Creating Cube-Maps - Real World Photography

Creating Cube-Maps - Real World Photography
To create a cubemap from
photography, you will need :
1. A set of images (preferably very
high resolution) shot in a way to
cover a full 360° panorama from
a single point of view.
2. A 360° image built from stitching
the set from step 1 together usually a spherical, cyndrilical,
or equirectangular image.
3. Software
that
converts
combined 360° image from
step 2 into the cubemap format.
When combined, this set of images provide a 360° viewing angle of Bangkok. Note that the exposure is quite different
between images and that stitching lines are very apparent. For many images, this can be cleaned up automatically,
but the best cubemaps are created by those willing to provide a lot of high-quality RAW format HDR images and put
in the time to manually blend exposures between images and remove stitchlines that automatic solutions might miss.
Once you have your set of images,
assemble them using the appropriate
stitching method. There are many
pieces of software that will auto-stitch
photos together into a single 360°
image.
The best free solution for for fully
automated
image
stitching
is
Microsoft’s Image Composite Editor.
However, If you prefer more manual
control, you can also do this in
any recent version of Photoshop
using the “Photomerge”
action
(File>Automate>Photomerge...)and
then adjust in the assembled file as
needed.
This 360° Image of coral reef near Heron Island is a equiractangular spherical panorama, meaning if you wrapped
this rectangular image around the interior geometric faces of a sphere and viewed it from the inside, it would then
appear undistorted.
After building the combined image,
you then need to pass the file into
another piece of software that will
convert it into a cubemap, which is an
image format that provides 6 images
for each of a cube’s faces. When the
6 images of a cubemap are applied to
the proper interior faces of a cube and
viewed from the inside, the image will
appear seamless and undistorted.
Then in the Output column, click the
“New Output Format” dropdown and
make sure it’s on the “Transformation”
option.
In the Settings tab, change the “Type”
dropdown to “Cube Faces,” change
the “Face Names” dropdown to
“‘front’..’down’” and choose an image
size of at least 2048x2048.
Note
that
this
resolution
is
The most versatile software package recommended for each face to
for doing this is Pano2VR. Pano2VR provide maximum pixel density in
will allow for many input formats and Gear VR.
will export out each of the 6 images
needed to give you a full cubemap.
Under the “Output” section at the
In Pano2VR, in the Input column on bottom on the “Format” dropdown,
the left, click “Select Input” and match choose PNG (.png) to ensure minimal
the input format to the combined compression artifacts.
360° stitched image file created in the
previous steps.
Note that this image is far too low resolution / quality to be acceptable for
the cubemap format in VR. Remember, whatever resolution a panorama is,
it will be stretched across a 360 degree view, so always provide as high a
quality set of starting images as possible.
Creating Cube-Maps - Computer Generated Imagery
Creating a cubemap from computer
generated images (CGI) is much more
succinct in procedure explaination
than with photography due to no
stitch lines and exposure being
uniform thanks to CG cameras not
being bound by physical properties of
lenses and image sensors.
Most CG shops should have no
problem rendering out a cubemap
from various professional software
solutions, but here is an example
anyway generated from 3DS Max :
Up
Right
Front
Left
Down
Back
V-Ray’s box camera option was used
to render and export this format
directly out from 3DS Max without
having to use any additional tools.
Note that the benefits of using pure
CGI for generating cubemaps aren’t
just limited to uniform exposure and
elimination of stitch lines, but that it
allows also for additional viewpoint
possiblities,
re-exportation
of
infinintely higher resolution cube faces
for future use in higher pixel density
display technology, exportation for
use on different geometry besides
that used for cube-maps, and even
stereoscopic rendering.