This is more a proof of concept than a serious image. I came up with the idea to
combine the eggshell-ghost-idea by Bill Pragnell with the makecloud-macro by
Gilles Tran.

The ghost is a MakeHuman model posed with Blender. My small experience with that
explains the stiff pose a bit :( There is an inner ghost, which is scaled down a
small amount along the normals, substracted from the original outer ghost. The
difference is run through the makecloud-macro, resulting in a df3 file used for
an emitting media. Clothing and hair were created with a very small
shell-extrude (Wings 3D) along the normals and converted to df3 like the ghost.
I had to left out the eyes since there is still an artefact there I could not
track down so far.

The scene is inspired by an old pen and paper role playing game I played ages
ago. So it is pure fantasy and not intended for the "Legends"-round at TC-RTC.

Credits: For the fire place I started with the POV-Comp entry by Jussi Kantola,
modified the densities and ended with a version not much more convincing than
the original.

Seems my next project should be about radiosity...

Best regards,
Michael

Post a reply to this message

Attachments:
Download 'evocation1.jpg' (111 KB)

Preview of image 'evocation1.jpg'

On 26-9-2013 8:35, MichaelJF wrote:
> This is more a proof of concept than a serious image. I came up with the idea to
> combine the eggshell-ghost-idea by Bill Pragnell with the makecloud-macro by
> Gilles Tran.

I think that worked well so far.

>
> Credits: For the fire place I started with the POV-Comp entry by Jussi Kantola,
> modified the densities and ended with a version not much more convincing than
> the original.

The fire is a bit too small to be able to judge the problem. The sparks 
maybe?

Thomas

Post a reply to this message

Seeing the transparency in your ghost reminds me of what I did that I forgot
about. After many dissatisfying tests I eventually made my ghost opaque, using
the media shell to provide glowing edges. Although this wasn't the effect I
originally set out to produce (I was thinking of the classic movie effects that
we see in, e.g. Ghostbusters), it worked ok in my image because there wasn't
much scenery behind the ghost.

I remember having trouble with the eyes, too :/

Having also seen Thomas' tests with the head mesh, perhaps a better (if
time-consuming) technique would be to use dense, dark media for an inner mesh,
with brighter media in the shell to provide the edges...

Post a reply to this message

On 26-9-2013 12:56, Bill Pragnell wrote:
> I remember having trouble with the eyes, too :/

The technique I am using does not have those problems though (see my 
latest test).

>
> Having also seen Thomas' tests with the head mesh, perhaps a better (if
> time-consuming) technique would be to use dense, dark media for an inner mesh,
> with brighter media in the shell to provide the edges...
>

Yes, I think so too. However, I also think that the edges should not be 
too bright in that case.

Thomas

Post a reply to this message

Thank you Thomas and Bill for your comments. Yes, first the fire is to small and
second the sparks are a bit annoying. The sparks are my addition and I wanted
them look like the glowing logs of the fire, so I used the same texture. But I
will address this later.

And for ghosts: Yes Bill, if you will have a ghost like in ghostbusters one has
to use another approach. In this case Thomas approach is superior. But I fear
that using a spherical density works well with an object not very far away from
convexity - as Thomas has shown us - like a head but worse with a whole body.
Further ghosts are described in history in many forms. Take a look at the first
"photography" at

http://en.wikipedia.org/wiki/Ghost

That is more the kind of ghost I had in my mind. I tried such ghosts earlier
after your original approach but yielded some annoying white spots mainly at the
joints.

http://www.tc-rtc.co.uk/imagenewdisplay/stills/index395.html

(In the current scene you can see white spots at the armpits of the ghost, but
this are stars from the background by chance. With the next version I will have
rotated the stars away from this points.) With an df3 approach this should not
be possible (in theory).

BTW: the artefacts I considered as radiosity artefacts turned out to be a miss
scaled sky pigment producing this strange Moire effects. I never expected that,
but you will find the most stupid errors ever at last.

