Thanks, Jérôme: it works like a charm. Instead tracing from random
points around the mesh, with hgpovray I can trace from points just in
front of each vertex, so I can reach places that would be inaccessible
with the classic method. Here is a test with a knot from Wings3D (I just
had to invert the mesh, as the normals seem to come inverted on the
POV-Ray export).

Now I just have to model a teddy bear with Wings3D... :)

--
jaime

Post a reply to this message

Attachments:
Download 'knot_fur_test-01-2m29s.jpg' (100 KB)

Preview of image 'knot_fur_test-01-2m29s.jpg'

Jaime Vives Piqueres <jai### [at] ignoranciaorg> wrote:
> Thanks, Jérôme: it works like a charm. Instead tracing from random
> points around the mesh, with hgpovray I can trace from points just in
> front of each vertex, so I can reach places that would be inaccessible
> with the classic method. Here is a test with a knot from Wings3D (I just
> had to invert the mesh, as the normals seem to come inverted on the
> POV-Ray export).
>
> Now I just have to model a teddy bear with Wings3D... :)
>
> --
> jaime


Nice as always...

Margus Ramst wrote a nice fur macro in 2000
(http://news.povray.org/povray.binaries.scene-files/thread/%3C39A10353.9F08EA9E@peak.edu.ee%3E/?ttop=292702&toff=850).
It isn't really fast and it needs much memory, but it is very versatile
regarding form, gravity influence and texturing. Beside that it covers objects
very good.


Norbert

Post a reply to this message

Attachments:
Download 'material_1740.jpg' (831 KB)

Preview of image 'material_1740.jpg'

> Margus Ramst wrote a nice fur macro in 2000

   Another one I missed back on the day... it's very good, specially the
gravity thing. Should take a look at the code, but I'm fearing it will
be very complex...

--
jaime

Post a reply to this message

On 8-3-2016 21:33, Norbert Kern wrote:
> Jaime Vives Piqueres <jai### [at] ignoranciaorg> wrote:
>> Thanks, Jérôme: it works like a charm. Instead tracing from random
>> points around the mesh, with hgpovray I can trace from points just in
>> front of each vertex, so I can reach places that would be inaccessible
>> with the classic method. Here is a test with a knot from Wings3D (I just
>> had to invert the mesh, as the normals seem to come inverted on the
>> POV-Ray export).
>>
>> Now I just have to model a teddy bear with Wings3D... :)
>>
>> --
>> jaime
>
>
> Nice as always...
>
> Margus Ramst wrote a nice fur macro in 2000

<grin> I was going to mention that same macro indeed.

I appreciate the possibility to reach difficult locations on the mesh.


-- 
Thomas

Post a reply to this message

Jaime Vives Piqueres wrote on 08/03/2016 18.18:
> Thanks, Jérôme: it works like a charm. Instead tracing from random
> points around the mesh, with hgpovray I can trace from points just in
> front of each vertex, so I can reach places that would be inaccessible
> with the classic method. Here is a test with a knot from Wings3D (I just
> had to invert the mesh, as the normals seem to come inverted on the
> POV-Ray export).
>
> Now I just have to model a teddy bear with Wings3D... :)
>
> --
> jaime
Oh! This knot is interesting too...
;-)
Paolo

Post a reply to this message

No teddy bear yet... but I tried with a towel model from archive3d
(tried to model one with Megapov simcloth, but didn't get anything decent).

   I figured a better method than tracing the threads: instead, I just
loop over all the triangles, then for each triangle I place 20 hairs at
randoms points on the triangle surface. I searched on the internet for
methods to do this, but as I didn't understand any of them due to my
poor knowledge of formal math notation, I ended up doing my own thing to
find random points on a triangle surface... it's ugly, slow and surely
inefficient, but works fine visually.

   Then there was The Last Problem... I wanted to texture the threads so
that they follow the uv_mapped texture on the mesh, but eval_pigment
will not work on this scenario. At that moment I realized the mesh data
extraction functions on hgpovray are missing one function: something
like get_uv_coordinates().

   But squeezing a bit more my poor brain I figured a workaround: copy
and paste the uv_vertex info from the mesh include file into a separate
file, formatted as an array, and then use that with eval_pigment and a
flat version of the image_map. Fortunately, the uv_vertex and
uv_vertices are correlative, so it works fine. But it would be much
better (and faster) to have an hgpovray function for it.

   Regards,

--
jaime

Post a reply to this message

Attachments:
Download 'hanging-towel-04-p4m21s-r12m48s.jpg' (185 KB)

Preview of image 'hanging-towel-04-p4m21s-r12m48s.jpg'

Le 17/03/2016 21:03, Jaime Vives Piqueres a écrit :
>  Fortunately, the uv_vertex and
> uv_vertices are correlative, so it works fine. But it would be much
> better (and faster) to have an hgpovray function for it.

Oh, I see, the towel is for the sweat once the work has been done...

so far no work done, so the towel is still fresh :-)

Post a reply to this message

Le_Forgeron <jgr### [at] freefr> wrote:

> so far no work done, so the towel is still fresh :-)

I think the only thing _fresh_ here is YOU    :D    ;)


Great work, Jaime - it looks very natural and has a wonderful texture.
I think you could use your method for some interesting other textures - moss, or
the fuzz on a peach, or those plants with leaves with fine hairs covering them
(psoralea?) - certainly violets.

Thanks as always for sharing!

Post a reply to this message

On 17-3-2016 21:03, Jaime Vives Piqueres wrote:
>    No teddy bear yet... but I tried with a towel model from archive3d
> (tried to model one with Megapov simcloth, but didn't get anything decent).
>
>    I figured a better method than tracing the threads: instead, I just
> loop over all the triangles, then for each triangle I place 20 hairs at
> randoms points on the triangle surface. I searched on the internet for
> methods to do this, but as I didn't understand any of them due to my
> poor knowledge of formal math notation, I ended up doing my own thing to
> find random points on a triangle surface... it's ugly, slow and surely
> inefficient, but works fine visually.

I never got to tracing threads on something like this and suspected that 
there would be problems with the folds. Your hgpovray methods seems to 
really work well indeed. It forms an excellent proof-of-concept for 
regular development on POV-Ray's main branch one day :-)

-- 
Thomas

Post a reply to this message

El 18/03/16 a las 08:55, Thomas de Groot escribió:
> I never got to tracing threads on something like this and suspected
> that there would be problems with the folds. Your hgpovray methods
> seems to really work well indeed. It forms an excellent
> proof-of-concept for regular development on POV-Ray's main branch one
> day :-)

   Yes, it would be fantastic to have these mesh data functions
integrated into official POV-Ray. I can't see these features breaking
anything else, but I'm not an expert, of course. The icing on the cake
would be, as I said, a function to get the uv coordinates at a given
vertex (this one alone would overcome the problem with eval_pigment()
and uv_mapped pigments).

   Here is a new test, with another towel, but this time folded and
resting over a surface (again the model is from archive3d.net, but the
uv mapping on this one was messed up, so I didn't use it). Still needs a
bit more threads per triangle, but it's already made of 897940 threads
(20 per triangle). It takes 7 minutes to parse, mainly due to my ugly
method to find 20 random points over the surface of the triangle.

--
jaime

Post a reply to this message

Attachments:
Download 'folded-towel-01-p7m5s-r48m5s.jpg' (269 KB)

Preview of image 'folded-towel-01-p7m5s-r48m5s.jpg'