StereoView after StackFocuser

[NeLaS]

Hello!

I have a rgb stack I would like to transform in a stereo view image. I was
doing the sequence:

>Open image sequence
>Run Extended Depth of Field Plugin (EDF)
>Run Anaglyph plugin (G.Landini) and mark to generate a stereo image

However, the topology map created by EDF does not seem to reflect the
stack's resulting topology, but only the single output image's topology. If
you compare the
heightmap<http://i4.photobucket.com/albums/y115/NeLaS/HeightMap.png>generated
by the Stack Focuser plugin and the
Topology <http://i4.photobucket.com/albums/y115/NeLaS/Topology.png> from EDF
you can see the difference. This is a 2-cell echinoderm embryo, where the
cell on the left is on a different focus plane of the cell on the right
(checkout a slice <http://i4.photobucket.com/albums/y115/NeLaS/2cell.png> to
get the idea). You can only notice this difference on the heightmap image.

So, as I want to give a perspective to a onfocus 2D image, using stereo
view, I'd rather use stack focuser's height map because it seems to preserve
the original stack information.

I tried to use Anaglyph plugin with the Height Map, but with no success! The
resulting image is absolutely distorted. I assigned the output and heightmap
to 32bit and 8bit, but it made no difference.

Does anybody know how to build a stereo view image using the resulting
images from Stack Focuser (focused and heightmap)?

thanks!!!

bruno

ps: sorry for my ignorance, but is heightmap = topology ?


--
Bruno C. Vellutini
organelas.com
ccnelas.org

Centro de Biologia Marinha (CEBIMar)
Universidade de São Paulo
Av. Manoel H. do Rego km 131,5
11600-000 São Sebastião, SP, Brasil
http://www.usp.br/cbm/

[Gabriel Landini]

On Friday 13 April 2007 02:23:44 NeLaS wrote:
> doing the sequence:
> >Open image sequence
> >Run Extended Depth of Field Plugin (EDF)
> >Run Anaglyph plugin (G.Landini) and mark to generate a stereo image
>
> However, the topology map created by EDF does not seem to reflect the
> stack's resulting topology, but only the single output image's topology.

Maybe you are seeing the results of different methods to decide what is on
focus when you use semi-transparent objects.

These plugins would work perfectly on stacks of images of opaque objects. The
problem with semi-transparent objects (like those cells) is that they can be
on focus at several places in the z plane. The plugins try to maximise some
focus function in the z direction. Depending on the implementation this may
be found at different places.

Just curious, which of the topology maps give you the best focused result?

Can you upload the stack for testing?

> I tried to use Anaglyph plugin with the Height Map, but with no success!
> The resulting image is absolutely distorted. I assigned the output and
> heightmap to 32bit and 8bit, but it made no difference.

> ps: sorry for my ignorance, but is heightmap = topology ?
Yes, although the heightmap is not guaranteed to be meaningful (measure-wise)
unless your stacks are all taken at the same step depth.

Regards

G.

[NeLaS]

Hi Gabriel!

Maybe you are seeing the results of different methods to decide what is on

> focus when you use semi-transparent objects.
>
> These plugins would work perfectly on stacks of images of opaque objects.
> The
> problem with semi-transparent objects (like those cells) is that they can
> be
> on focus at several places in the z plane. The plugins try to maximise
> some
> focus function in the z direction. Depending on the implementation this
> may
> be found at different places.

Hum, interesting point!

Just curious, which of the topology maps give you the best focused result?
>

That's a difficult question, but I think the stack focuser plugin is giving
better ones. If I use kernel size of 3, the focused image turns out
"average" because of the abundance of "noise", final image is grainy. If I
use kernels bigger than 100 I get really nice resolved focus in most areas
of the image, but the kernel patches appear explicitly (while using smaller
kernels these patches are so small they look like noise).

The stack focuser seems to interpret the stack's focus information as I
would expect, but EDF apparently mix the onfocus information (maybe because
of the opacity of images?). I have tried other configurations for EDF, on
the advanced mode, but the resulting images were not too different regarding
the topology.

Can you upload the stack for testing?


sure! the .zip can be downloaded
here<http://filexoom.com/files/12783/2cell.zip>.
It contains the stack (8bit), a focused image processed with k=3 and another
with k=150 (StackFocuser), and the output image from Extended Depth of Field
(Easy mode -> Speed/quality "high").

thanks!

bruno


On 4/13/07, Gabriel Landini <[hidden email]> wrote:

