I wondered how a poser figure would look like after occlusion map baking. So
here is a first attempt with a version of Poser 6 James I modified with a head
from a free 3d model (unfortunatelly I cannot remember where I have found it)
slightly.

I had to raise the threshold for the occlusion map to become it visible at all.
Here is James with Jaimes copper.

Best regards,
Michael

Post a reply to this message

Attachments:
Download 'james.png' (159 KB)

Preview of image 'james.png'

On 7-11-2012 23:14, MichaelJF wrote:
> I wondered how a poser figure would look like after occlusion map baking. So
> here is a first attempt with a version of Poser 6 James I modified with a head
> from a free 3d model (unfortunatelly I cannot remember where I have found it)
> slightly.
>
> I had to raise the threshold for the occlusion map to become it visible at all.
> Here is James with Jaimes copper.

This is looking great indeed. What happened to the nails?

[note to self: another thing to test out <sigh>]  ;-)

Thomas

Post a reply to this message

>
> This is looking great indeed. What happened to the nails?
>
> [note to self: another thing to test out <sigh>]  ;-)
>
> Thomas

I accidentelly omitted them. In the meantime I have merge the meshes (in the
image above they are not connected) which yielded new problems (hidden vertices
in the poser mesh caused a crumpled neck, not visible in Wings, PoseRay and
Meshlab but in POV, I didn't notice such issues with the Sphinx-round of the
TC-RTC). The nails proved to have a completely other occlusion map than their
surroundings. A miracle to me. Here I have now six meshes with their own uv-maps
and I baked the occlusion maps in an animated sequence but all against the whole
figure. I will give a new example if I have an own scene for it. Using Jaime's
is fine for comparision but one likes to have his own. I will postpone the roman
glasses issue and go for a labyrinth scene (that is why I'm in need of this
special version of James) even if someone has already entered his fingerprint to
the TC-RTC, a comparable but completely different scene.

Best regards,
Michael

Post a reply to this message

On 9-11-2012 18:13, MichaelJF wrote:
>
>>
>> This is looking great indeed. What happened to the nails?
>>
>> [note to self: another thing to test out <sigh>]  ;-)
>>
>> Thomas
>
> I accidentelly omitted them.

I thought so. They are easily forgotten, I have experienced.

> In the meantime I have merge the meshes (in the
> image above they are not connected) which yielded new problems (hidden vertices
> in the poser mesh caused a crumpled neck, not visible in Wings, PoseRay and
> Meshlab but in POV, I didn't notice such issues with the Sphinx-round of the
> TC-RTC). The nails proved to have a completely other occlusion map than their
> surroundings. A miracle to me. Here I have now six meshes with their own uv-maps
> and I baked the occlusion maps in an animated sequence but all against the whole
> figure. I will give a new example if I have an own scene for it. Using Jaime's
> is fine for comparison but one likes to have his own. I will postpone the roman
> glasses issue and go for a labyrinth scene (that is why I'm in need of this
> special version of James) even if someone has already entered his fingerprint to
> the TC-RTC, a comparable but completely different scene.

Looking forward to it indeed.

/Someone/ has always to pave the way it seems ;-) I have about finished 
my own contribution. Only the final render to do. No labyrinth though.

Thomas

Post a reply to this message

Here is the next version of "James". As you can see, there is a second character
(the Diego-Version of Poser 8 Ryan), a ball of wool and Ariadnes thread within a
cretan ground labyrinth. So Theseus task is a little bit pointless, since first
he could easily reach the minotaur directly, and second, the labyrinth has only
one path, no thread needed... And he has lost his club...

With Baking occlusion maps the main problem with Poser figures is that they
consist of more than one mesh and one has to bake the maps one after the other
for every mesh involved against the union of all meshes in an animated sequence.
The ball of wool is not from Poser and my own creation but shows the same
difficulty. Most problems derived from Diego's dynamic hair. PoseRay converts
the lines to tubes but gives only one mesh with all the tubes (hairs). The
uv-maps are overlapping heavily. This is o.k. for simply applying an uv-mapped
texture but deadly for use of the mesh-camera since POV cannot find the actual
faces. So what to do? My decision was to analyse the tubes-group mesh and
isolate every single hair with a third party software, which is more suited to
such tasks. Here I had 5103 hairs and I had to rebuild every single hair
(vertex_vectors, normal_vectors, uv_vectors, face_indices and so on) from the
original PoseRay tubes-group. The programming of this task was done within an
hour, the time of computation was a little bit longer... (To address an other
thread I started some weeks ago: even with the possibility of having a kind of
readln-directive this would have been possible with POV, but I didn't dare to
calculate the parsing time needed. If FlyerX read this: For use of the mesh-cam
it would be fine to have a mesh for every single Poser-"line".) Of course I
could have baked occlusion maps for all 5103 hairs (and ended Christmas next
year). I decided to have only a quarter of the hairs by random due to the
special structure here. I wanted to model a kind of bronze statue, so only a
small amount of hairs were really needed.  The most of the original dynamic
hairs from Poser are overlapping and not needed, so I left them out. But some
hairs are still missing. So here is the next WIP of "baked" Poser-Figures (grass
by Gilles Tran - of course).

