I have been working on making a 3d model of a head but I was wondering what I may be doing wrong that it turns out like a klingon or something abnormal. ?
hmm we're not in 2005 anymore - why even bother starting from model ?
nowdays you can use zbrush to sculpt face from a cube, and then do retopology afterwards to get your basemesh ..
It makes the task much, much easier if you start off with a good basemesh. Also, this method ensures you know how to place your edge loops properly for when you do retopologize. Also, starting from a model blueprint, it also ensures that your main facial features are correct.
To to OP:
There are some nice head modelling tutorials on the polycount wiki, including some nice pictures on where and how to properly place your edge loops. Huge help!
Also, I'd recommend you not using a drawing as your model. Use a real-life front and profile picture to do so. Then you can exaggerate the features to make it match the model a bit more. But that's just me.
hmm we're not in 2005 anymore - why even bother starting from model ?
nowdays you can use zbrush to sculpt face from a cube, and then do retopology afterwards ..
ugh... that's like saying "why bother learning to draw on paper with a live model, when you can download some photo's and copy it in photoshop".
Prophecies : that's counter productive at his level - to worry about anatomy and edgeflow at the same time.
I don't think modeling a basemesh from thin air is easier than retopology, and sculpting from a cube is a great way to learn basic facial anatomical proportions.
ugh... that's like saying "why bother learning to draw on paper with a live model, when you can download some photo's and copy it in photoshop".
huh ? not even close. Edit:But I see the reason for the confusion .. I was referring to making a basemesh that way, not the actual head.
I think the method of starting from a basemesh from thin air is very hard and outdated, it's much easier to first get the anatomical proportions off with zbrush and then retopology afterwards. (at which time, you could concentrate on topology, without worrying about fixing the anatomy all way through)
I use the same basemesh for most of heads, but I highly recommend doing the above method rather than downloading a premade basemesh, it teaches both anatomy and correct edgeflow modeling.
The first problem I see is that your reference images could be better. If you're new to modeling heads then I highly recommend modeling from a professional concept or from photos. Having nice orthographic images are great, but you'll rarely find side views that match perfectly with the front view so don't be too surprised if your model seem off even though you think you're matching it to both images. The head is not a perfect box shape so get some good 3/4 views as well and make sure your head model is also conforming to these views.
I also recommend analyzing your model under different lighting situations. Set up some lights similar to how they appear to be in your reference images and see if you can spot the differences in how the shadows are forming. There are a lot of common misconceptions about how the head is structured that can be avoided if you forget about what you think you know and really analyze the forms.
I'd also hold off on adding too much detail until you're more confident in the proportions. Don't get too caught up in trying to get perfect topology. It's best to accept that your first head is just for learning purposes, and focus on conveying the proper shapes. You can always go back and fix the topology later.
Well, your concept isn't the best one in the world to be honest, considering it you're doing pretty good, assuming you'll do tweaks/mesh cleanups later.
I would recommend getting better concept, or photo. And use reference planes just for figuring out basic facial/cranial masses and placing strategic parts, like eyes, mouth and maybe chin/jaw corner. Next thing would be hiding those planes, analyzing actual reference and figuring out form in perspective. Human head is not ak74 - tracing blueprints will only get you so far
i agree with chai, at this stage fighting with edgeloops as opposed to work with form is not the best idea, we have tools these days that let yu forget the oh so technical modelling part and let you concentrate on what counts at this stage the most, how your model looks.
while i agree that a good basemesh can fasten things, i don't agree that the sculpting basemesh should have any defined loops, it needs to be evenly spaced and might have some more detailled areas here and there, having an already defined "final" edge flow can stop you from beeing creative with your forms as you wouldn't want to destroy your holy topology
i understand the thinking behind the "learn how to model a good model before you consider sculpting stuff" method, the thing is, the whole meshflow, topology techy thing can be learned on top of a already well defined sculpt as well, you don't need to start from scratch to learn how a good meshflow works
Cooljay, you need to adjust the proportions and smooth your polygons, but in max, with that mesh it would be a mess/nuisance. If you are modelling a head, using subdivs you can use less polygons and modify the forms/volumes with ease.
