Re: small problem that we can solve

[Joe Crawford]

> ... if there's some small problem that we can solve in the short term
> for you, let's tackle it so you're finally using K-3D on a regular
> basis, even if it's just using the pipeline to automate some process
> with a script.  BTW, this is why I'm putting some effort into a
> render-engine for Daniel, so he can start generating some useful results.

Ok.  I can definitely go around and ask artists about what tool they
are missing, or if there's a particular problem k3d could solve for
them. Something that would get them using K3D, even if only for a
small part of the process.

Also, from my own experience, here's a thought relating to that. The
idea is fairly small in scope and UI, but the technical coding
challenge might be too much.

A while back there was a lot of talk on the list about quad remeshing.
 In the games and film industry, retopology tools are a *huge* need
right now, and most of them aren't very good.  Zbrush, 3D coat, and
topogun are all tools that help with it, but non of them work very
well.  Also, they have no kind of intelligence or automation.

So if we made the quad remeshing work well, and gave it a good UI,
we'd have something right away that people would want to use. (I'd use
it constantly... lol, and so would basically every high poly
modeler/sculptor.)    It could get lots of people immediately using
k3d, even if all they did was import a model, change it, and export
the model.  It would at least get the software into the pipelines.

I work a lot with artists who follow the common process of:

Model Low Poly >  Sculpt Highpoly > Retologize so it rigs well

This is pretty much standard at every game studio now, and well as in
most films featuring heavy CG work that isn't just backgrounds.

Some artists/studios even skip the low poly modeling part.  In fact in
some situations, people use scanned models, or voxel based models,
which means they start with the worst polygon triangle mesh to start
with.

Tim, what are your thoughts on this?  Do you know much about the very
math heavy ideas behind quad remeshing?  I know a lot about the
theory, but not at all how to code it. (As usual...  lol).


Anyway, the tool would have to:

- Import and understand multi-resolution geometry somehow  (see Footnote 1)
- Allow the artist to draw some key lines on the surface of that
model, to indicate important edge flows (see Footnote 2)
- Based on the high res model, generate a new model with good edge
flow, for which the highest subdivision level matches the original
surfaces.

Footnote 1:  Perhaps by importing muliple versions of the model at
different levels, since a standard format for multi res doesn't really
exist.  This might also be a place where k3d could help form a
standard.

Footnote 2: these could be nurbs curves, polygons, or even just
texture painting... anything that K3D could see to understand where
the artist wanted edges.  Its just a network of curves that the user
can draw and move around)


This might all seem kind of hard to understand.  I'll also see if I
can dig up some videos about this stuff.  I'm sure Blender has some.
Then, depending on how much interest there was, I could make a
screencast to explain it.  By the way, you can see a lot of the stuff
I'm talking about in Blender, though it has no real form of remeshing,
other than doing it by hand.


I don't know what the current state of our remeshing is.  I've never
been able to make it work. Also I don't think it accepts much in the
form of guidance.

Thoughts?


Sincerely,

Joe Crawford
___________________________________
Owner - Celestine Studios   and   Joetainment Enterprises
Cell: 604-866-3050
Email: joetainment@...
Web: http://celestinestudios.com

------------------------------------------------------------------------------
Crystal Reports - New Free Runtime and 30 Day Trial
Check out the new simplified licensing option that enables unlimited
royalty-free distribution of the report engine for externally facing
server and web deployment.
http://p.sf.net/sfu/businessobjects
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

[Daniel Scott Matthews]

On Wed, Jun 17, 2009 at 4:51 PM, Joe Crawford<joetainment@...> wrote:

