Recommendation for software that provides quick, easy solution to retopology?
NightmareHero
Posts: 18
I do alot of group modeling, where I box model some pieces and mishmash them together to make my 3d models, akin to Lego modeling. As usual I have been told that this is not the standard way of modeling and makes the mesh hard to uv map because there are overlapping elements of the obj groups that are inside and not seen from the outside facing meshes of a multigroup obj mesh that I model in Hexagon. I have been looking at retopology as a solution to turn my existing non-manifold meshes into singular meshes that can be easily UVed and turn out AO bake maps without errors.
So far the solutions I have seen are manual and tedious, then I remembered Poser Tool Box and a certain feature of fitting a cylinder mesh to the contours of a female poser figure to create a makeshift dress/skirt.
My question is this, is there some type of algorithm, that can express what was done for turning the hollow cylinder mesh into a dress, but apply it to a sphere, and literally, analogously, fit the contours of the non-manifold mesh to register all the external faces correctly, and subdivide the sphere till there are enough faces to express the mesh without internal geometry?
If anyone of you know of such a program already in existence that avoids the hassle of doing retopology manually please let me know, for anyone wanting to make some money on noob friendly programs, take this as food for thought to make a program just like that. I'd be willing to pay $50 for a program that ONLY does that.
Daz 3D is part of 
Connect
DAZ Productions, Inc.
224 S 200 W, Suite #250
Salt Lake City, UT 84101
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
Not sure I quite understand.. Your described method of modling sounds like it would be just fine when used to make certain types of objects, what are you modelling and what do you mean by over lapping elements? I dont get why can't you weld and bridge these seperate parts together.
Have you got some screen grab examples handy?
That would make a nightmarish mesh for all but the simplest of objects
Not for beginners but perhaps Shrinkwrap methods might help you out ?
The short answer is no.
You can try Blender's Shrinkwrap, but that's not going to produce very clean geometry afterward either. That's kind of wallpapering over your problem anyway, which is that you need to learn to strip model (box modeling is not a good way to produce organic meshes in most cases). The solution to having bad meshes is not to make bad meshes. Nothing you do to fix them is going to produce quality results when they've got issues to begin with.
@Sickle, Im a bit shakey the term 'strip model' seriously though, is that a new / more specific term for poly / edge modeling or is it something else?
hehe, y never heard of 'strip model' either and didn't find anything on Google. SickleYield is right though, it may sound harsh but the answer is to learn how to model correctly and make good topographical meshes in the first place. Remember, good topography starts at home ;) and... when tempted by all of the 'quick' and hackish ways to slap something together, 'just say no.'
To make Blender's 'Shrink Wrap' work, or to nab pieces of various objects and merge them together with any semblance of a decent object when done requires the understanding of what makes a decent mesh in the first place. Automated systems will not make up for that at this time or in the very near future or everyone would be using them. After all, why bother trying to learn how to do something if you can just get out the magic box.
I will say, learning good topo brings rewards, it is worth the effort.
Strip modeling and poly modeling are different words for the same thing. Here's a tutorial for the OP's benefit, although I wouldn't start with a biped face for a first project. I've heard the term used most by other Blender users, but there are a couple of 3ds Max forum posts and tuts calling it that, so I didn't just make it up. ;)
The reason this method is so useful is that if you start by laying strips of polygons in the shape you want, then connecting and filling in between them, you can have complete control over the shape and flow of your polygons and avoid N-gons that Poser and DAZ don't like (you mostly need to have triangles or quads). Further, if you strip-model using larger polys, it's much easier to duplicate, subdivide and sculpt it nondestructively, so that you have a low-poly and a high-poly version (something advanced users love as well as something you can use for texture baking to the low-poly version).
This can also be the difference between a clean, stretchless UV map and no ability to map at all.
I was going to do boolean union on my mesh, but Hexagon maxed out on how many times I could use boolean union, so I couldn't finish doing boolean union.
I mostly build spaceships and war robots. My problem for example, is turning a leg of a robot, in this case a shin, which is composed of individual groups that are welded together to give a unique shape to the shin, into a single mesh that can be UVed easily.
If mesh retopology is not the solution, then can anyone recommend a free program that doesn't max out on boolean unions to the point that I can make a body part like the shin into a single mesh? An example of what I do can be found here:
http://nightmarehero.deviantart.com/gallery/?offset=48#/d515kgn
@Sickle, thanks for the explaination - that makes sense now. I guess then (for the most part) I tend to start with 'boxing' but end up 'stripping'. :)
@TNH, Looking at you link, I cant tell much about the topology, but the model looks ok. From the sounds of it, your modeling technique is fine for this sort of thing. a robot like this is made up of lots of seperate parts.
