PDA

View Full Version : modelling ettiquette?


assistant pimp
08-16-2005, 09:07 PM
Ok I need an opinion here. When I model (lets say a flat panel monitor) objects I tend to pull off pieces that require more geometry and place them on their own layer . I then do all my cutting, shifting, bandsawing, then transefer it back to the main layer. Then if it needs further editting I do a select connected cut it to another layer then bring it back. Now I did this precise thing for a small design gig. It rendered well and came cool. But I was told by someone with more knowledge than me that my method is not proper modelling and that I should try to keep everything as on big mesh. Is this correct? It does not make sense to me that if I have a button panel on my model that requires denser mesh in a concentrated area, and if I do a bandsaw it adds mesh around the whole object, in places like the side that is happy just being made of 4 polys. Is my thinking on this worng?. Wouldnt it be better to just add mesh where you need it then paste back into the main layer? I know there is no definitive way to do things, but I basically want to know the harm in what I am doing beside the fact that when you view my wire the mesh is dense in some areas and you can tell it is a separate piece.

Advice appreciated.

KillMe
08-16-2005, 09:18 PM
guess it comes down to the individual - sure its nice having everything as one clean mesh but sometimes that might not be practical or require a larger poly count which could hit rendertimes so - if your happy working that way and you dont need it to be all one mesh then cut it up

all that said i'm no pro but seems logical to me

Badllarma
08-16-2005, 09:18 PM
I keep everything as layers until I know the final model is correct then merge them together but it all depends on what I'm doing different on charaters to static models such as a house but IMO what ever works is the correct way it's like saying to get a great picture you have to paint in a cirtain way with a cirtain brush.................................all crap what matters is the finished article not the method of getting there.

assistant pimp
08-16-2005, 09:21 PM
thanks guys....I am asking because I am one day hoping to go into product/industrial design. Maybe working at a firm. And since I am teaching myself I dont want to develop bad habits that will not transfer well into the work world. And what i was told is that studios look at your meshes more than anything, and wont give you the time of day with chopped up meshes. Kinda scared me because I got pretty good at modelling this way.

uncon
08-16-2005, 09:43 PM
As far as the only real etiquette I have learned, always keep the model clean and efficient. It's very bad form to hand a model over to the rigger or animator that has a million unnecessary polygons and bogs down the scene. For closeup detailed work, I like a single mesh because it's seamless. Don't add detail where it won't be used/seen.

Also, no elbows on the desk, back straight, and make sure your pinkie is fully extended while typing, using the mouse and sipping tea.

Jarrede
08-16-2005, 10:05 PM
The only issue I can see coming up from that it's how it looks when it's rendered, depending on the angle that lights hit it, and what type of lights they are, having something like buttons a separate mesh can give results in the render that are un-expected. But, there are situations where A. the time it takes to incorporate high density meshes with low ones is too long and B. the total polycount doesn't benefit that much from it.

So, my answer is, it's not the "wrong" way to do it, just depends on the model.

blaqDeaph
08-16-2005, 11:48 PM
Firstly, ultimately, only the final result, and the time required in creation really matters. It doesn't matter (not much anyway) how the mesh looks, so long as the final render is fine. Yes, it's much better if the polyflow is right etc, but at the end of the day, the client isnt going to be seeing the polyflow anyway!

There's no right approach to modelling. Yes, there are certain steps that you can take to improve efficiency, but, art is art, and no artist should dictate the type of brush, or the color of paint that you use to draw.

As for cutting to another layer, modifying then pasting back. However, this created all sorts of problems for me as I kept forgetting to merge back the points.

IMHO, a slightly better approach then cutting and pasting is to use Hide and Unhide. Instead of selecting the polys, cutting to another layer, and pasting, I usually select the polys, and press [=], or hide unselected. This causes everything that I'm not selecting to be hidden. That way, I can apply all the tools to only the visible portion of the model, then press [\] to unhide all. The only down side to this, is that you can't see the rest of the model (as in background mode with the multi-layer approach).

Really, it's your choice though.

