TriDef Photo Transformer - Tutorials
Advanced Depth Map Settings
Training objectives
At the end of this tutorial you will be able to:
-
Adjust Search Radius
-
Adjust Search Tolerance
-
Adjust Source Blur
-
Adjust Bias X,Y
Depth Map Options
There are several options available to you which affect the
way in which Photo Transformer turns Depth Data points into
a Depth Map. You will find that different types of image
require different types of settings to produce the best
Depth Map from the fewest Depth Data points. These settings
affect the entire image, so you have to seek a balance to
get the best result.
All settings are found in Edit > Preferences >
Depth. You can also leave these settings alone if you
wish and just add a few more Depth Data and Transparent
points to complete your Depth Map. This is the recommended
approach for beginners. These settings become most useful
when working with either larger image or slower machines,
when having fewer points helps to keep Photo Transformer
responsive.
Search Radius
Search Radius determines how much of the image around a
Depth Data point is searched for potentially matching depth
values. The smaller the Search Radius, the more Depth Data
points you have to add to the image. A larger Search Radius
can yield a more accurate Depth Map with fewer Depth Data
points, but it takes longer to update the Depth Map.
Whether a depth value in this area is added to the Depth
Map depends on the Search Tolerance.
Search Tolerance
Search Tolerance determines which parts of the image in the
Search Radius area around a Depth Data point qualify as a
matching depth value. These matching points are then added
to the Depth Map. If your image has quite high color
contrast between adjacent objects at different depths, then
you can make the Search Tolerance higher and use fewer
Depth Data points. If your image has poor contrast between
these objects you should use a lower Search Tolerance and
add more Depth Data points.
Tip: Remember that Transparent Points are very
useful in low contrast image.
Source Blur
Source Blur can be used to soften a grainy Depth Map and
can eliminate high-contrast speckles which sometimes cause
visible flaws in the Depth Map. It is a useful tool for
visually complex image (patterned shirts, long grass,
crowds), but can lead to loss of fine detail.
Bias XY
Bias X,Y helps to reduce noise outside the border of
objects in an image. It is useful for noisy image where
there is a lot happening in the background behind the main
object . For example, in an image of a baseball player
standing well in front of a crowd, the people are visually
complex with lots of different colors, but they only
require a single flat depth. Bias X,Y can help Photo
Transformer to ignore the background immediately around the
ball player to reduce the bleeding effect.
Adjusting Search Radius
-
Open Tutorial4.TriDefPhoto
This image was Depth Mapped at a Search Radius of 40
and a Tolerance of 40.
-
Open the Depth Window (View menu), and the
Preferences Window, Depth panel (Edit menu).
-
In the Depth window, select the Source checkbox
and drag the slider to 50.
-
In the Preferences window, drag the Search
Radius slider down to about 10, then click
Apply.
You will see a polka dot effect on the Depth Map which
would need many more Depth Data points to correct.
Lower Search Radius values are useful for precision in
complex image.
-
Drag the Search Radius slider up to 100, and click
Apply.
-
Return the Search Radius to 40 and click Apply.
This is a good average setting and will do for most
cases.
Adjusting Search Tolerance
-
Drag the Tolerance slider down to about 10, then click
Apply.
See how lots of holes appear in the Depth Map? You
would need to add several more Depth Data points to
fill them if you used this setting.
Low Tolerance settings are useful where your image is
composed of complex objects (with poor contrast)
overlapping each other at different depths, although
you do need to add more Depth Data points.
-
Drag the Tolerance slider up to about 200, then click
Apply.
The Depth Map becomes much less defined, particularly
around the tyre. To correct this problem you would need
to add a lot of Transparent points to stop the
overflow.
High Tolerance settings are useful where your image is
composed of complex or noisy objects at the same depth,
such as a highly patterned shirt.
-
Return Tolerance to 40 and click Apply.
This is a good average setting and will do for most
cases.
Adjusting Source Blur
-
Drag the Source Blur slider down to 0, then
click Apply
You will see some speckles and noise appear in the
Depth Map (you may need to deselect the Source checkbox
to see tihs clearly). You could correct these errors
with more Depth Data points, but Source Blur does it
for you.
Low Source Blur settings are useful when your image
are composed of well defined, highly contrasted and
detailed objects.
-
Drag the Source Blur slider up to 5, then click
Apply.
The speckles will disappear, but now there may be a
loss of definition. This is because Photo Transformer
can no longer 'see' some of the details in the
image--they have been blurred out of existence--and it
now would be difficult to Depth Map these features
accurately even with a lot of Depth Data and
Transparent points.
High Source Blur settings are useful when your image
are composed of objects which are visually noisy but
have simple outlines.
-
Return Source Blur to 2 and click Apply.
This is a good average setting and will do for most
cases, unless you are 'losing' important low contrast
details. In that case, drop Source Blur to 1 or 0.
Adjusting Bias X,Y
-
Click the Bias X,Y checkbox to clear it, then
click Apply.
You may see some small errors appear on the edges of
objects. You could correct these by careful placement
of extra Depth Data points right at the edge, but
points so close to edges can cause extra spillage and
more corrective action.
It is better to not use Bias X,Y when your image is
composed of simple shapes against a clear background,
as this will allow you to Depth Map with fewer Depth
Data points.
-
Click the Bias X,Y checkbox to select it, then
click Apply.
The little edge errors have been corrected.
Bias X,Y is a useful setting on most types of image. At
worst, leaving it on will only require you to
occasionally add a few extra Depth Data points.
Remember that these settings affect the entire image: you
can't use high Tolerance for one half and low Tolerance for
the other half.
This has only been a brief overview of each settings
behavior in isolation. In practice, these settings can
interact in quite complex ways, so give it a try and let
experience be your guide.
The End
This concludes the current series of Photo Transformer
tutorials. For more information on particular topics,
review the main Help topics. As you become more familiar
with the program, consider using the range of Keyboard
Shortcuts available to speed up your work.
|
|