Im a Max user so I have never tried hexagon, but isnt there a way in that program, to simply attach your meshes together? e.g. group all the meshes that make up the lower leg in to one mesh?
I dont really see the need to join all these parts up by their verts and faces, doing so would probably ruin the model. Since you werent UV mapping each mesh as you went along, you will have to spend a bit of time sorting all those clusters out to build your map, but it shouldnt take that long.
I think what your asking for though is impossible, I have never seen anything on this earth that auto-generates complicated topology with out making a a dogs dinner out of it.
You must have missed the part about 'quick and hackish ways...' as you are attempting to use booleans ;p
Remember, just say no..
As for box modeling being limiting and giving poor results vs starting with a surface patch etc... I would suggest that the problem isn't with starting with a box modeling technique but rather in viewing the technique in a limiting way. However one starts out, loops, edges, cuts, stitches, joins etc.. all come into play at some point so it's more a matter of whatever works most efficiently for any given area imo. Weld a box modeled section onto a lathed section with a patch if it those are each the most efficient for the sections of the model in question.
Ok, I hesitated to give up this link because it is such a great resource imo... but in case anyone is interested the first link covers modeling basics in a very digestible and thorough way using Blender. The same concepts can be used by any modeling program ofc with some adaptation: http://www.youtube.com/playlist?list=PL2BA5BDF79FF50122
He also has lessons in texturing, zBrush, UDK, Unity, etc..
My opinion,
We cant fairly judge his technique without seeing his work properly, I feel that his slightly cavalier attitude is maybe clouding judgment abit.
The picture of the Mech he posted actually looks pretty good.
And for the other points.
IMO You can potentially start off a complex model with either a box or a plane primative, depends on what is it you are doing, sometimes it can even be a case of apples and oranges, as 1 hour later you wont even be able to tell, or even remember which one you started with.
Although Ive not used them for a long time, even booleans have their uses on occassion, in fact, there is nothing automaically wrong with using a boolean by default, they can be fine in some situations - so long as you know what they are about, prepare the topology in advance and then tidy it up afterwards.
Obviously if you use them to punch a tiny round hole in the middle of a single giant face with no sub-divisions, the resulting topology will be nothing short of horrific.
It really depends....
here is model I did 8 years ago for a Silent Hunter 3 mod, all those side windows on the nose and fuslage are booleaned. It was a bit of a frankenstien model with about 10% of it being reworked from an older mesh imported in obj (triangulated areas).
Triangles are not much of problem on lowish poly game models though. I wouldnt dare to have tris on peice of clothing for Genesis.
There may be a better solution - try Meshlab : http://meshlab.sourceforge.net/ It's free :)
There's a feature for removing occluded faces someplace. I saw a tutorial over at www.fractalforums.com for using it to do that very thing when optimizing fractal obj files, which always have tons of junk on the inside).
I'll look for the toot later...
i use wings3d but the concepts should be the same. you might find more help in the wings forum. http://www.renderosity.com/mod/forumpro/showforum.php?forum_id=12445
in wings you connect the 'pieces' into a single mesh by joining a polygon from each piece with the bridge command.
I already use the group function and the weld function in Hexagon to get all the parts in one group. The reason that I want to fix up my models is so that I can sell them online at a place like DAZ with UV mapping already defined competantly. I have been told by the DAZ vendor Valander that I need to make my stuff work without internal mesh data if it is going to be acceptable for sale. Also if I ever sell any of my concepts to an art/movie studio, I'm pretty sure they would demand as much of the mesh.
I am a firm believer in my modeling technique, because I am modeling in a style that generates a symbolic element to my concepts, I use somewhat symbology on the contours and structures of my designs to influence the aesthetics. You can probably see faces in my spaceships like the Warrior Lord; the pig face in the middle of the bottom of the ship is a God of salvation, and the front of the spaceship is made to resemble a duck's face, that is a reference to a warrior tyrant from another planet that became a behemoth monster akin to the hulk, as my back story for that ship indicates.
http://nightmarehero.deviantart.com/gallery/#/d5c4nc3
Likewise if you look at my final design for my war mech and the others in this following picture, you will understand my desire to sell it as Poser merchandise. I was checking the internet online and I found references to voxel sculpting that is like boolean modeling on the fly and certain programs like 3DCoat that do just that, however I was hoping for something in my price range that could actually do something akin to a boolean union on all the mesh faces and combine the mesh without disfiguring it. I was trying a couple of programs last night and I couldn't make some of them work. If anyone has a suggestion on a free program that does something akin to voxel sculpting and takes a grouped obj and combines it to remove internal geometry and leaving the external geometery intact, then please let me know, as I really want to sell my meshes online at a site like this.
http://nightmarehero.deviantart.com/gallery/#/d5dnuv7
I was looking at that program last night and I found it complicated, I'm more of a layman and I didn't know what to do with it. maybe it works for what you say, but I didn't know where to start to clean up my mesh.
reading the advice from the experienced vendors here are giving to you, it sounds to me like they are saying you shouldn't use the Boolean method to create content for selling. just my 2 cents worth.