You should learn at first some aspects of subdivision modelling. It's not much, and it can give you a lot of advantage using Max tools. Triangles and pentagonos are keys to do nice subdiv models with a few poligons.
And i must say i'm not with Chai at all
We could not be in 2005, but i have one thing clear, the 3D has not changed a lot. I'm still doing the same things as in 2004 but only a bit faster (better tools and and better computers, that's all).
I always start drawing concepts, then i BUILD my models "from thin air" without blueprints (in that way i train my perspective perception), always starting from something solid i know how it is. With experience, you reach a point that topologies are no longer an expensive calculation for your brain (like normal maps for actual videocards ), and not only topologies, but also seams for the mappings and topologies for proper deformation (animation). You end taking into consideration too many things in an automatic way, things that are heavy important in real production.
To be technical is to really know the tools we are using (and not the raw use). You can be more or less technical, but you NEED to have some knowledge. The more technical, the better, you will be more useful and appreciated in any job.
Each person is a world, and having worked from a cube, and later a retopo, i can say that at least for me, that way is counter-productive when you are working with a concept, and why? because you lose time, and because you can read the topology to be builded from the concept (why don't do things well from a very beginning? because we are lazy?).
Topologies are very important and you need to train that aspect very hard. You can sculpt very well, but later, your retopologized mesh can be a mess because you don't have any idea of topologies (with bad smooth and topology, and artifacts, using triangles where they are not supposed to be, etc), so useless at the end. All artists need to be technical in a good %.
I know too many artists that have been commissioned for Zbrush sculpts, and all because they didn't show knowledge on real modelling. One guy did the sculpt, another the subdiv model + mapping, and another the texturing+shading. You must be very very very good at sculpting if you want to be contracted for a sculpt work.
practice is all.
If you don't know what you are doing... to go for the easy way may be a good option (the more artistic way), but i rather prefer to play with a pencil and paper, much faster, and i don't end modelling something i will trash as too many in this forum.
Having a solid concept... and later need to do retopo... err sorry but i think that's not good, you have a broken workflow and you will end wasting more hours/days in something you could have done in less time.
furthermore, meshes from retopologies are very smoothed, and you need to be tweaking a lot in too many cases, so more work at the end.
And, please, don't talk me about game art models and its workflow because that's not the subject of this thread and we all know it's different :shifty:. I'm talking about production subdiv models if someone don't read me well.
blaizer, while i understand and agree your points, i still don't get why you are so agressively against this workflow, you are a great modeler, no questions about that, some of the best models i've seen in the las few years i have in my inspirations folders are from you.
But all you say also applies on the sculpting and then retopo workflow, you can learn the exact same things, you can learn topology from it as much as you can learn it from modelling you just learn it in a different order.
Instead of learning topology and then form you first learn form and then topology and from all my years as a tutor i have to say that sculpting and then retopo is faster to learn and easier to understand, a tool loke zbrush, mudbox or 3dcoat is just easier to learn and when you got that its easier to make the step into another app - but thats just my experience from my students.
when you say meshes are very smoothed when retopologized,i tend to agree but thats something you can tweak with practice as you can learn it the other way around with practice
you have to admit, that every sculpting tools will smooth the basemesh once you subdivide it the first time, the inssue stays the same and you will again have to tweak afterwards if you want it to be the same sharp model
depending on th style you way might apply more or less, but from a beginners point of view i'd say sculpting first applies more, as its as you agree the more artistic way and the technical part just doesn't apply that much yet, the model will be for th trashbin anyways, as every beginners model is, no matter how you create it.
I don't prefer one of the workflows over the others it just is a matter of style and from what i see he wants a pretty organic less clean stylized look and thats definitely a thing you can start easier with a sculpt
I think workflow is an important thing to develop, for consistency and many other reasons. However, while there are several different workflows, I don't think that you need to stick to a certain one yet. If you want to make a base mesh in Max, then take it into Zbrush for sculpting, that would be equally fine to creating the whole head from start to finish in Max. You should try all the different methods (box model, plane edge extrusion, spline, etc.) and see which works best for you.
