interior of my friend's intrument repairman shop.
Pictures and output dimensions:
4 shots + 1 zenital, Nikon D100, 8mm 2.8 Fisheye-Nikkor, shooted
in jpg from a good tripod and Agnos Quickly QT head. Images given,
the equirectangular output was set at an optimum resolution of 6000x3000
Enblend 2.3 - commmand line "-v"
Smartblend 1.14 - command line empty
Selfbuilt in a well cooled big tower case:
P5AD2 Asus based PC, 3Ghz Intel 775 dual core, 3Gb Corsair 600Mhz
DDR2 in 2+2 configuration, 2 Seagate SATA 150 Barracuda200 HD, running
It is always important
to declare all conditions and specifics of a test. You may able
to discover errors on it or avoid misunderstandings. If the goal
is to compare the two softwares, what is interesting is not the
absolute speed, but the difference in speed between them and comparision
of the results.
What I meant for speed test was to follow a very standard setup,
5 images from a 6Mpixel camera, one of which is barely used but
has its amount of time consumption, that end up on its optimum size
This is not an F1 race, nor a mathematical test. Time was taken
in seconds on a good USSR mechanical stopwatch, from appearing of
the dos shell of the blending software to its disappearing.
If interested, I
also give the elapse of time PTstitcher takes to process the 5 images
(this is ment to give an idea of the computer speed, and you can
make a comparison to yours).
It is around 25".
A piece of a flare ghost,
present and far larger on one of original images, and a piece
of the head (always a piece in, when using circular fisheyes!).
Please note the change in the floor texture and illumination.
||Little stitching error on the
top! however a very good zenit came out of this stitching.
||Is there something wrong? the
flare present in one of the original images was completely removed,
as all the head pieces.
Usually the top shot is made to save light faloff and poor border
quality. And usually pttools lets it apart. I do not comprehend
how it decides the seams shape. However here smartblend did