I'm not opposed to doing it without boolean, but reinventing my modeling strategy which relies on symbolic elements would be detrimental to me at this point. If there's some way to remove internal geometry without boolean when you have meshes grouped/welded together to create one mesh group, I'm all for using it. I'm just hoping it's in my price range and they'res a tutorial behind it.
I'm not opposed to doing it without boolean, but reinventing my modeling strategy which relies on symbolic elements would be detrimental to me at this point.
And there's the bottom line.
They say picture says a thousand words, and the above is good example, now I fully understand your problem.
And - No Im quite sure You cannot boolean your way out of that one. Even if you could, it would be a very messy mesh by the end of it.
If you have low poly version (without mesh smooth) you can manually edit that - otherwise a rebuild would be in order.
I know its not what you want to do, but I cant see any other option Im afraid.
A side note, I'm not sure where you are getting the parts from that you are welding together but hopefully you realize that you can't take them from other meshes and redistribute the composite if any part comes from a model that has a limit on the license that would contradict said form of distribution.
Also, part of the problem with welding parts, whether from different models or from your own is that to get good uv maps, one needs to have good flow to the topology and have somewhat consistent polygon size. Also, things like n-gons, mixes of tris and polys, excessive use of polys in areas where not needed, flow of poly direction where there will be bends, and verts with excessive polys coming together all effect the quality of the final product and therefore it's end value. That was why I put up the Blender link btw, it goes into all of that.
Yep, N-Gons are a no-no in Poser or DS, as they dont display properly, While tris are often considered as bad topology, they can be ok, like on flatish surfaces, and objects with sharp edges, but when they are mixed in with quads on a curved surface they are bad news - since they tend to break up smoothing groups, they dont subdivide well and they can also be a bit of a pain when UV mapping.
thenightmarehero:
I understood your issue from the first post, because I have modeled this way in the past many, many years ago.
It is called "primitive modeling" (for those of you who don't know), and the completed model is composed of morphed/shaped primitives, and grouped together to form final mesh.
I was instructed (by professionals) in this model building technique via books for both RayDream3D, and TrueSpace4.
Looking back at it, it is one of the worst ways (next to Booleans) to model for Poser/DS by today's standard. Sure, it was fine back in 97', and even still o.k. now for personal projects, but if you want to sell commercially, forget about it.
I created robots and spacecraft this way with DOGA as well. This technique has it's merits, but highly undesirable for Poserverse.
There is even one technique/workflow that yields what you want, a shrink-wrapped solid mesh shell in the shape of your underlying primitive group compound parts.
Metaballs...
But, like primitive modeling, is not fit for today's standard in regards to end mesh topology. It is hard for you now, because you are working from after the fact. If say you started with creating those primitives, and grouping it all together in TrueSpace, or Organica (which would be my choice) you would end up with the one solid mesh shell you now desire. Though, some manual retopo might still be in order. This technique, and end result is also stuck in 97'.
I understand you want to continue to build your models in this way, and I agree with you, aesthetically, they are pleasing, and very dynamic. Technically, however, they are a nightmare. You could continue to model this way, but you will just have to learn another modeling skill set to offset, and make corrections.
If it were my model, personally, I would do a little prep work, and then eradicate all the occluded geometry in Metasequoia (my facet/poly/strip modeler of choice: FREE version available).
The prep work is two-fold; First, you need to discern and assign contour lines via edge topology with coincident vertices to line up in each individual primitive to be welded together. This is important, because you need to define where your outer shell begins, and where the occluded geometry to be deleted lies. Second, once you have defined shell geometry you need to select all polys that will belong to the final shell, and assign them a new material (a temp material is fine).
With the prep work finished, I would then select the remaining polys (occluded) via material selection (of original materials), and then hit delete. Viola... No more occluded inside mess to worry about. The only thing left would be to weld all your coincident vertices, and then go about re-selecting/re-assigning your polys to your desired material zones. Now, onto UV mapping, which should be a breeze...
I hope this helps.
If you would like, you could break me off a piece of the model (complete shin setup, for instance) and send it to me, and I will perform the techniques, create screen grabs, and send you back the final so you can follow along.
If not, no worries. This is just how I would correct the mesh.
