Smooth Surface Modeling; alternative method

[R Reynolds]

Nothing earth shattering; just a different way to layout patches based on the idea of retopolgy; i.e. using an existing model as a template for a new mesh.

 

The first image shows the original model - the red hood of a 1955 Chevy truck - modeled by Stian and generously lent to me. As you can see he formed the surface in the typical fashion by drawing splines along the major contour lines. But as usual when you have a cylindrical/spherical detail in a rectangular surface you can't avoid generating 3 and/or 5 point patches; which, as you can see in the second image, sometimes are not realistically smooth no matter how much you tweak the bias handles. Believe me I tried.

 

After trying a number of alternate, unsuccessful variations to minimize the 3 and 5 pointers, I realized could likely get the smoothness I wanted if I used only four sided patches. So I made a green copy of the hood in the same location in modeling space as the red original. Next, as shown in the third image, I just did a simple re-arranging and extension of the existing splines to form only four pointers. It was easy to move the new green cp's to lay in the red surface (fourth image). After a little bias tweaking I got the more than acceptable smoothness seen in the fifth image. Granted patch count is a bit higher but it's a small price to pay.

 

Your mileage may vary depending on the desired shape but making a sacrificial template model might be the answer to a particularly vexing surface.

[itsjustme]

Thanks, Rodger!

[Xtaz]

Perfect.... great result !!!

[Rodney]

Sweet! Thanks for the tip Rodger.

 

Such a simple and elegant solution.

 

 

making a sacrificial template model might be the answer

 

Over the years there have been several times where this underlying idea of creating sacrificial templates has worked really well with splines (I recall sacrificial templates being useful in layering materials as well ala Anzovin). Another example is where we might lathe a sphere but by first lathing a cylinder to get the ideal curvature for the half circle we need to lathe the final spherical splines. It's a lot more easy to do than describe...

 

To achieve better/smoother surfaces with more evenly spaced four point patches I can easily see the benefit of using your methodology.

[robcat2075]

Good example , Rodger!

[jakerupert]

hi Rodger,

 

Great find!

 

I guess this is, where some retopologytool like "snap cps to underlying surface"

would come in handy?

 

Isn`t the adjustment of all the biashandles very tedious?

[agep]

Interesting! Is porcelain added to any of the examples? If not, I would love to see how that would turn out

[R Reynolds]

Is porcelain added to any of the examples? If not, I would love to see how that would turn out

The first thing I did when I opened your Chevy flat bed was strip out that material. I've never been satisfied with the effect of averaging normals so I rarely use it. And the attached image just confirms that. As before the green hood uses only four pointers while the red is your original hood with a 3 & 5 pointer plus a hook. The normal weight increase from 0% to 100%, left to right. More averaging certainly improves the red hood but not as much as eliminating the odd patches. Perhaps more surprisingly, more averaging degrades the green hood, IMHO.

 

Isn`t the adjustment of all the bias handles very tedious?

Sometimes almost stupefyingly so. But it's the price we pay for having almost total control over a surface's shape. Sometimes you just have to turn up the background music and get all zen on the process.