The topology of this head isn't good, and while I agree that Cooljay shouldn't be worrying about topology and anatomy and everything else at the same time, it doesn't hurt to practice them. As I stated in the duplicate thread, Cj needs better reference. If he follows front/side refs, the basic anatomical problems should be solved. And for topology, I think he should at least put in the loops around the eyes and mouth, because those are at least the very basic ones.
It would help to study other head models' edge loops:
I'm not saying anyone's suggested workflow is wrong (in fact they're all good ), all I'm trying to tell Cooljay is that he needs to find what workflow works best for him, or at least try all the various ones before he decides which he likes best. Anyway, we don't know for certain that he has Zbrush.
It takes a long time to get even a decent head. Just be patient and keep practicing.
Replies
nowdays you can use zbrush to sculpt face from a cube, and then do retopology afterwards to get your basemesh ..
To to OP:
There are some nice head modelling tutorials on the polycount wiki, including some nice pictures on where and how to properly place your edge loops. Huge help!
Also, I'd recommend you not using a drawing as your model. Use a real-life front and profile picture to do so. Then you can exaggerate the features to make it match the model a bit more. But that's just me.
Hope this helped a little!
ugh... that's like saying "why bother learning to draw on paper with a live model, when you can download some photo's and copy it in photoshop".
I don't think modeling a basemesh from thin air is easier than retopology, and sculpting from a cube is a great way to learn basic facial anatomical proportions.
huh ? not even close.
Edit:But I see the reason for the confusion .. I was referring to making a basemesh that way, not the actual head.
I think the method of starting from a basemesh from thin air is very hard and outdated, it's much easier to first get the anatomical proportions off with zbrush and then retopology afterwards. (at which time, you could concentrate on topology, without worrying about fixing the anatomy all way through)
I use the same basemesh for most of heads, but I highly recommend doing the above method rather than downloading a premade basemesh, it teaches both anatomy and correct edgeflow modeling.
I also recommend analyzing your model under different lighting situations. Set up some lights similar to how they appear to be in your reference images and see if you can spot the differences in how the shadows are forming. There are a lot of common misconceptions about how the head is structured that can be avoided if you forget about what you think you know and really analyze the forms.
I'd also hold off on adding too much detail until you're more confident in the proportions. Don't get too caught up in trying to get perfect topology. It's best to accept that your first head is just for learning purposes, and focus on conveying the proper shapes. You can always go back and fix the topology later.
I would recommend getting better concept, or photo. And use reference planes just for figuring out basic facial/cranial masses and placing strategic parts, like eyes, mouth and maybe chin/jaw corner. Next thing would be hiding those planes, analyzing actual reference and figuring out form in perspective. Human head is not ak74 - tracing blueprints will only get you so far
cheers
while i agree that a good basemesh can fasten things, i don't agree that the sculpting basemesh should have any defined loops, it needs to be evenly spaced and might have some more detailled areas here and there, having an already defined "final" edge flow can stop you from beeing creative with your forms as you wouldn't want to destroy your holy topology
i understand the thinking behind the "learn how to model a good model before you consider sculpting stuff" method, the thing is, the whole meshflow, topology techy thing can be learned on top of a already well defined sculpt as well, you don't need to start from scratch to learn how a good meshflow works
You should learn at first some aspects of subdivision modelling. It's not much, and it can give you a lot of advantage using Max tools. Triangles and pentagonos are keys to do nice subdiv models with a few poligons.
And i must say i'm not with Chai at all
We could not be in 2005, but i have one thing clear, the 3D has not changed a lot. I'm still doing the same things as in 2004 but only a bit faster (better tools and and better computers, that's all).
I always start drawing concepts, then i BUILD my models "from thin air" without blueprints (in that way i train my perspective perception), always starting from something solid i know how it is. With experience, you reach a point that topologies are no longer an expensive calculation for your brain (like normal maps for actual videocards
To be technical is to really know the tools we are using (and not the raw use). You can be more or less technical, but you NEED to have some knowledge. The more technical, the better, you will be more useful and appreciated in any job.
