Hi all, I am new here... and wanted to ask some 3d veterans about this.
If you create complex models for game assets or artwork, do they need to be attached to the same single overall mesh? Sometimes it seems so much easier just to let them "float" close or intersect them (I know clipping is a sin, but just throwing it out there).
For instance, if you had a cylinder with 40 faces and wanted to attach it to a plane with only four vertexes, is it worth it to remesh the plane and tweak it until they meld together into a single overall mesh? After that, say you need to attach a round ball to the end of the cylinder... do you do the same process and work out all the edges and verts to connect as seamlessly as possible? These are just some simple examples, but some objects get very complex. Do you spend all that time ensuring you have no extra faces or edges and pray that it subdivides nicely?
I spend a lot of time trying to ensure I never get ngons and everything is a quad, and I am curious if that is what most people do when learning to make complex objects. I feel like I am making some fundamental mistakes.
TLDR: Do single objects with many complex parts/pieces need to be a single continuous mesh? Thanks!
I'm not a veteran but if you use (simple) clipped geometry and the engine where you use it in has no problems with it .. and you also don't want to make any smooth transition.. even only in the texture.. then again you may be fine with it.. if you wanna make a subdivided mesh for getting some normal map.. well this would not.. work.. and so on..
There may be also a shorter (and general) answer:
It depends.
And: There might be some veteran thinking of this already..
...on what that cylinder and plane represent, on how close you get to it and/or how large it is, on how much time you have to produce the asset, etc.
Clipping isn't a sin. It potentially wastes texture space, but balancing efficient use of resources with efficient use of your time and energy is something you will constantly be doing in production.
Ngons are ok if they triangulate well, and quads are only relevant to animation and sub-d meshes (and sub-d can handle ngons if they're on flat surfaces).
With experience you'll get a feel for how things subdivide.
Making objects for games means the more you can optimize them the better for the level. Making artwork I take to mean designs or objects you may eventually want an optimized model from, or just need for a commercial shot and/or anything in between. Below is a pretty complex object.
Its a hipoly model in zBrush and made of loads of separate parts. If you wanted it for a still image texturing it would be enough and it could remain a disconnected model. If you wanted a short animation you could rig it so only the parts you needed to move (like the trigger or magazine) could be rigged to do that. If you wanted it for a lengthier or a reoccurring animation like a tv series or a bunch of ads you would optimize the mesh and perhaps use a displacement approach that would help lower the rez for animation work. If you wanted it for a game you would build a low rez mesh around this hipoly and use it to extract textures that you can use with materials and describe the surface detail. If you wanted an optimized game model you could animate, then the same as above, make a low poly and separate low polys for the bits that need to move.
Replies
Hi all, I am new here... and wanted to ask some 3d veterans about this.
If you create complex models for game assets or artwork, do they need to be attached to the same single overall mesh? Sometimes it seems so much easier just to let them "float" close or intersect them (I know clipping is a sin, but just throwing it out there).
For instance, if you had a cylinder with 40 faces and wanted to attach it to a plane with only four vertexes, is it worth it to remesh the plane and tweak it until they meld together into a single overall mesh? After that, say you need to attach a round ball to the end of the cylinder... do you do the same process and work out all the edges and verts to connect as seamlessly as possible? These are just some simple examples, but some objects get very complex. Do you spend all that time ensuring you have no extra faces or edges and pray that it subdivides nicely?
I spend a lot of time trying to ensure I never get ngons and everything is a quad, and I am curious if that is what most people do when learning to make complex objects. I feel like I am making some fundamental mistakes.
TLDR: Do single objects with many complex parts/pieces need to be a single continuous mesh? Thanks!
I'm not a veteran but if you use (simple) clipped geometry and the engine where you use it in has no problems with it .. and you also don't want to make any smooth transition.. even only in the texture.. then again you may be fine with it.. if you wanna make a subdivided mesh for getting some normal map.. well this would not.. work.. and so on..
There may be also a shorter (and general) answer:
It depends.
And: There might be some veteran thinking of this already..
;-)
It depends... 🤣
...on what that cylinder and plane represent, on how close you get to it and/or how large it is, on how much time you have to produce the asset, etc.
Clipping isn't a sin. It potentially wastes texture space, but balancing efficient use of resources with efficient use of your time and energy is something you will constantly be doing in production.
Ngons are ok if they triangulate well, and quads are only relevant to animation and sub-d meshes (and sub-d can handle ngons if they're on flat surfaces).
With experience you'll get a feel for how things subdivide.
Making objects for games means the more you can optimize them the better for the level. Making artwork I take to mean designs or objects you may eventually want an optimized model from, or just need for a commercial shot and/or anything in between. Below is a pretty complex object.
Its a hipoly model in zBrush and made of loads of separate parts. If you wanted it for a still image texturing it would be enough and it could remain a disconnected model. If you wanted a short animation you could rig it so only the parts you needed to move (like the trigger or magazine) could be rigged to do that. If you wanted it for a lengthier or a reoccurring animation like a tv series or a bunch of ads you would optimize the mesh and perhaps use a displacement approach that would help lower the rez for animation work. If you wanted it for a game you would build a low rez mesh around this hipoly and use it to extract textures that you can use with materials and describe the surface detail. If you wanted an optimized game model you could animate, then the same as above, make a low poly and separate low polys for the bits that need to move.
More considerations here
http://wiki.polycount.com/wiki/Normal_Map_Modeling#Low-Poly_Mesh