Re: Northern California in 3d

[P3D Larry Berlin <lberlin@xxxxxxxxx>]

>Date: Tue, 23 Sep 1997
>From: P3D  Adam L. Beckerman 
>
>In response to some comments I received about my first stereo images:
>
>1.  I have drastically reduced the size of the thumbnails so that they load 
>faster (rather than the server resizing the original image to be a thumbnail, 
>it is now the physical size of the thumbnail).

******  Usually the better option so download times are improved. The other
way causes the full image to download.

>.......................
>4.  Since I seem to only be able to view wide-eyed, and not cross-eyed (for 
>most images), can anyone else attest to the fact that Gabriel notes:
>>One minor point, some of the L-R-L pictures were actually L-R-R.<
>I find this hard to believe considering the way I went about composing the
L-R-L 
>pictures.  If anything, they _could_ be R-L-R, but I find it hard to
believe that 
>they are L-R-R.  Can anyone else comment??  If some ARE L-R-R, which ones?

*****  None of the images on your site today are LRR. You don't have to be
able to view in cross-eyed mode to tell what's going on in the images. If
you view wide-eyed, then the pair on the left will appear normal and the
pair on the right should appear pseudoscopic. If by mistake you got two RR
instead of the intended RL, then it would appear flat and have no depth
either direction. It's good practice if you aren't familiar with these
things, to intentionally put together wrong combinations and see what they
look like in whatever viewing method you use. Then you can easily recognize
when mistakes happen.

>
>5.  ..................  However, as is apparent, it is also a 
>good example of what can get lost when digitizing images.  I tried to play
with 
>the brightness and contrast settings but to no avail.  If you look hard,
you can 
>pick the deer out of the center of the frame.  If this is simply not worth
displaying 
>b/c the deer is lost, I'd happily consider removing it.

*****  Not much is lost due to digitizing in this image. The deer is just a
very low contrast set of markings almost lost amidst the leaves. When it's
noticed it does jump out. A bit of selective brushing or contrast work just
around the deer itself might help it stand out better. It would have helped
if the deer was standing in the sun instead of shadows, but try telling that
to the deer. ;-)

One possible change would be to crop a lot of the image away so that the
deer is a larger percentage of the total image being presented. You can scan
a smaller area of the original at a good resolution so the deer is the main
part of the image. Eliminating much of the foreground would help bring out
the dear deer. The current image has several lines of red pixels at the
bottom that need trimmed anyway. Basically a great shot.

>
>6.  Dan mentions that by exaggerating the effect so that it is slightly
hyper often 
>enhances digital images.  When creating them in Photoshop, I previewed them 
>and created the best effect possible (be it hyper or normal or even hypo
(although 
>I don't know if there actually are any hypos)) IMHO.
>
>7.  Some of my scenics may look flat, but it isn't b/c of the way the
'mounting' 
>was done, it's more b/c I ran out of depth range (I had subjects in the
foreground 
>in decent 3D, but the background was so far away, it flattens out to a
backdrop).

*****  Both comment 6 and 7 relate to the fact of your use of a Realist
camera for the images. They are all approximately Orthos. The camera has a
fixed base and is best for subjects that are relatively close. Subject
matter at any significant distance ends up rather flat especially in scanned
versions. The only way to combat that is with separate cameras where you can
increase the base, or by using the weight shift method with the Realist and
taking two pairs of each scene. No amount of image filtering will increase
the depth factors. You could increase stereo parallax by hand but it takes a
bit of knowledge and lots of time. This is where you learn what a camera
will and will not do and plan your shots accordingly.

Muir Woods is a beautiful image but could have used a bit more foreground
content. That's true of other scenics where the fixed base makes the
distance pretty flat. In several of the images, just squatting down and
positioning rocks and other things in the closer foreground would have added
useful interest.

>With regard to this pink light leak streak on the right view of the 
>Olmstead Point image, I'm having a tough time seeing it.  I switched 
>back and forth between the left and right, and only noticed some light 
>density changes in the corners (maybe to do with vignetting), but no 
>obvious pink streak.  This has been pointed out by a few people.  Is 
>it in the top right??  Or somewhere else?

It's running vertically through the whole image (Olmstead point) but is
strongest in the upper half. It looks like a streak in the film perhaps
caused by heating or maybe a light leak. It appears only in the R image near
the right edge, so creates retinal rivalry for stereo viewing. It's hard to
fix, but can be improved somewhat by selecting the discolored area,
feathering the edge and color correcting the selection. Do it on a duplicate
image so you can discard any unwanted messes that might result before you
get it right.

A great job overall. Keep up the good work.

Larry Berlin

Email: lberlin@xxxxxxxxx
http://www.sonic.net/~lberlin/
http://3dzine.simplenet.com/


------------------------------

End of PHOTO-3D Digest 2307
***************************
***************************
 Trouble? Send e-mail to 
 wier@xxxxxxxxxxxxxxxxxxxxxxxx 
 To unsubscribe select one of the following,
 place it in the BODY of a message and send it to:
 listserv@xxxxxxxxxxxxxxxxxxxxxxxx 
   unsubscribe photo-3d
   unsubscribe sell-3d
   unsubscribe overland-trails
   unsubscribe icom
 ***************************