: RE STACK FOCUSSER

[Goldsmith, Noel]

Hi,
NeLaS,

NeLaS said.

    I've been testing the Extended Depth of Field, Stack Focuser and Z
    Projections and the results are getting better each time I try something
    different! It seems that what you need to do really depends on the
original
    photos properties... so each set is a new challenge... Next time I'll
take
better photos :-)!

Noel replies
True so true.


I have one question about file format and one about taking photos for
focusing and 3D reconstruction.

1- I wanted to compare the results of focusing a JPG stack and a TIFF stack=
,
but I don't have any TIFF stack right now. Could the noise and abnormalitie=
s
present in my focused picture be a consequence of the jpg compression?

Noel says,
Absolutely, do not use JPG for this.

I know this sounds a bit obvious but maybe a high quality JPG does not
differ too much (or at least not visually) when compared to a focused TIFF
stack.
Just use Tiff or some other uncompressed unaltered data.



2- The minimum number of photos for a building a focused stack should be th=
e
Z axis object size / Depth of field. This assures that every part of the
study object has an "in focus" section. Is that correct?

Noel says  Seems to be true. However, what is the value for DOF  (depth of
field) (or is that depth of focus).
Consideration of how you can tell an image is in focus or not comes down to
two necessary conditions, which mean similar things from a different point
of view.

 Firstly from the energy point of view, the energy at focus is maximum. For
Luminous parts of the image, this means the in focus bright points are
brightest. For dark parts of the image the intensity is minimum.
Out of focus is featureless grey.
Secondly, from the fine detail point of view, when the change from pixel to
pixel is a maximum, you are at focus. That is the differential signal is
maximised.

The stack focusser works by looking at the changes between pixels. And the
wavelet focusser probably does a similar thing.

The question you are asking is: what change in the focus causes some
perceptible/measureable change in the image.
My experience is that the zone in focus slowly moves across a sloping
surface, and it has a finite width. The difficult thing is estimating just
where its edges lie. The space between really sharp and really blurred is
finite.
We look at metallic fractures and other opaque objects and it is my
experience that a finely etched surface, with lots of fine detail and
texture works really well with the sharpness based evaluations.
My calculations of the DOF based upon the 1/4 wavelength criterion (Rayleigh
Limit) indicate a number somewhat smaller than that which is perceptible.
If a 1/2 wavelength is chosen then this is more similar to what is seen.
However, in the interests of taking a few extra samples, ie oversampling
then there is no reason, apart from time constraints etc, to making the
slice depth any suitable arbitrary size.
Of course, if you obtain noise in your images and have no strategy to cope
with that you may end up with all of the noise in the result.



Now my question is: to get better results during Focusing stacks should I
take the minimum number of pictures or should I take more and overlap the
"in focus" areas of the object study??? don't know if I was clear enough...=
!

Noel says.
Take at least as many as the minimum, otherwise you get fuzzy areas, and
maybe add some more in a controlled way to see if it makes any difference.
The advantage of this is that you may be able to choose a good number for
the slice depth, and the smooth sliding of the in focus zone down a slope
may well define things more finely.

Hope this helps.
Noel