P3D Re: 3D realism and focus cues

["Greg Wageman" <gjw@xxxxxxxxxx>]

From: John W Roberts <roberts@xxxxxxxxxxxxxxxxx>


>If all scenes are bright-sunlight scenes, and if everything is kept pretty far
>away from the camera, then assuming no change in focus with distance is a
>pretty good approximation. But those conditions do not always hold.


Yes, I agree there conditions under which the depth of focus of the eye is
insufficient to keep everything sharp; I notice this particular when viewing
things very close to me.

But, consider the depth of focus scale on our stereo cameras.  At f/16, the
hyperfocal table on the Revere indicates that the image is sharp from about 3
feet to infinity.  Of course this range is dependent upon the size of the circle
of confusion chosen, and is dependent on the lens design, and the focal length
of the eye is somewhat variable, so the numbers don't directly translate to that
of the eye...  I wonder if any of the list members have data on the equivalent
f-number range of the average human eye, and what the approximate depth of focus
is?  I suspect that for the "average" stereophotograph, with objects no closer
than 7 feet and extending to infinity, that the eye is quite capable of keeping
all of that in focus in daylight conditions.

>Another possible issue - depth of focus as a function of brightness of
>illumination of the scene [Note]. (Though if you're looking at the depth
>that was in focus in the original photograph, it might match up.)


Interesting.  A dimly-lit scene that is in sharp focus might look less 'rea'
because had we actually been there, it wouldn't have been sharp front-to-back.
Certainly possible.

>[Note: In some of the old 2D movies that are broadcast on television, there
>are certain scenes which appear to be intended to depict outdoor nighttime
>scenes (cues from story line and darkness of the scene), but which appear
>very likely to be daytime scenes (sharp vertical shadows, maybe even cumulus
>clouds), shot with a dark filter in front of the lens. They look so
>unconvincing by modern standards as nighttime scenes, I'm not even sure
>that's what they were intended to be. Have others noticed this effect in
>old movies? I wonder what cinematic techniques now in use will be chuckled at
>50 years from now. :-) ]


Part of the issue here is videotape transfers.  Videotape encodes a much smaller
dynamic (luminance) range than film.  For example, have a look at most video
transfers of "Return of the Jedi".  During scenes with head-shots of the
Emperor, an annoying black blob is visible just adjacent to his eyes.  It seems
that someone didn't like the way the hood looked on him, and somewhere in
post-production, they "rotoscoped" (probably digitally, but the term still
applies) in a dark area.  Obviously, they used a much "blacker black" than that
of the adjacent area.  This most likely didn't show up on film (I can't imagine
them missing it and/or allowing it), but when transferred to video, the
re-mapping of luminance levels caused it to be visible.  This is true of may of
the special effects shots; the masks used on many of the TIE fighter ships in
the composite shots are visible on video, where they weren't on film.  I suspect
the same is true for these "night shots"; they looked fine on film, but suffer
from the video transfer.

     -Greg W. (gjw@xxxxxxxxxx)





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