Or, he could duplicate edges that make up the defining outer edges, delete everything else, and use those edges to recreate the structure with a clean topology. This would be my preferred workflow in most cases.
I'm not saying this is a better method btw, just an alternative.
* I really liked Raydream also. It had a sweet watch tutorial which included the gears and would look fantastic today, 10 years later. Carrara does not seem the same for some reason, can't put my finger on it...
I see what you are saying Gedd.
Yeah, that would work, but he would need to be skilled enough to recreate the contour lines where the primitives meet. Most of the outer edges of each primitive would still reside in occluded space.
If I were to use that technique I would definitely use Hexagon then, and once I have all my extracted edges I would recreate the polys.
I still use RayDream from time to time, but by my count it's 15 years (bought mine alongside Poser3 in 97'). What gets me from RD to Carrara is RayDream has 3D paint, but my Carrara 6 Pro does not. Yet, C7Pro has 3D paint again. I feel cheated everytime I open C6Pro and I want to 3D paint.
I'm wondering if SickleYield and I were basically talking about the same thing re: his comment on strip modeling and my referencing pulling lines from the original mesh to build up a new mesh. There is a technique I've heard referred to as 'spline patch modeling' as it uses splines to define the framework to build up the basic form of a surface, as shown here:
Spline Patch Modeling in Modo, P1
There are also coons patches in Hexagon which I haven't played with yet so can't give a +/- on, but plan on trying out.
I do think building parts up using these and other techniques (lathe, etc..) then welding them together is the most efficient way to model, but the fundamental understanding of what makes a good mesh regardless of creation method is critical in getting good results. Understanding how to create meshes using different methods that once combined look like a well developed continuous mesh that belies the fact it was welded together (along with taking into account flow etc...) is the key imo.
There is another note here, some tutorials show how to 'cut' a mesh and insert polys using different methods to 'retopo' the mesh so that it flows better. Retopo doesn't have to be this complicated thing if done during the modeling, particularly if done when the mesh is still relatively low poly. If one watches many tutorials, one of the key concepts that comes out is that building correct structure starts at the lowest poly stage and continues through the mid poly stage. Anything higher then mid the correct structure should already be established as fixing it at the point of any higher poly count then low mid is a nightmare that often requires an entire mesh retopo using retopo tools, which is very labor intensive and not the preferred method. In fact, at that point it is often easier to rebuild the mesh from scratch using better methods.
If one is to retopo a full mesh, consider 3D Coat as it seems to have some of the best retopo tools at the moment. There are videos at their site showing what is involved. Again, an understanding of proper flow is necessary to use these tools also. There is no shortcut to needing to be able to know good mesh design.
I'm wondering if SickleYield and I were basically talking about the same thing re: his comment on strip modeling and my referencing pulling lines from the original mesh to build up a new mesh. There is a technique I've heard referred to as 'spline patch modeling' as it uses splines to define the framework to build up the basic form of a surface, as shown here:
Spline Patch Modeling in Modo, P1
........
If one is to retopo a full mesh, consider 3D Coat as it seems to have some of the best retopo tools at the moment. There are videos at their site showing what is involved. Again, an understanding of proper flow is necessary to use these tools also. There is no shortcut to needing to be able to know good mesh design.
No, I thought he should toss the original mesh and start over, but then I also haven't done much with spline modeling - it's not Blender's best thing, so I'm not as familiar. ;)
I have heard good things about 3dcoat, for its texture painting as well as its retopology, but I do not own it yet.
I'm wondering whether your symbolism and your modeling are as closely tied as you think.
Symbolism is essentially another design element overlaid onto the basic structure and there are several ways of incorporating that into a design, you can do it on the mega scale, in which the whole design has a certain aesthetic (for instance the common saucer and nacelle design of Star Trek starships), or at the micro-scale in which design elements repeat throughout the structure (c.f. tiling patterns in Mosques, or gothic arches in a cathedral), or there is your system in which, if I understand it correctly, you build the structure by welding smaller symbolic elements together.
Might it be possible to lift your symbolism out of the basic structure, and either incorporate it into the initial design concept at the macro scale, or superimpose it post-build as micro-scale surface detailing (greebles!), or even move it out of the modelling entirely and into texturing?
The pros seem to be saying your current modelling method won't make the grade as a vendor, you say you want to take the step to becoming one, something has to give. Changing your modelling method doesn't necessarily mean dropping symbolism, but it does seem to demand you change how you are currently doing it.
I only use my parts, all the parts I use on my spaceships and robots are my original design.
Well then as long as you don't rely on booleans, you shouldn't be far off base. Sounds more like you just need to expand a bit :)
I recommend you check out those links I posted earlier, it's using Blender but he breaks it down very simply and you can use the concepts in Hex if you prefer.