jo6nnughi3ngrjml71a5h1h7mko6b11dfe@4ax.com...
> I eventually did come across a reference to normals in the mesh2
> documentation - but I assumed OBJuvPOV was correctly dealing with
> them. Also, I believe the facetted effect is not so bad if area lights
> are not used - or so it would seem when switching between the 2 types
> when prototyping.

Another possibility is that it's simply the result of both Poser's and
POV-Ray's specific shortcomings with mesh normals. Poser has its usual
problem with joints and POV-Ray may display facets outlines under some
lighting conditions (the explanation is in the docs). I did have some ugly
conversion problems with normals back in 1998 when I first tried early
OBJ -> mesh2 converters, but I think that OBJuvPOV is OK.

BTW, Jeremy Engleman did some experiments about facetted effects on nudes,
if you're interested.
http://art.net/~jeremy/cg/figure5.3.html#img

G.


--
**********************
http://www.oyonale.com
**********************
- Graphic experiments
- POV-Ray and Poser computer images
- Posters

Post a reply to this message

"IMBJR" <no### [at] spamhere> wrote in message
news:e4cnnuoovdlfvgcaceejin4d40rcvj2315@4ax.com...
> On Sun, 08 Sep 2002 21:22:15 +0200, Fabien Mosen
> <fab### [at] skynetbe> wrote:
>
> >IMBJR wrote:

[snip]
> >But I like the hard edges in your picture.  It becomes a graphic
> >element in itself.
>
> I rather like them myself.
>

I don't. I find they scream 'bad 3D' and detract from an otherwise impressive
texture and lighting setup. In fact, it is so impressive I would really like to
see a version with the normals corrected.

The composition is oddly interesting as well. Why headless?

Post a reply to this message

"Batronyx" <bat### [at] alliancecablenet> wrote in message
news:3d7c1416$1@news.povray.org...
> "IMBJR" <no### [at] spamhere> wrote in message
> news:e4cnnuoovdlfvgcaceejin4d40rcvj2315@4ax.com...
> > On Sun, 08 Sep 2002 21:22:15 +0200, Fabien Mosen
> > <fab### [at] skynetbe> wrote:
> >
> > >IMBJR wrote:
>
> [snip]
> > >But I like the hard edges in your picture.  It becomes a graphic
> > >element in itself.
> >
> > I rather like them myself.
> >
>
> I don't. I find they scream 'bad 3D' and detract from an otherwise impressive
> texture and lighting setup.

        I disagree. Yes, it might 'scream bad 3D' but if that was painted in oils,
the artist would be acclaimed for it. I like it as it was intended.

    ~Steve~

Post a reply to this message

On Sun, 8 Sep 2002 23:00:44 +0200, "Gilles Tran" <git### [at] wanadoofr>
wrote:

>

>jo6nnughi3ngrjml71a5h1h7mko6b11dfe@4ax.com...
>> I eventually did come across a reference to normals in the mesh2
>> documentation - but I assumed OBJuvPOV was correctly dealing with
>> them. Also, I believe the facetted effect is not so bad if area lights
>> are not used - or so it would seem when switching between the 2 types
>> when prototyping.
>
>Another possibility is that it's simply the result of both Poser's and
>POV-Ray's specific shortcomings with mesh normals. Poser has its usual
>problem with joints and POV-Ray may display facets outlines under some
>lighting conditions (the explanation is in the docs). I did have some ugly
>conversion problems with normals back in 1998 when I first tried early
>OBJ -> mesh2 converters, but I think that OBJuvPOV is OK.
>
>BTW, Jeremy Engleman did some experiments about facetted effects on nudes,
>if you're interested.
>http://art.net/~jeremy/cg/figure5.3.html#img

They certainly have some charm about them.

>
>G.

--------------------
http://www.imbjr.com

Post a reply to this message

On Sun, 8 Sep 2002 22:22:05 -0500, "Batronyx"
<bat### [at] alliancecablenet> wrote:

>"IMBJR" <no### [at] spamhere> wrote in message
>news:e4cnnuoovdlfvgcaceejin4d40rcvj2315@4ax.com...
>> On Sun, 08 Sep 2002 21:22:15 +0200, Fabien Mosen
>> <fab### [at] skynetbe> wrote:
>>
>> >IMBJR wrote:
>
>[snip]
>> >But I like the hard edges in your picture.  It becomes a graphic
>> >element in itself.
>>
>> I rather like them myself.
>>
>
>I don't. I find they scream 'bad 3D' and detract from an otherwise impressive
>texture and lighting setup. In fact, it is so impressive I would really like to
>see a version with the normals corrected.

The quotes around bad 3D are so necessary - since it's all very
subjective.

As for correcting the normals, I wouldn't know where to begin any
ideas? It's, remember, a mesh2 generated by OBJuvPOV.

>
>The composition is oddly interesting as well. Why headless?

