Hi Frank,

I would like to know what you did to check the camera exif data with
Realsoft camera object.
This morning I did a test with my camera because I wanted to know, once and
for all, how good the match is.

So I made a setup with a simple box like object. The size of the object was
100x65x70 cm. I put the camera at 900 cm distance and measured every other
variable as good as possible. This time I did it much preciser as with the
couch. Because of the distance I could use maximum zoom. In that case
straight lines remain straight.

I reconstructed the same object and camera viewpoint in RS and loaded the
picture as a backdrop.
The exif data mentioned that focal length was 48,25 mm (or 191 mm in 35mm
film)
When I filled out the 191 mm in the focal length field the 3d object was
quite a bit bigger than the object in the picture. I had to set the slider
back to 160 mm to get a correct match. By tweaking the banking and other
position values a few centimeters the match was even perfect.

But of course I started to wonder about the difference between the 160 and
191 mm. I exported the scene to check in Cinema4D. Cinema4D gave a much
closer match (187 mm). But still not something you can count on. Especially
when distances and objects get bigger. The deviation will grow the same.

I went back to Realsoft to see if I could get a better correlation. I
started to play with the film width and height fields (I never had a look at
them before). When you change them you can see the value of the focal length
change too.

Now comes the most important part:
To be able to use the 191 mm. value, I had to change the film width and
height to 0,036x0,027. In that case I got the perfect match I wanted.
So my question is:
Is this film width and height value a constant factor for my camera? So when
I use this setting can I trust that I can safely fill out the stored focal
length data in RS to get a perfect camera object.

If you don't know the answer I won't blame you ;), maybe somebody else does.

Arjo.

Reply via email to