SUbD and quad workflow advices?

thomthom

Marcello's video tutorial is well worth the time watching it! It's a really good introduction to the concepts of working with quads and subdivisions - best of all within SketchUp!

But remember that quads and subdivision challenges is universal - so remember to not narrow your searches to SketchUp or SUbD when looking for info.

There are some links on the Quad page of SUbD:
http://evilsoftwareempire.com/subd/quads

And as pilou suggested, there are lots of really good info on the subject on other sites that serve other modelling packages.

The OpenSubdiv project also has their own set of tips:
http://graphics.pixar.com/opensubdiv/docs/mod_notes.html

I normally search for "topology quads" and absorb whatever comes up there.

thomthom

@cotty said:

Here's Richs great example as V8...

And here's a screenshot of Cotty's re-save of Rich's example model:

derei

Thank you all!

Here is my result! ... is a combination of all information I got. but i like the result...is clean.

plate corner.jpg
plate corner_sub.jpg

nlipovac

Marcello, WOOOW.... great tutorial, and a great service to plugin.

panixia

thanks! glad to ear that was appreciated and helpful!

nlipovac

I tried to apply Marcello tut on this ugly chair but failed on quad for texturing with Fredo tools.
Can someone see where did I go wrong and hoe to make model better for SubD.
I do not know how to add a pic, or upload a file... so I am posting Google drive links.
https://drive.google.com/file/d/0B3tzIJ59An6WT2tiOTE2Ry1HRWs/view?usp=sharing
https://drive.google.com/file/d/0B3tzIJ59An6WS1NhV2FuMS0wdU0/view?usp=sharing

panixia

fredo paints fails if you try to add the material with quad method on a mesh with poles.
you should group pieces of the mesh so that they have not poles and then run trupaint on each single piece as i quickly did here.
then you can explode all of that again and select as the "face method" the one for "all adiacent WITH SAME MATERIAL AND UV MODE" to fine tune it.
sort of "poor man unwrap".

you also applyied the material on the limit mesh, wich is pointless, because when you toggle off sub-d, mapping will be lost.

LAST AND MORE IMPORTANT: please next time share with sub-d toggled off, no need to share a 35mb file when you can share a (less than) 1mb file

CHAIR FIXED.skp

nlipovac

@panixia said:

fredo paints fails if you try to add the material with quad method on a mesh with poles.
you should group pieces of the mesh so that they have not poles and then run trupaint on each single piece as i quickly did here.
then you can explode all of that again and select as the "face method" the one for "all adiacent WITH SAME MATERIAL AND UV MODE" to fine tune it.
sort of "poor man unwrap".

you also applyied the material on the limit mesh, wich is pointless, because when you toggle off sub-d, mapping will be lost.

LAST AND MORE IMPORTANT: please next time share with sub-d toggled off, no need to share a 35mb file when you can share a (less than) 1mb file

Thank You for advice and good will for tutorials.
I do usually make cuts in model prior texturing, I use SketchUV and Roadkill (kinda poor man uv[sup:2dqh4gfk]2[/sup:2dqh4gfk])...
I just hoped that quad modelling would end those stretch or compressed parts

panixia

@nlipovac said:

I do usually make cuts in model prior texturing, I use SketchUV and Roadkill (kinda poor man uv[sup:24ao8030]2[/sup:24ao8030])...

i can hear you! luckily enough i often work for rich people running zbrush in their office so i can benefit of zbrush wich is kinda rich man unwrap[sup:24ao8030]32[/sup:24ao8030] that software is amazin my second favourite modeler ever after sketchup

@nlipovac said:

I just hoped that quad modelling would end those stretch or compressed parts

i'm pretty confident in thomthom he's the man.. he is playing around with opensubdiv.. who can say what's coming next?

@thomthom btw i noticed some times ago the same control meshes in 3dsmax 2014 (regular catmull clark no opensubdiv)
smoothed by turbosmooth modifier, does't strech narrow control loops' UVs as sub-d does, any ideas
could be helpful for you to dig into it, if next time i try to export some textured quad.obj from max?

thomthom

@panixia said:

@thomthom btw i noticed some times ago the same control meshes in 3dsmax 2014 (regular catmull clark no opensubdiv)
smoothed by turbosmooth modifier, does't strech narrow control loops' UVs as sub-d does, any ideas
could be helpful for you to dig into it, if next time i try to export some textured quad.obj from max?

The next big goal for SUbD is to implement OpenSubdiv. That will be the default mode and I will phase out the legacy mode. This means that the features of subdivision will be dictated by the OpenSubdiv project. Already it features much more than my own version in SUbD. And given how it's becoming the industry standard for subdivisions I would expect it to provide the best set of features.

As for the UV stretching, SUbD currently do a linear interpolation, while OpenSubdiv offer other options. (See my screenshot from an older Pixar research paper here: http://sketchucation.com/forums/viewtopic.php?f=397%26amp;t=63826%26amp;start=60#p585774)
So - while I've get to hook up the various UV options and experiment - I'm pretty sure there are options there that will improve things.

panixia

can't wait

xiombarg

I've noticed that doing extrusions using Joint Push Pull and it's affiliated extrusion tools often creates weird geometry and artifacts when converting to SUBD, it doesn't seem to help by using Quadface tools and trying to convert the mesh. Any solution to this?

Here you can see that the top row is normal, but where it was extruded by Joint Push Pull it gets those weird triangulations on the bottom few rows.

xiombarg

I think I was wrong in thinking it was caused by Joint Push Pull, or at least it's not directly caused by Joint Push Pull. It seems that when I extrude and if I accidentally have the "Borders" settings so that there are "individual borders for each face" and then go in and erase these after the fact, this is what causes problems. Or if the "Finishing" setting is set to "Thicken" and leaves the original extruded face inside the model, you can't just go in and erase the interior edges and faces and still have the geometry turn out correctly when Subdividing.

I even tested erasing the faces and keeping the edges to try and use Quad Face Tools "Convert wireframe to quades" and that doesn't work.
Cleanup doesn't make any difference either.

So it would seem the only answer is to be sure and do your extrusions with the proper setting the first time, or in my case I used the pencil tool to redraw the wireframe of just the bad portions of the geometry and then used quadface tools to covert the wireframe to mesh. And that worked fine.

thomthom

Do you have a sample model so we can see what is going on?

xiombarg

I fixed the problems on the original model, but I was able to recreate the issue.
It's not necessarily anything of a serious problem, but it does create unnecessary geometry when subdivided, yet still smoothed.

The odd geometry only appears on portions of the model.

example.skp

thomthom

Some of the diagonals have cast-shadows enabled:

If you have only soft+smooth diagonals (like from Sandbox Tools) then use the Sandbox Tools Quads to QuadFace Tools Quads function in QFT. That will ensure they are treated as quads and not separate triangles.

I got plans for further tools to make it easier to detect this.

xiombarg

Hmmm, it seems to do it somewhat randomly. I have tried doing push pulls with different options selected, and sometimes it will have cast shadows checked on the edges and sometimes it doesn't. The only consistency I could see is that cast shadows become selected the more the model has been monkeyed around with which I'm sure means something specifically is causing it, but I can't pinpoint what it is. I thought it would always have something to do with the options selected on a push pull, but that only generates cast shadows on edges sometimes. Odd?

pbacot

Wouldn't JPP produce more softened edges? And why would these be expected to follow quad rules? I may not be following the method correctly as others clearly know more about this... just thought I'd offer this.