"Bill Pragnell" <bil### [at] hotmailcom> wrote:
> What ho
>
> I've been trying to make some HDR light probes over the last few weeks. It's
> tricky. :) The "stainless steel" balls I got from a garden centre were not up
> to scratch... much deformation was apparent on the surfaces. They weren't even
> stainless steel - plastic with a chrome coating probably, and even the
> strongest permanent magnet I could lay hands on failed to grip the surface more
> than fractionally.
>
> So, I purchased a 60mm ball bearing from ebay instead, and this seems to work
> very well. It has some slight scuff-marks which I can no doubt polish out in
> time, but otherwise gives very sharp reflections.

Your ball bearing is about 2.3 times bigger than mine, so it should be able to
give much better images. How much did you pay for it? I'm don't know where to
look for bigger ball bearings where I live...

My one inch ball bearing occupies about 850 pixels square with my 8 megapixel
camera (Canon IXUS 860). You should be getting 2000 square pixels or so?

Make sure you take the photo at maximum zoom to reduce the distortion of the
ball - this will help in stitching the two images together.

From what I can tell, the combination of magnification and bright light sources
(the sun or point-like interior lights) brings out even the smallest
imperfections in the polishing. I got my ball chrome plated, and that made a
huge difference. Almost no flare from the sun from the ball bearing surface
(but still a little from the cheaper lens on the point and shoot camera).

One last point - my ball bearings (I got three) started rusting after a couple
of months of use. I'm not sure exactly what your one is made of, but don't
touch it with your fingers, and clean it with a lint free cloth whenever you
use it or store it. I'm hoping the chrome plating on my third one will mean it
will now never rust...

> My camera seems adequate to the task, although it would have been very difficult
> without the CHDK upgrade to the Canon firmware (thanks to tgq for the link). I
> have two tripods, which I have 'calibrated' for HDR-gathering (even drawing a
> line round the neck of the larger one to mark the elevation!), and a great
> little plastic tub-thingy which sits on the larger tripod and supports the ball
> nicely.

CHDK is an absolute marvel :-) I used the custom grid function to draw a circle
and crosshairs that precisely matches the position of the ball on screen to aid
in getting it at exactly the right distance. I also used the "override focus"
setting to force the camera to focus on the minimum distance at maximum zoom.
Getting the ball in perfect focus had been a real problem up until that.

> However, I am having a few problems, particularly with image alignment. I'm
> using HDRShop for the whole process, following the tutorial here:
>
> http://gl.ict.usc.edu/HDRShop/tutorial/tutorial5.html
>
> It works a treat, but often I find that the rotated image doesn't quite match
> the other image after unwrapping. The points I select in the editor match
> perfectly, naturally, but the images seem to diverge near the edges. I'm not
> entirely sure why this is happening, but making sure the ball is centred in the
> frame when taking the pictures seems to help considerably. Can any resident
> HDR-makers shed any light on this?

I use HDR shop to convert the images from spherical mirror projection (i.e. the
HDR photograph) into Latitude/Longitude format, then do the stitching in
Photoshop. I think there is an HDR version of GIMP - Cinepaint? Everything is
much simpler to do in square lat/long format, and POV can use the resulting
images just fine.

Having the two images taken at max zoom will help, as will making sure that they
are taken from exactly the same elevation and that the camera is absolutely
level. I find I have to do a little stretching and fudging in Photoshop even
when I'm really careful about that stuff - it's just a case of minimising the
work required.

> In any case, here's my test scene so far, using the probe I made at the weekend.
> I don't have a light-dome setup so I've cheated a little by using a single
> light-source to account for the sun.

Looks pretty great already - I'm really jealous of the extra resolution you are
able to get!

> Bill

Cheers,
Edouard.

Post a reply to this message