At last I think that the idea to apply occlusion map baking to Poser figures
were worth all the effort, but (naked) they are too smooth to yield good
results, even after having morphed Diego to a bodybuilder. Maybe I should place
the waiting Ariadne at the empty spot. She should have more not convex areas for
occlusion;-)

Best regards,
Michael

Post a reply to this message

Attachments:
Download 'labyrinth_pbi.png' (3655 KB)

Preview of image 'labyrinth_pbi.png'

On 06/12/2012 8:27 PM, MichaelJF wrote:
[Snip]

Truly a Herculean task ;-)

> She should have more not convex areas for occlusion;-)
>

Yea, and text underneath, 'Here dwells Benedick the married man'? ;-)


-- 
Regards
     Stephen

Post a reply to this message

Stephen <mca### [at] aolcom> wrote:
> On 06/12/2012 8:27 PM, MichaelJF wrote:
> [Snip]
>
> Truly a Herculean task ;-)
>
> > She should have more not convex areas for occlusion;-)
> >
>
> Yea, and text underneath, 'Here dwells Benedick the married man'? ;-)
>
>
> --
> Regards
>      Stephen

LOL, thanks a lot.
Michael

Post a reply to this message

On 6-12-2012 21:27, MichaelJF wrote:
> Here is the next version of "James". As you can see, there is a second character
> (the Diego-Version of Poser 8 Ryan), a ball of wool and Ariadnes thread within a
> cretan ground labyrinth. So Theseus task is a little bit pointless, since first
> he could easily reach the minotaur directly, and second, the labyrinth has only
> one path, no thread needed... And he has lost his club...

Just as an aside here: have you experienced that squinting eye problem 
with the Poser 8 figures (Ryan/Alyson and avatars)? I found out that the 
problem occurs when passing from one pose to another, e.g. from zeropose 
to some other; the respective sideways angles of the eyes (-2 +2 
respectively) become zero, which happens to be the squint.

>
> At last I think that the idea to apply occlusion map baking to Poser figures
> were worth all the effort, but (naked) they are too smooth to yield good
> results, even after having morphed Diego to a bodybuilder. Maybe I should place
> the waiting Ariadne at the empty spot. She should have more not convex areas for
> occlusion;-)

Imo, even as smooth figures the result is as expected and certainly 
worthwhile.

Of course, your main trouble is coming from the animation goal. For a 
still the procedure would be much easier as you can combine all the 
individual meshes into a single one using any modeller, and use that for 
the baking.

Thomas

Post a reply to this message

Thomas de Groot <tho### [at] degrootorg> wrote:
> Just as an aside here: have you experienced that squinting eye problem
> with the Poser 8 figures (Ryan/Alyson and avatars)? I found out that the
> problem occurs when passing from one pose to another, e.g. from zeropose
> to some other; the respective sideways angles of the eyes (-2 +2
> respectively) become zero, which happens to be the squint.

Yes, may be. I have observed problems with the eyes in some of my poser scenes
yielding a squint. But it was with Poser 6 Jessie or James I still prefer over
Alyson/Ryan.

>
> Imo, even as smooth figures the result is as expected and certainly
> worthwhile.
>
> Of course, your main trouble is coming from the animation goal. For a
> still the procedure would be much easier as you can combine all the
> individual meshes into a single one using any modeller, and use that for
> the baking.
>
> Thomas

Hm, this must be a misunderstanding. I used an animation to compute the
occlusion maps one after the other for every mesh involved. Having 1107 meshes
in a union I had to bake the occlusion map for ever mesh against their union.
Every single mesh has to have its own uv-mapping. The animation was only a
technique to bake all the maps needed. To explain this a little bit more: I have
one mesh for the body of Theseus, one for his head, one for his filled holes
(with Meshlab), one for his penis and 1103 for his locks. The "animation" was
simply to switch through all this meshes and bake occlusion maps for every
single mesh. In the end I put them all together in applying the results to the
proper objects. For use with the mesh-cam you must have an uv-mapping for every
single mesh involved. You cannot combine them. That was the task here. With the
mesh-cam uv-mapping has to be different from simple texturing. With simple
texturing you can have overlapping areas, like PoseRay does it. FlyerX modifies
the uv-mapping slightly from tube to tube most likely to have a certain
variation with the texture but this did not address the issues of the mesh cam.
Here you need a one-to-one allocotion from the uv-map to the point at the object
surface. That was the reason to tear the PoseRay-results into parts and render
them one after the other in an "animated" sequence.

Best regards,
Michael
Best regards,
Michael

Post a reply to this message

On 7-12-2012 20:52, MichaelJF wrote:
> Hm, this must be a misunderstanding. I used an animation to compute the
> occlusion maps one after the other for every mesh involved. Having 1107 meshes
> in a union I had to bake the occlusion map for ever mesh against their union.
> Every single mesh has to have its own uv-mapping. The animation was only a
> technique to bake all the maps needed. To explain this a little bit more: I have
> one mesh for the body of Theseus, one for his head, one for his filled holes
> (with Meshlab), one for his penis and 1103 for his locks. The "animation" was
> simply to switch through all this meshes and bake occlusion maps for every
> single mesh. In the end I put them all together in applying the results to the
> proper objects. For use with the mesh-cam you must have an uv-mapping for every
> single mesh involved. You cannot combine them. That was the task here. With the
> mesh-cam uv-mapping has to be different from simple texturing. With simple
> texturing you can have overlapping areas, like PoseRay does it. FlyerX modifies
> the uv-mapping slightly from tube to tube most likely to have a certain
> variation with the texture but this did not address the issues of the mesh cam.
> Here you need a one-to-one allocotion from the uv-map to the point at the object
> surface. That was the reason to tear the PoseRay-results into parts and render
> them one after the other in an "animated" sequence.

I have not much experience with the mesh cam yet, which explains my 
comment :-) Thanks for the explanation.

Thomas

Post a reply to this message