Each person is a world, and having worked from a cube, and later a retopo, i can say that at least for me, that way is counter-productive when you are working with a concept, and why? because you lose time, and because you can read the topology to be builded from the concept (why don't do things well from a very beginning? because we are lazy?).
Topologies are very important and you need to train that aspect very hard. You can sculpt very well, but later, your retopologized mesh can be a mess because you don't have any idea of topologies (with bad smooth and topology, and artifacts, using triangles where they are not supposed to be, etc), so useless at the end. All artists need to be technical in a good %.
I know too many artists that have been commissioned for Zbrush sculpts, and all because they didn't show knowledge on real modelling. One guy did the sculpt, another the subdiv model + mapping, and another the texturing+shading. You must be very very very good at sculpting if you want to be contracted for a sculpt work.
practice is all.
If you don't know what you are doing... to go for the easy way may be a good option (the more artistic way), but i rather prefer to play with a pencil and paper, much faster, and i don't end modelling something i will trash as too many in this forum.
Having a solid concept... and later need to do retopo... err sorry but i think that's not good, you have a broken workflow and you will end wasting more hours/days in something you could have done in less time.
furthermore, meshes from retopologies are very smoothed, and you need to be tweaking a lot in too many cases, so more work at the end.
And, please, don't talk me about game art models and its workflow because that's not the subject of this thread and we all know it's different :shifty:. I'm talking about production subdiv models if someone don't read me well.
concept>model>mapping>sculpting>texturing>rigging>lighting>render.
My two cents
But all you say also applies on the sculpting and then retopo workflow, you can learn the exact same things, you can learn topology from it as much as you can learn it from modelling you just learn it in a different order.
Instead of learning topology and then form you first learn form and then topology and from all my years as a tutor i have to say that sculpting and then retopo is faster to learn and easier to understand, a tool loke zbrush, mudbox or 3dcoat is just easier to learn and when you got that its easier to make the step into another app - but thats just my experience from my students.
when you say meshes are very smoothed when retopologized,i tend to agree but thats something you can tweak with practice as you can learn it the other way around with practice
when you say the workflow should be as follows:
concept>model>mapping>sculpting>texturing>rigging> lighting>render.
you have to admit, that every sculpting tools will smooth the basemesh once you subdivide it the first time, the inssue stays the same and you will again have to tweak afterwards if you want it to be the same sharp model
depending on th style you way might apply more or less, but from a beginners point of view i'd say sculpting first applies more, as its as you agree the more artistic way and the technical part just doesn't apply that much yet, the model will be for th trashbin anyways, as every beginners model is, no matter how you create it.
I don't prefer one of the workflows over the others it just is a matter of style and from what i see he wants a pretty organic less clean stylized look and thats definitely a thing you can start easier with a sculpt
I think workflow is an important thing to develop, for consistency and many other reasons. However, while there are several different workflows, I don't think that you need to stick to a certain one yet. If you want to make a base mesh in Max, then take it into Zbrush for sculpting, that would be equally fine to creating the whole head from start to finish in Max. You should try all the different methods (box model, plane edge extrusion, spline, etc.) and see which works best for you.
The topology of this head isn't good, and while I agree that Cooljay shouldn't be worrying about topology and anatomy and everything else at the same time, it doesn't hurt to practice them. As I stated in the duplicate thread, Cj needs better reference. If he follows front/side refs, the basic anatomical problems should be solved. And for topology, I think he should at least put in the loops around the eyes and mouth, because those are at least the very basic ones.
It would help to study other head models' edge loops:
http://www.subdivisionmodeling.com/forums/showthread.php?t=8911
http://img254.imageshack.us/img254/874/polyregionsif6.jpg
http://img222.imageshack.us/img222/5171/newhead2hp2.jpg
I'm not saying anyone's suggested workflow is wrong (in fact they're all good
It takes a long time to get even a decent head. Just be patient and keep practicing.