The headlessness corresponds to the hand jesture towards the viewer -
as if she is asking 'why' or even 'where'.

>
>

--------------------
http://www.imbjr.com

Post a reply to this message

> >I don't. I find they scream 'bad 3D' and detract from an otherwise impressive
> >texture and lighting setup. In fact, it is so impressive I would really like
to
> >see a version with the normals corrected.
>
> The quotes around bad 3D are so necessary - since it's all very
> subjective.

Yep, subjective indeed. For example if it were me, and I chose to
keep the lines as I put it to canvas, I would likely choose to make
that forward knee considerably more subtle. It's the thing bugging
me the most.

But then, Boris Vallejo is my favorite artist, I think
Van Gogh was suffering from astigmatism, and I'm convinced Picasso
was just a lucky pretender. So what do I know? :)


>
> As for correcting the normals, I wouldn't know where to begin any
> ideas? It's, remember, a mesh2 generated by OBJuvPOV.

Several things come to mind. A combination of things may be required.
I haven't used OBJuvPOV, but if it has an option to increase the
smoothing angle of the normals that would be the first thing to try.
That seems to be the most obvious problem, around the arms. 3Dwin and
Crossroads both support this option. But I think only 3Dwin can deal
with the uv coords.

For that knee, you would almost have to pull it into a modeller and
manually tweak the points, into a smoother position. Same for the far
hip, but it could ride as is the shadows. Anim8or, for example, is free
and supports the .obj format. Point selection can be tedious but zooming
in and learning the virtual trackball helps a bunch. If you do this
you might want to save the uv coords as a separate file and re-apply them later.
UVMapper can assist with this.

As Gilles pointed out some of this may be due to the way POV renders
meshes, particularly at glancing angles. Double illuminate could help
a little, but I saw a post here some months ago suggesting that a second
copy of the mesh translated ever so slightly behind the first ( say .00001
or something like that) can also help overcome those artifacts.


> >The composition is oddly interesting as well. Why headless?
>
> The headlessness corresponds to the hand jesture towards the viewer -
> as if she is asking 'why' or even 'where'.

Ok. I see 'where' more than 'why', but I see it now. Thanks.

Post a reply to this message

Wasn't it IMBJR who wrote:

>The quotes around bad 3D are so necessary - since it's all very
>subjective.
>
>As for correcting the normals, I wouldn't know where to begin any
>ideas? It's, remember, a mesh2 generated by OBJuvPOV.

Did you actually remove the head in Poser, or did you do it with some
external technique afterwards? It's my guess that you may have got the
head removal slightly wrong, resulting in the set of deleted normals not
being exactly the same as the set of deleted points, thus resulting in
the remaining normal_indices being associated with the wrong
face_indices. 

Or possibly the wrong uv_indices being associated with the wrong
face_indices - I'm not 100% convinced that the facets are due to wrong
normals, they look to me like the wrong bits of the image map might be
being applied to each triangle. It's much easier the get the uv_indices
wrong if you modified the INC file created by OBJuvPOV, because they are
presented in a different order to that of the face_ and normal_indices.

The Poser-PVJuvPOV-POVRay combination usually gets the normals and uv
mapping correct, so if any unusual additional processing were performed
I'd tend to suspect that.

-- 
Mike Williams
Gentleman of Leisure

Post a reply to this message

On Tue, 10 Sep 2002 09:25:49 +0100, Mike Williams
<mik### [at] econymdemoncouk> wrote:

>Wasn't it IMBJR who wrote:
>
>>The quotes around bad 3D are so necessary - since it's all very
>>subjective.
>>
>>As for correcting the normals, I wouldn't know where to begin any
>>ideas? It's, remember, a mesh2 generated by OBJuvPOV.
>
>Did you actually remove the head in Poser, or did you do it with some
>external technique afterwards? It's my guess that you may have got the

Do you know - I can't recall - but I strongly suspect it was done in
Poser and not OBJuvPOV.

>head removal slightly wrong, resulting in the set of deleted normals not
>being exactly the same as the set of deleted points, thus resulting in
>the remaining normal_indices being associated with the wrong
>face_indices. 

That surely would not effect the whole body.

>
>Or possibly the wrong uv_indices being associated with the wrong
>face_indices - I'm not 100% convinced that the facets are due to wrong
>normals, they look to me like the wrong bits of the image map might be
>being applied to each triangle. It's much easier the get the uv_indices
>wrong if you modified the INC file created by OBJuvPOV, because they are
>presented in a different order to that of the face_ and normal_indices.
>
>The Poser-PVJuvPOV-POVRay combination usually gets the normals and uv
>mapping correct, so if any unusual additional processing were performed
>I'd tend to suspect that.

It would appear that area lights really bring out this effect, normal
spotlights are far more forgiving.

--------------------
http://www.imbjr.com

Post a reply to this message