>
> On Friday 13 April 2007 02:23:44 NeLaS wrote:
> > doing the sequence:
> > >Open image sequence
> > >Run Extended Depth of Field Plugin (EDF)
> > >Run Anaglyph plugin (G.Landini) and mark to generate a stereo image
> >
> > However, the topology map created by EDF does not seem to reflect the
> > stack's resulting topology, but only the single output image's topology.
>
> Maybe you are seeing the results of different methods to decide what is on
> focus when you use semi-transparent objects.
>
> These plugins would work perfectly on stacks of images of opaque objects.
> The
> problem with semi-transparent objects (like those cells) is that they can
> be
> on focus at several places in the z plane. The plugins try to maximise
> some
> focus function in the z direction. Depending on the implementation this
> may
> be found at different places.
>
> Just curious, which of the topology maps give you the best focused result?
>
> Can you upload the stack for testing?
>
> > I tried to use Anaglyph plugin with the Height Map, but with no success!
> > The resulting image is absolutely distorted. I assigned the output and
> > heightmap to 32bit and 8bit, but it made no difference.
>
> > ps: sorry for my ignorance, but is heightmap = topology ?
> Yes, although the heightmap is not guaranteed to be meaningful
> (measure-wise)
> unless your stacks are all taken at the same step depth.
>
> Regards
>
> G.
>

--
Bruno C. Vellutini
organelas.com

Centro de Biologia Marinha (CEBIMar)
Universidade de São Paulo
Av. Manoel H. do Rego km 131,5
11600-000 São Sebastião, SP, Brasil
http://www.usp.br/cbm/

[NeLaS]

I just realized that if I use Sobel or Variance as Sharpness Estimator on
the EDF, I get similar results to stackfocuser! But If I use Real or Complex
Wavelets, the topology gets mixed.

Kind of obvious since Stack Focuser uses Sobel... :-)

On 4/13/07, NeLaS <[hidden email]> wrote:

>
> Hi Gabriel!
>
> Maybe you are seeing the results of different methods to decide what is on
> >
> > focus when you use semi-transparent objects.
> >
> > These plugins would work perfectly on stacks of images of opaque
> > objects. The
> > problem with semi-transparent objects (like those cells) is that they
> > can be
> > on focus at several places in the z plane. The plugins try to maximise
> > some
> > focus function in the z direction. Depending on the implementation this
> > may
> > be found at different places.
>
>
> Hum, interesting point!
>
> Just curious, which of the topology maps give you the best focused result?
> >
>
> That's a difficult question, but I think the stack focuser plugin is
> giving better ones. If I use kernel size of 3, the focused image turns out
> "average" because of the abundance of "noise", final image is grainy. If I
> use kernels bigger than 100 I get really nice resolved focus in most areas
> of the image, but the kernel patches appear explicitly (while using smaller
> kernels these patches are so small they look like noise).
>
> The stack focuser seems to interpret the stack's focus information as I
> would expect, but EDF apparently mix the onfocus information (maybe because
> of the opacity of images?). I have tried other configurations for EDF, on
> the advanced mode, but the resulting images were not too different regarding
> the topology.
>
> Can you upload the stack for testing?
>
>
> sure! the .zip can be downloaded here<http://filexoom.com/files/12783/2cell.zip>.
> It contains the stack (8bit), a focused image processed with k=3 and another
> with k=150 (StackFocuser), and the output image from Extended Depth of Field
> (Easy mode -> Speed/quality "high").
>
> thanks!
>
> bruno
>
>
> On 4/13/07, Gabriel Landini <[hidden email]> wrote:
> >
> > On Friday 13 April 2007 02:23:44 NeLaS wrote:
> > > doing the sequence:
> > > >Open image sequence
> > > >Run Extended Depth of Field Plugin (EDF)
> > > >Run Anaglyph plugin (G.Landini) and mark to generate a stereo image
> > >
> > > However, the topology map created by EDF does not seem to reflect the
> > > stack's resulting topology, but only the single output image's
> > topology.
> >
> > Maybe you are seeing the results of different methods to decide what is
> > on
> > focus when you use semi-transparent objects.
> >
> > These plugins would work perfectly on stacks of images of opaque
> > objects. The
> > problem with semi-transparent objects (like those cells) is that they
> > can be
> > on focus at several places in the z plane. The plugins try to maximise
> > some
> > focus function in the z direction. Depending on the implementation this
> > may
> > be found at different places.
> >
> > Just curious, which of the topology maps give you the best focused
> > result?
> >
> > Can you upload the stack for testing?
> >
> > > I tried to use Anaglyph plugin with the Height Map, but with no
> > success!
> > > The resulting image is absolutely distorted. I assigned the output and
> > > heightmap to 32bit and 8bit, but it made no difference.
> >
> > > ps: sorry for my ignorance, but is heightmap = topology ?
> > Yes, although the heightmap is not guaranteed to be meaningful
> > (measure-wise)
> > unless your stacks are all taken at the same step depth.
> >
> > Regards
> >
> > G.
> >
>
>
>
> --
> Bruno C. Vellutini
> organelas.com
>
> Centro de Biologia Marinha (CEBIMar)
> Universidade de São Paulo
> Av. Manoel H. do Rego km 131,5
> 11600-000 São Sebastião, SP, Brasil
> http://www.usp.br/cbm/
>

--
Bruno C. Vellutini
organelas.com

Centro de Biologia Marinha (CEBIMar)
Universidade de São Paulo
Av. Manoel H. do Rego km 131,5
11600-000 São Sebastião, SP, Brasil
http://www.usp.br/cbm/

[Mikhail Umorin]

I am going to work on Stack_Focuser. I shall try to introduce a different edge
detector. If you have any questions or suggestions -- let me know.