>> ... if there's some small problem that we can solve in the short term
>> for you, let's tackle it so you're finally using K-3D on a regular
>> basis, even if it's just using the pipeline to automate some process
>> with a script.  BTW, this is why I'm putting some effort into a
>> render-engine for Daniel, so he can start generating some useful results.
>
> Ok.  I can definitely go around and ask artists about what tool they
> are missing, or if there's a particular problem k3d could solve for
> them. Something that would get them using K3D, even if only for a
> small part of the process.
>
> Also, from my own experience, here's a thought relating to that. The
> idea is fairly small in scope and UI, but the technical coding
> challenge might be too much.
>
> A while back there was a lot of talk on the list about quad remeshing.
>  In the games and film industry, retopology tools are a *huge* need
> right now, and most of them aren't very good.  Zbrush, 3D coat, and
> topogun are all tools that help with it, but non of them work very
> well.  Also, they have no kind of intelligence or automation.
>
> So if we made the quad remeshing work well, and gave it a good UI,
> we'd have something right away that people would want to use. (I'd use
> it constantly... lol, and so would basically every high poly
> modeler/sculptor.)    It could get lots of people immediately using
> k3d, even if all they did was import a model, change it, and export
> the model.  It would at least get the software into the pipelines.
>
> I work a lot with artists who follow the common process of:
>
> Model Low Poly >  Sculpt Highpoly > Retologize so it rigs well
>
> This is pretty much standard at every game studio now, and well as in
> most films featuring heavy CG work that isn't just backgrounds.
>
> Some artists/studios even skip the low poly modeling part.  In fact in
> some situations, people use scanned models, or voxel based models,
> which means they start with the worst polygon triangle mesh to start
> with.
>
> Tim, what are your thoughts on this?  Do you know much about the very
> math heavy ideas behind quad remeshing?  I know a lot about the
> theory, but not at all how to code it. (As usual...  lol).
>
>
> Anyway, the tool would have to:
>
> - Import and understand multi-resolution geometry somehow  (see Footnote 1)
> - Allow the artist to draw some key lines on the surface of that
> model, to indicate important edge flows (see Footnote 2)
> - Based on the high res model, generate a new model with good edge
> flow, for which the highest subdivision level matches the original
> surfaces.
>
> Footnote 1:  Perhaps by importing muliple versions of the model at
> different levels, since a standard format for multi res doesn't really
> exist.  This might also be a place where k3d could help form a
> standard.
>
> Footnote 2: these could be nurbs curves, polygons, or even just
> texture painting... anything that K3D could see to understand where
> the artist wanted edges.  Its just a network of curves that the user
> can draw and move around)
>
>
> This might all seem kind of hard to understand.  I'll also see if I
> can dig up some videos about this stuff.  I'm sure Blender has some.
> Then, depending on how much interest there was, I could make a
> screencast to explain it.  By the way, you can see a lot of the stuff
> I'm talking about in Blender, though it has no real form of remeshing,
> other than doing it by hand.
>
>
> I don't know what the current state of our remeshing is.  I've never
> been able to make it work. Also I don't think it accepts much in the
> form of guidance.
>
> Thoughts?
>

LOD (level of detail) is part of the Renderman standard and we can
implement something like it fairly simply by having special check
point nodes in the pipeline that basically say "On level (x) stop here
and render what you have got." By render I mean GL or External

e.g.

[Simple Mesh]  -> [Some Operation Sequence] -> [LOD (1)] -> [Some
Operation Sequence] -> [LOD (2)] -> [Some Operation Sequence] -> [LOD
(3)]  etc.


The [LOD (x)] can also have additional conditional logic that
activates it only under given conditions, including a calculation such
as the distance from the current camera to the nearest edge of a
bounding shape.

I'd like to have a tool to let me displace a mesh with a bitmap where
new faces are added according to the tightness of the resulting local
curvature of the mesh. This Displace by BitMap operation can be
repeated multiple times. When I add renderman displacement to this I
get three or more levels of detail very easily. If I can then use the
GIMP API to paint these LOD displacement Maps (which each can have
different sizes and amounts of repeat) in a live 3D view I would then
have a great sculpting tool.

------------------------------------------------------------------------------
Crystal Reports - New Free Runtime and 30 Day Trial
Check out the new simplified licensing option that enables unlimited
royalty-free distribution of the report engine for externally facing
server and web deployment.
http://p.sf.net/sfu/businessobjects
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

[Timothy M. Shead]

Joe Crawford wrote:
>> ... if there's some small problem that we can solve in the short term
>> for you, let's tackle it so you're finally using K-3D on a regular
>> basis, even if it's just using the pipeline to automate some process
>> with a script.  BTW, this is why I'm putting some effort into a
>> render-engine for Daniel, so he can start generating some useful results.

> So if we made the quad remeshing work well, and gave it a good UI,
> we'd have something right away that people would want to use. (I'd use
> it constantly... lol, and so would basically every high poly
> modeler/sculptor.)    It could get lots of people immediately using
> k3d, even if all they did was import a model, change it, and export
> the model.  It would at least get the software into the pipelines.

> I don't know what the current state of our remeshing is.  I've never
> been able to make it work. Also I don't think it accepts much in the
> form of guidance.

Plenty of food for thought here, I'll make the observation that this
isn't simple or small ... LOL :)

Aside from the mathmatics of remeshing, there's a more general technical
challenge with this class of plugin, which is the lack of stable results
- basically, due to limited-precision math, you aren't guaranteed to get
repeatable outcomes with multiple runs on the same machine, let-alone on
different platforms or architectures.  That's not a show-stopper, it
just suggests a new type of plugin that acts as an "operation" on a
FrozenMesh, rather than being inserted into the pipeline.  I could
certainly work on this part, but we would need someone else to handle
the actual remeshing part.

Cheers,
Tim


[tshead.vcf]

---

begin:vcard
fn:Timothy Shead
n:Shead;Timothy
org:www.k-3d.org
email;internet:tshead@...
title:Founder
x-mozilla-html:FALSE
url:www.k-3d.org
version:2.1
end:vcard

---

------------------------------------------------------------------------------
Crystal Reports - New Free Runtime and 30 Day Trial
Check out the new simplified licensing option that enables unlimited
royalty-free distribution of the report engine for externally facing
server and web deployment.
http://p.sf.net/sfu/businessobjects
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development