View Full Version : Changing topology?

07 July 2002, 02:31 PM
If you have a fairly finished human head mesh(about 70% finished), and you realize the topology is not the best. Do you:

a) Keep working on that mesh. Change the topology by deleting a bunch of edges, moving verts around, and cutting new edges in to form the topology you want, and then tweak it back to good shape.

b) Start from scratch and work the new topology in from the base on up.

c) Get frustrated and go kick your neighbor's cat.

I'm facing that problem right now. I took choice b, and started from scratch. About 50% into it, I realized it might be faster to just tweak the old mesh. So, I went back to the old mesh and took route a.

Well, to tell you the truth, I'm not sure if one choice was better/faster than the other. The time I spent cutting in new topology on the old mesh and moving verts/deleting edges probably was about as long as it would've taken me to finish the 50% finished new head.

Oh well. At least I didn't kick the neighbor's cat. :)

07 July 2002, 03:12 PM
I would go route a.

Something I worked on for a while isn't worthless enough to simply throw away.
Tweaking a mesh is like working out a puzzle, in a way. It can be fun!
How complex is the mesh you have now? If it just something simple with a low poly count, It might be better to just start over.
I would throw it away if it REALLY sucked and gave me fysical pain.

07 July 2002, 03:18 PM
I would have to go with a and c. Rework the topology & kick the cat repeatedly:thumbsup:

07 July 2002, 05:53 PM
c always works for me

07 July 2002, 06:40 PM
yeah, a i think.
But c looks much easier, faster and, of course, sounds
like some big fun :D "Go for the window kitten! *kick*"
*boom* "Damn! she hit my neighbor" :surprised

keep working on your mesh, it's always worth a try.

07 July 2002, 11:04 AM
Either A or B is good. With A you already have a base to start with. With B you can skip over all the mistakes you made the first time, and probably have a lighter cleaner mesh.

But either way somewhere along the line C is going to look pretty inviting.;)

07 July 2002, 12:57 PM
if the mesh is so screwed up that it would take longer to fix it than to build it new i would go for a good c) and then do a b). otherwise, when you can fix it, fix it.

07 July 2002, 09:39 PM
(This was a tip for 2d, but why wouldn't it work in 3d too...)
If you choose b, then you should retain the wrong piece, maybe make it somewhat transparent and unselectable, so you can clearly see what you did wrong, and won't do it again.


07 July 2002, 10:59 PM
it would depend on the density of your mesh. if it is a reasonable ammount of geometry, i would say just rework it. by reasonable i mean not 'poly smoothed'!
i would say really really try to rework what you already have, even if its practicaly the whole mesh.
and make sure you do it right this time. edges fanning around the shoulders etc. etc.
good luck

07 July 2002, 11:00 PM
I'd never kick a cat.

11 November 2004, 09:26 PM
REcently had the same dilema and did a combo used the old mesh as an underlying structure and snapped the points of the new better topology to the surface of the old mesh to help spead up the process.

11 November 2004, 11:03 PM
Maybe something different... subdivide that mesh you have once or twice, then 'paint' new, correct topology using that new topology brush in Silo, or in Paraform if you have access to it.... This could work, there is demo of silo if i remember correctly.

11 November 2004, 12:09 PM
Silo's topology brush seems the way to go for this job.

11 November 2004, 03:21 PM
StephanD is true, Silo's enables to "paint" a topology over an existing mesh... Should be cool. Next time, you should try to work first on your model's topology and refine the mesh then. Your neighbor's cat could live a bit longer (I like cats :) ) if you do so. I'm working on a script for maya so as to fit a new topology to an high-res 3d scanned model... it would work in your case too.

11 November 2004, 11:11 AM
Only an idiot would kick a cat, i'd go with route A and change the topology.
Silo's topology brush is a good idea too..

11 November 2004, 06:35 PM
I'd go with b and c. For me it's impossibe to tweak anything after it's been finished because of the high polygon count i work with. To give all of you an idea, that ting in my avatar is a balisong or butterfly knife and it's only 76503 polygons.

11 November 2004, 06:47 PM
using lots of polygons doesnt make a good modeller ;). less is more.

11 November 2004, 07:04 PM
using lots of polygons doesnt make a good modeller ;). less is more.

I agree that a lot of polygons dont make a good modeller but unless you are working for the game industry less is not more. I use 3ds max and instead of using meshsmooth i make the mesh very detailed. I get better results.

11 November 2004, 07:28 PM
even in film, having too many polygons is bad. lets say your making a full cgi film, and you overdo the polygons for just one object, but you need to fill an entire screen with detailed objects?. take into account the number of polygons that will be on screen, then the textures for each object.. characters.. cloth.. etc. making things managable is the number one priority.

usually ill use displacements for microdetails, or just smart details in areas that need em. but never overdo the polygons for just one object.

11 November 2004, 04:07 PM
One painful aspect of changing topology is when I'm using morph targets. I'm looking forward to the day when topology-independent morph modeling is more widespread, ala Mirai & Modo. I haven't had the time to check out these tools in-depth yet, but hopefully the UVs, creases, and material assignments will also be propagated properly across all the targets when the topology needs to be altered drastically. Can anyone share their experiences with this?

11 November 2005, 12:49 AM
Well for me I like going back and starting over cause you learn from your mistakes so you end up not making the same mistakes you did the first time and then you get more accustomed to doing it the right way from scratch.

11 November 2005, 04:15 AM
Well, I tired option "C" once, and well...It just didn't seem to fix any of my modeling issues, so I would have to say don't bother with option c, it's a waste of time...and if your neighbour is bigger than you, it might waste more than just your time... o.0

11 November 2005, 04:16 PM
Hey, I'm the neighbours cat and I would choose either A or B.

11 November 2005, 09:25 AM
I use route A because I'll already have the existing head available for reference if I have to re-model the whole thing, which is never. I usually just need to change polyline direction in a few areas and leave the rest of the head alone.

11 November 2005, 04:47 PM
Download the spinFaces script from (then kick the cat)

Its perfect for adjusting topolgy without making a huge change to your model

11 November 2005, 03:16 PM
mm polyboost 1.5 for 3dmax also has a brush that lets you paint over another model, thus creating new polygons.

i think it was first introduced in 1.5 so i would advice you checking out that video.

i havent used it myself, but looking at the video it seems that it can also be used to create low poly versions using a high poly as base. )

CGTalk Moderation
11 November 2005, 03:16 PM
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.