Jarrede
08-16-2005, 11:53 PM
I use hide and unhide a lot as well, and recommend it, although sometimes I've noticed if you have a really dense mesh, it's doesn't speed up opengl tumbling as much as it should.

plotz
08-17-2005, 12:12 AM
If you're talking industrial design I think they might have been referring to how the mesh is delivered...not how you model it.

I'm not an industrial designer, but I do a lot of work with parts and products that have been designed by industrial designers. When they send me a file it's usually in some kind of CAD/CAM format like .SAT. Some send models in VRML format too.

When they give me a mesh everything is in one object/layer but each functional component of that object has been assigned a unique surface so that I can quickly select it and copy it out for mesh reduction or clean up. (In Rhino or Maya)

Personally I wouldn't think it matters how you model something as long as you deliver the final mesh in the form and format that's expected.

Jarrede
08-17-2005, 12:49 AM
Personally I wouldn't think it matters how you model something as long as you deliver the final mesh in the form and format that's expected.

I agree.

@assistant pimp

In the past I've always had people try and tell me how proper modeling should be done, a lot of people will try and tell you all kinds of crazy **** just stick with what your comfortable with, often times techniques people have presented to me would compromise my confidence in getting it done on time and looking good, I've pissed off some supervisors in the process, but they've always been happy with the end result.

another thing to keep in mind that while most often it doesn't matter what kind of "paint" an artist uses, if your working with a team you should be sympathetic to what others prefer to work with, and if it doesn't take too long to accommodate to their needs it should be considered.

I've been thinking about the issue a bit more, and have run some tests. Basically I believe it depends on where your doing it in the mesh and how noticeable it is. Is surface smoothing going to break where it should continue? How close is the camera going to get to the seam, and can the object realistically be constructed that way in the sense of appearance? Is the object going to be deformed? Will the deformation cause the objects to behave poorly?

Most often or not when I run into this, the objects are connected in some way, but it's not feasible to stitch them together in time. If it's a hard surface object like you mentioned, you can usually get away with it. But if say they want to apply rigid bodies to the object, having separated meshes may give unpredictable rigid calculations. In some situations you'd want to break the object up into different pieces depending on how it's used.

My new answer is, if it looks good in the render after smoothing and texturing have been completed, and that's the only plan they have for the object then it's perfectly ok.

But if they want to apply softbody deformation, or rigid body deformation, having separate pieces may or may not be the best way to set the model up.

assistant pimp
08-17-2005, 07:33 AM
Good stuff guys....I am going to tryout modelling all in one mesh for a few objects to see if I can polish up on that. I am currently toying with it now, and first off I see it requires more planning. I still dont fully understand how to maintain poly flow but I may get it as I practice more. this is kinda depressing...I am still learning modelling after 3 mos of LW and I still have to learn texturing, morphs, animation......the list goes on. Seems like I am going to suck for a while.

Anyway thanks for the insight folks, it really helped me.

doc sampson
08-17-2005, 10:39 PM
If a client cuts you a paycheck it was modeling properly :)

blaqDeaph
08-17-2005, 11:57 PM
If a client cuts you a paycheck it was modeling properly :)

Depends on what you do really, most clients are more interested in the final render than the mesh (although some, like architects want the mesh as well).

Aegis Prime
08-18-2005, 12:04 AM
If a TD gets his/her shot out without having to rebuild and/or fix your model then it's modelled correctly...

otacon
08-18-2005, 01:13 AM
If a TD gets his/her shot out without having to rebuild and/or fix your model then it's modelled correctly...

That is soooo true.

essencedesign
08-18-2005, 04:28 AM
Agreed and agreed with above......I'm wondering what type of builds the person that told you this has been involved in......?


...sometimes when you just get going and building things can get a bit ...umm...of a mess.....but when your rolling along who want's to stop and clean up junk layers....

This is one of the things that makes LW's Modeler so user friendly.....

..My last build toped in at a bit of a wopping 150 something layers 3 million in modeler ...but then just flatten all the layers down , check for any 2 point polys or stray points....do a merge, unify align ect for your deliverable model,,,and keep your Build model as messy as you want.... :argh: :thumbsup:

cheers
J~

CGTalk Moderation
08-18-2005, 04:28 AM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.