Mikhail Umorin.


On Friday 13 April 2007 10:15, NeLaS wrote:

> I just realized that if I use Sobel or Variance as Sharpness Estimator on
> the EDF, I get similar results to stackfocuser! But If I use Real or
> Complex Wavelets, the topology gets mixed.
>
> Kind of obvious since Stack Focuser uses Sobel... :-)
>
> On 4/13/07, NeLaS <[hidden email]> wrote:
> > Hi Gabriel!
> >
> > Maybe you are seeing the results of different methods to decide what is
> > on
> >
> > > focus when you use semi-transparent objects.
> > >
> > > These plugins would work perfectly on stacks of images of opaque
> > > objects. The
> > > problem with semi-transparent objects (like those cells) is that they
> > > can be
> > > on focus at several places in the z plane. The plugins try to maximise
> > > some
> > > focus function in the z direction. Depending on the implementation this
> > > may
> > > be found at different places.
> >
> > Hum, interesting point!
> >
> > Just curious, which of the topology maps give you the best focused
> > result?
> >
> >
> > That's a difficult question, but I think the stack focuser plugin is
> > giving better ones. If I use kernel size of 3, the focused image turns
> > out "average" because of the abundance of "noise", final image is grainy.
> > If I use kernels bigger than 100 I get really nice resolved focus in most
> > areas of the image, but the kernel patches appear explicitly (while using
> > smaller kernels these patches are so small they look like noise).
> >
> > The stack focuser seems to interpret the stack's focus information as I
> > would expect, but EDF apparently mix the onfocus information (maybe
> > because of the opacity of images?). I have tried other configurations for
> > EDF, on the advanced mode, but the resulting images were not too
> > different regarding the topology.
> >
> > Can you upload the stack for testing?
> >
> >
> > sure! the .zip can be downloaded
> > here<http://filexoom.com/files/12783/2cell.zip>. It contains the stack
> > (8bit), a focused image processed with k=3 and another with k=150
> > (StackFocuser), and the output image from Extended Depth of Field (Easy
> > mode -> Speed/quality "high").
> >
> > thanks!
> >
> > bruno
> >
> > On 4/13/07, Gabriel Landini <[hidden email]> wrote:
> > > On Friday 13 April 2007 02:23:44 NeLaS wrote:
> > > > doing the sequence:
> > > > >Open image sequence
> > > > >Run Extended Depth of Field Plugin (EDF)
> > > > >Run Anaglyph plugin (G.Landini) and mark to generate a stereo image
> > > >
> > > > However, the topology map created by EDF does not seem to reflect the
> > > > stack's resulting topology, but only the single output image's
> > >
> > > topology.
> > >
> > > Maybe you are seeing the results of different methods to decide what is
> > > on
> > > focus when you use semi-transparent objects.
> > >
> > > These plugins would work perfectly on stacks of images of opaque
> > > objects. The
> > > problem with semi-transparent objects (like those cells) is that they
> > > can be
> > > on focus at several places in the z plane. The plugins try to maximise
> > > some
> > > focus function in the z direction. Depending on the implementation this
> > > may
> > > be found at different places.
> > >
> > > Just curious, which of the topology maps give you the best focused
> > > result?
> > >
> > > Can you upload the stack for testing?
> > >
> > > > I tried to use Anaglyph plugin with the Height Map, but with no
> > >
> > > success!
> > >
> > > > The resulting image is absolutely distorted. I assigned the output
> > > > and heightmap to 32bit and 8bit, but it made no difference.
> > > >
> > > > ps: sorry for my ignorance, but is heightmap = topology ?
> > >
> > > Yes, although the heightmap is not guaranteed to be meaningful
> > > (measure-wise)
> > > unless your stacks are all taken at the same step depth.
> > >
> > > Regards
> > >
> > > G.
> >
> > --
> > Bruno C. Vellutini
> > organelas.com
> >
> > Centro de Biologia Marinha (CEBIMar)
> > Universidade de São Paulo
> > Av. Manoel H. do Rego km 131,5
> > 11600-000 São Sebastião, SP, Brasil
> > http://www.usp.br/cbm/