. I found that it's faster if I created autogroups with UV. Just give every hard-edge their own uv island, no need to unfold or pack them.
Just my 2 cents.
One question guys - I'm making a lot of these sheet metal car parts and this method works really well with them, with one small exception. After making the "sheet", applying Shell modifier in MAX , sending the model to Zbrush and smoothing it, the edges of the sheet get all rounded and not very "metal"-y (upper left on the pic). Do you know a way to prevent this and keep only the edges sharp?
One would be to make the shell strip a different UV shell from the top and bottom then use polygroups by UV, then hide the shell border, apply a polish to the main areas you want rounded, inverse polygroup, polish edge slightly to have sharper result.
Better than MR round corner trick, better than double subd trick, exponentially faster than chamfering subd. You can sculpt easily on it (weldings and stuff) as the mesh is homogeneous. It's non destructive, let you re-use your booleans setup for the low poly...
If you assign a multi/sub material to operands, mat IDs will be preserved (you can use the material modifier to assign an id without converting to poly). Could be used later to polish by polygroups, etc...
http://www.scriptspot.com/3ds-max/scripts/instant-meshes-bridge#comment-33954
Whenever I dynamesh an object, I always get ugly edges like this:
Instead of nice clean edges
Cheers
Did you polish after the dynamesh?
Nope, that would smooth out the edges wouldn't it? I was going for something with pristine crisp edges like the gun piece in the OP
I think the only doubts I have with this method is that, is this a good method I should learn (as a noob in modeling)??
On one hand, I feel like this is amazing and believe many things can be created (some even seem impossible to do in sub-d w/o giving you a migraine). Then, on the other hand, I'm wondering if it's bad to learn this because people will think it's lazy and bad because of the topology (like the "quads quads quads!") but I've heard somewhere that this method was used to make The Divisions weapons??!
unless.... it's actually really easy but manual to get nice topology afterwards...
please shine some light to my ignorance xD (side note: ... i dont even understand where the low poly comes from when making high poly ;~;)
Notice how crisp and perfect all of the edges are on this model after the Dynamesh?
Unless I'm misunderstanding... This model at that point in the opening post has been dynameshed, hasn't it? D=
@Perna, fair enough yeah - I'll probably give this a go with Dynamesh Master instead to see if that yields a better result for me; just a bit confused as why generally people's meshes are coming out super clean with Dynamesh, like the one of the gun in the opening post in comparison to mine. Cheers though!
Basically the edges will still look like the edges you were getting, but the geometry will be so densely packed that each face becomes small enough to the point where those imperfections wont even be visible unless you zoom way in.
You can see how to use dynamesh with boolean objects at the end of this video. Maybe that can help.
The secret is to use polygroups before you dynamesh. After that, you can use the polish by features to make your bevels.
watch at 4'40"
I think you should learn it. Then you can decide whether to keep using it based on direct experience. This also goes for subdivision modeling.
The techniques that other people call lazy are the ones you should try and learn immediately. Spending less effort for the same result is a quality that's easily lost on people who spent time mastering less efficient techniques that are now obsolete.
10 years ago I knew Polymodeling would be outdated in the way it was used at the time. I made some predictions for myself and worked based on that; Polymodeling would be replaced in part with faster techniques; Auto-retopo would become availible (Like Z-Remesher and Dynamesh); Auto UVW would become availible; Auto-packing of UVS's;. Everything has come true in the last 10 years. I lost quite a bit of time learning stuff I knew would be less useful later on. Therefore I focused on learning the basics and never went overboard learning Poly-modeling like it was used 8-10 years ago. When a new technique presents itself, or if I think of such a technique myself, I start using it.
it looks pretty good if it works as advertised and without crashing a lot like mayas other booleans tended to do when I tried them in the past.
http://polycount.com/discussion/164254/maya-surface-based-asset-generation/p1