Here is the next version correcting the pigment error, exchanged the ground from
a badly scaled height_field to a mesh and rotated the stars out of the ghost.
Accidentally the new ground moved the randomly placed trees, but may be it is
better to have the ghost in front of a tree. The ghost is now scaled according
to Cousin Ricky's proposal, but it still looks younger as intended. The small
deformations of the arms and legs are due to a slight warp to the density and
not to the original MakeHuman figure.

Best regards,
Michael

Post a reply to this message

Attachments:
Download 'evocation2.jpg' (93 KB)

Preview of image 'evocation2.jpg'

On 26-9-2013 22:16, MichaelJF wrote:
> Thank you Thomas and Bill for your comments. Yes, first the fire is to small and
> second the sparks are a bit annoying. The sparks are my addition and I wanted
> them look like the glowing logs of the fire, so I used the same texture. But I
> will address this later.

I think the sparks should be smaller and brighter. I leave it to you ;-)

>
> And for ghosts: Yes Bill, if you will have a ghost like in ghostbusters one has
> to use another approach. In this case Thomas approach is superior. But I fear
> that using a spherical density works well with an object not very far away from
> convexity - as Thomas has shown us - like a head but worse with a whole body.
> Further ghosts are described in history in many forms. Take a look at the first
> "photography" at
>
> http://en.wikipedia.org/wiki/Ghost
>
> That is more the kind of ghost I had in my mind. I tried such ghosts earlier
> after your original approach but yielded some annoying white spots mainly at the
> joints.

I agree. I prefer the "ectoplasm" effect of an internal media. I 
remembered the descriptions in "The land of mist" by Arthur Conan Doyle 
(when did I read that...? more than 50 years ago!) where the spirits are 
described as taking shape from a kind of thin mist emanating from the 
medium.

The added spherical density in my ghost is awkward indeed and certainly 
not to be used everywhere. However, marble or agate might give 
interesting results when properly scaled.

>
> http://www.tc-rtc.co.uk/imagenewdisplay/stills/index395.html
>
> (In the current scene you can see white spots at the armpits of the ghost, but
> this are stars from the background by chance. With the next version I will have
> rotated the stars away from this points.) With an df3 approach this should not
> be possible (in theory).
>
> BTW: the artefacts I considered as radiosity artefacts turned out to be a miss
> scaled sky pigment producing this strange Moire effects. I never expected that,
> but you will find the most stupid errors ever at last.
>
> Here is the next version correcting the pigment error, exchanged the ground from
> a badly scaled height_field to a mesh and rotated the stars out of the ghost.
> Accidentally the new ground moved the randomly placed trees, but may be it is
> better to have the ghost in front of a tree. The ghost is now scaled according
> to Cousin Ricky's proposal, but it still looks younger as intended. The small
> deformations of the arms and legs are due to a slight warp to the density and
> not to the original MakeHuman figure.

Note that my own interpretation of "25%" is at fault. Experimentally, 
from a box, I concluded that an x-scale 1.25 was about correct.

Thomas

Post a reply to this message

This is the last version of this exercise. The main issue with Poser- or
MakeHuman-models in this case is to ensure that there are no holes. That means
first, that after filling the holes with MeshLab e.g., better close MeshLab and
start it again with the resulting mesh to ensure that there is really no hole
left. I experienced twice that MeshLab stated to have closed all holes, but
found strange artifacts in the image, and yes, after reloading, a new hole
popped up (seems one has to use the "trivial" option for closing holes with
MeshLab). Second before running the mesh through PoseRay assure that there is
only one material, may be by loading the mesh in Wings, give all faces the same
material and then pass it to PoseRay (most likely PoseRay can do this, but I'm
to stupid to understand the settings within the material tab). It is paramount
that finally there is only one mesh and not a union of meshes. Otherwise Gilles
Tran's makecloud-algorithm will produce very strange noise artifacts.

Other changes: more, smaller and brighter sparkles and the fire is df3 too. And
some df3 clouds, just for the fun of it. Done with the original makecloud-macro
of course.

Best regards,
Michael

Post a reply to this message

Attachments:
Download 'evocation3.jpg' (199 KB)

Preview of image 'evocation3.jpg'