What have I done wrong? - Quads vs. Triangles vs. Beveling
polycounter lvl 10
Good morning polycounters,
I was recently working on a 3D car model, when I received some feedback regarding my modeling technique. I love getting criticism that helps me improve my modeling, and I've been racking my brain trying to get my head around the differences, google-ing topics, reading forums, checking tutorials and just experimenting with different techniques. Now I have reached a point where I'd like to see if anyone can explain things better than I'm trying to explain to myself.
The images below are of a car hood. I've included the original triangulated CAD model that was provided, the "correct mesh" (which was greatfully provided to me so that I could learn what I did wrong), and the "wrong mesh," which is the rejected mesh that I originally provided.
The correct mesh looks great in the viewport, it's smooth, has visible chamfers/bevels, and was the mesh that was approved by my boss. The wrong mesh, which I submitted was stated to have bad curves (especially around the middle of the hood), bad chamfering and I assumed that meant my Bevels also.
Here are the two together without wireframe:
Maybe it's a bit hard to tell from the above image, but basically without the wireframe, the two models looked the same to me, EXCEPT of course, that the meshes are different.
Where I'm getting stuck most, is that on the correct mesh, there are a lot of triangles, and 6-7 point poles, which I had originally always thought was a bad thing. I think there's a lot of broken edge flows too, but as I'm still a learner, I completely except that I am missing something. On my own mesh, I've made sure the entire model is made up of quads, constant edge flow, and when I add divisions, the polycount is still less than the correct mesh provided.
Closer look below:
I've been receiving feedback from an industry veteran who's been working for over 10 years, and as I've only been working for 6 months, I am fully aware that I just don't have the experience and knowledge that other people do. So if anyone can help me understand, I would greatly greatly appreciate it.
Thank you in advance!
I was recently working on a 3D car model, when I received some feedback regarding my modeling technique. I love getting criticism that helps me improve my modeling, and I've been racking my brain trying to get my head around the differences, google-ing topics, reading forums, checking tutorials and just experimenting with different techniques. Now I have reached a point where I'd like to see if anyone can explain things better than I'm trying to explain to myself.
The images below are of a car hood. I've included the original triangulated CAD model that was provided, the "correct mesh" (which was greatfully provided to me so that I could learn what I did wrong), and the "wrong mesh," which is the rejected mesh that I originally provided.
The correct mesh looks great in the viewport, it's smooth, has visible chamfers/bevels, and was the mesh that was approved by my boss. The wrong mesh, which I submitted was stated to have bad curves (especially around the middle of the hood), bad chamfering and I assumed that meant my Bevels also.
Here are the two together without wireframe:
Maybe it's a bit hard to tell from the above image, but basically without the wireframe, the two models looked the same to me, EXCEPT of course, that the meshes are different.
Where I'm getting stuck most, is that on the correct mesh, there are a lot of triangles, and 6-7 point poles, which I had originally always thought was a bad thing. I think there's a lot of broken edge flows too, but as I'm still a learner, I completely except that I am missing something. On my own mesh, I've made sure the entire model is made up of quads, constant edge flow, and when I add divisions, the polycount is still less than the correct mesh provided.
Closer look below:
I've been receiving feedback from an industry veteran who's been working for over 10 years, and as I've only been working for 6 months, I am fully aware that I just don't have the experience and knowledge that other people do. So if anyone can help me understand, I would greatly greatly appreciate it.
Thank you in advance!
Replies
I'm not all that familiar with current-gen modeling techniques but here are my guesses:
-triangulation is used to easily vary the amount of detail the mesh offers, so they can round the corners out without making the whole area denser.
-they try to keep the lines as straight and technical as possible because of highlights. In your version the highlights close to the corners would have a kink in the highlight.
-like said before: meshsmooth
-the points at the front of the hood have a weird diagonal crease, because of your adherance to complete loops, while their softly changes.
-looking at the meshes themselves you can see that the original has the same style topology as the red version ad your version is rather different.
P.S. - change your material to a very glossy shader to see the differences better.
http://wiki.polycount.com/SubdivisionSurfaceModeling
Also, Snader's spot-on about the material (etc. too!)... get a specular on there to help you see what's up.
Snader, I changed the material, and left the normals all hard, and saw a lot of what you meant (images attached). It's especially obvious on model #2.
For model #3 & #4, I've adjusted the normals to try and mimic the correct mesh, with mesh #4 having more polygons.
I think testing all my mistakes is making more sense.
Eric, thanks again for the link. I had watched it a while ago, but I think some things don't make sense until I'm actually doing it, so I'm going to go back and watch it a few more times to get my head around it.
I think I'm just struggling to get my head around the use of triangles and the "stars" (I think that's what they're called) present on the surface. Ok for hard surface, bad for organic?
I'll keep gleaning the wiki! ^_^
Thanks again guys!
You need to bring your edge loops closer together to make those ridges on the model. Your very first image shows this clearly... the middle mesh has tight edgeloops, while the bottom mesh does not.
Thanks again!