Jump to content

Help with hard-surface(d) poly model


Go to solution Solved by Cerbera,

Recommended Posts

Hi,

I am working on a part for a pair of headphones which I have managed to model and I am 95% happy with the results when rendered. Something is bugging me about it though and I feel the topology and edge flow could be better. 

 

The main issue is with control edges, where to get nice crisp edges the loops have to go through the entire model where it is not necessary and this creates ugly dense areas of geometry. 

 

Obviously this is a ‘nice to have’ and I am happy with the way it renders but I was wondering if anyone out there had any tips for me to be able to get this model a bit cleaner? I am completely open to starting the model again from scratch if perhaps I have approached this in the wrong way. 

 

One thought I had was to create the model again at very low poly and then bevel all the edges?

 

Screen shots and C4D are attached if anyone wants to look.

 

Thanks

 

Hard_Surface_Part.c4d

Screenshot 2020-11-17 18_0000_Screenshot 2020-11-17 18.30.36.jpg

Screenshot 2020-11-17 18_0001_Screenshot 2020-11-17 18.30.21.jpg

Screenshot 2020-11-17 18_0002_Screenshot 2020-11-17 18.30.18.jpg

Screenshot 2020-11-17 18_0006_Screenshot 2020-11-17 18.29.42.jpg

Screenshot 2020-11-17 18_0005_Screenshot 2020-11-17 18.29.46.jpg

Screenshot 2020-11-17 18_0004_Screenshot 2020-11-17 18.29.59.jpg

Screenshot 2020-11-17 18_0003_Screenshot 2020-11-17 18.30.03.jpg

Link to comment

Technically this is not a 'hard surface model' because it uses SDS and control loops !  Rather it is an SDS model of a form that has hard surfaces, which is different. We'll come back to that point. Before I start with the criticism there is a 'well done' due, to some extent. This is not an awful mesh by any stretch, and I can tell you have absorbed most of the poly modelling rules, and thought about flow and quad diligence to some degree, but it could definitely be better the 2nd go round...

 

So, accepting that you are going to build this with SDS, there are not quite enough control loops to fully support the geometry, particularly at some of the borders. There's also some rather odd decisions going on with contra flow edges but you seem to be largely getting away with it in terms of renderable surfacing. However, there are a few places where your borders are somewhat baggy and undefined-looking because you haven't followed the rule about reinforcing border edges on both sides, and not allowing changes in edge flow or point convergence to happen right on them where that is the case. In several places your geo only has one control loop where 2 are necessary, and that's what gives you the ragged edges and slightly strained looking geo, although the negative effects are somewhat mitigated here by the polygon density. Also, there are several places where you have dissolved a poly centreline to make a quad but then not slid the points out to make kite quads, thereby giving you degenerated polys that are interpreted by the Catmull Clark solver as triangles, even though they are technically quads*.

 

Here's one area that needs the additional borders, similar to what I have crudely drawn over the top with the doodle tool.... the arrows show points that need sliding out a little to avoid the aforementioned polygon convexity / internal angles of more than 179 degrees.

 

image.thumb.png.145e353ebd7f0b6a60e8d8b8ade21c75.png

 

Another one of those around the front where again our main form only has 1 out of 2 control loops.

 

image.thumb.png.e7f93243a9453d58d7d2d5483faf59f6.png

 

and here....more under-enforced borders...

 

image.thumb.png.76e74335692eecc0f4f38bb21f175149.png

 

The red cross shows where it is not a good idea to suddenly divert an edge at 45 degrees to the flow, right on a curve !

 

Other than that there is nothing catastrophically wrong with this, it just could be a bit better if you spent a bit more time on it and spent longer thinking about edge flow and where exactly you need reinforced border edges... likewise, as you noted, you have a lot of extraneous loops in other places, that could be solved down to simpler forms. In fact I would say it is that lack of solving and downstepping that makes the difference between a Grade A mesh, and this one, which is a solid B- Grade as it stands now.

 

You could probably learn a lot from building it again focusing on tactical shared loops where you can, and making sure you have equally weighted control loops either side of any border on the areas I mentioned. Then you would learn even more by building it a 3rd time as a proper hard surface model and seeing how much time you saved, and could compare the results with the SDS equivalent. I would bet that the HS version would have slightly nicer surfacing.

 

If I had more time I could start the debate about whether or not this should have been an SDS model in the first place, and indeed I see you have considered this ! 😉 There are arguments on both sides, but IMO rather more factors on the side of suggesting SDS is not the way to go here and that actually it would be a far more efficient and speedy process to build the mesh more simply and use true hard-surface techniques like procedural edge bevels and no subdivision at all. This would have negated the need for ANY control loops, and would have meant you could tolerate tris without having to solve them.

 

At various points in our modellings we will all come across exactly that dilemma as to whether to make something SDS or not, and indeed my own natural propensity is to go the SDS way even in some situations where others wouldn't, but in recent years I have become better at making the right decision in advance and my modelling efficiency has improved very much so as a result.

 

CBR

 

*and on a deeper level even quads are technically ALL seen as triangles, but that is unimportant in the context of my point.

Link to comment
  • Cerbera changed the title to Help with hard-surface(d) poly model
  • Solution

Here is your front part mesh with only the necessary loops for HS remaining. I have also evened out the cylinder loops, and fixed a lot of the other issues I mentioned above. I have also applied a bevel (Chamfer, 3 segments, uniform). I've also turned some loops back on themselves after they have left the circular bit to cut down the amount of loops going all around the model, and done that with 100% quads, although I didn't have to because a) this isn't using SDS, and b) I have correctly chosen to do that down-stepping on a totally planar area.

 

image.thumb.png.bc601c001ecbc0786ee26ce887e1eb82.png

 

image.thumb.png.51883e83133198e5084489b4d7ac8432.png

 

 

image.thumb.png.261158d8f3e951845353f84c22194f14.png

 

image.thumb.png.e844130ee398910dc547f24ab9a61c8a.png

 

Apart from the faceting we can see on some of curves (which is only there because this was designed with not enough segments to do without SDS) the actual surface quality and edge bevelling is actually slightly nicer than we would get under SDS and by breaking phong rounding we can get superbly flat looking hard surfaces, without the inherant 'puffiness' that is sometimes evident with SDS, even when there are enough control loops ! And now we have the added advantage that we can change our edge bevels at any time, and they are giving us subdivision only where we need it. And that makes the argument for HS here, and I remain fairly certain it wins the day in this situation.

 

CBR

Link to comment
  • 1 month later...

Hi,

 

Sorry for the late reply, I have finally had the time to look at this and yes you are right I definitely get a better result by doing this as a hard-surface model. @Cerbera thanks again for your expert advice! This is an interesting take on modelling as being fairly new to C4D I would by default go to SDS modelling so it's good to understand there are certain situations where this may not actually be the best tool.

 

I have attached the working file in case you are interested here: Headphone_Part.c4d

 

Thanks

 

Link to comment
×
×
  • Create New...

Copyright Core 4D © 2023 Powered by Invision Community