[Bf-committers] Updated Camera patch

Ejner Fergo ejnersan at gmail.com
Mon Mar 14 22:54:52 CET 2011


I loaded up my last camera projection scene, taken with an EOS 50D
(APS-C sensor). Before, I found the correct hfov using the BLenses
script in Blender 2.4x, which I then matched geometry against in 2.5x.
But this would give me a focal length based on the default Blender
sensor size (32x18), and even if the perspective was right*, the focal
length value made no sense, and the workflow was obviously not ideal.

* Originally I made the mistake to match against an image that was not
lens corrected, so even with location data (measuring the environments
length/distances) the geometry shapes and scales became warped...
Using a http://lensfun.berlios.de capable app (rawstudio, ufraw,
darktable, ...) fixed this.

With this updated patch I could select my cameras sensor size and
enter the correct focal length used (18mm) and the match was perfect.
Mathematically it makes sense, but the great part is that we can use
available information (via exif) and not need to "convert" anything.

As for testing, because of the mathematically nature things should
"just work", but please go ahead and test with your own material!

Stuff I haven't tested myself yet, here listed as the modified sources:


source/blender/blenlib/intern/uvproject.c <- Though I haven't tested
it just yet, I'm fairly sure it will work. I just *really* dislike the
way we have to subdivide geometry so much, for the projection to
"stick" properly..!

Also please test addons that may depend on the camera, as well as
importers/exporters (will look at FBX in a bit, but I can't test
Collada presently). I wrote an addon that exports a camera in .chan
format so I could import it into a Nuke camera, and that works
successfully. It still needs some work (import .chan in Blender and
maybe import/export geometry) but I will upload it if/when this camera
patch becomes default.

If you have experience with panorama, please test that as well. Depth
of Field should work, but in my attempt to include
Distance_Affects_FOV I think the way Blender uses DoF is a little...

If you can think of any other way to test this basic camera, please comment!


On Mon, Mar 14, 2011 at 12:51 PM, François T. <francoistarlier at gmail.com> wrote:
> I probably can find a matchmove I did with syntheyes which used a D90 (not
> fullframe backplate) or do another one if this can help ?
> I don't know what all this patch can do, me I'm interest of a match Lens
> value according to the filmgate anyway.
> F.
> 2011/3/14 Troy Sobotka <troy.sobotka at gmail.com>
>> @Mats Holmberg / @Francois T et al:
>> Is there a way we can establish a set of tests to definitively test this
>> patch?
>> Obviously matching a scene to real world units against a test
>> photograph is a possible option here, but I'm wondering if there is
>> something else we can do?
>> Sincerely,
>> TJS
>> _______________________________________________
>> Bf-committers mailing list
>> Bf-committers at blender.org
>> http://lists.blender.org/mailman/listinfo/bf-committers
> --
> ____________________
> François Tarlier
> www.francois-tarlier.com
> www.linkedin.com/in/francoistarlier
> _______________________________________________
> Bf-committers mailing list
> Bf-committers at blender.org
> http://lists.blender.org/mailman/listinfo/bf-committers

More information about the Bf-committers mailing list