P3D Re: Flash and 3D

[roberts@xxxxxxxxxxxxxxxxx (John W Roberts)]

>Date: Tue, 19 May 1998 13:38:50 -0600
>From: John Ohrt <johrt@xxxxxxxx>
>Subject: P3D Re: Flash and 3D

>John W Roberts wrote:
>>
>> >The problem with on-camera flash is that light falls off very quickly as
>> >distance increases.  (the inverse square law I think)
>> 
>> Therefore, purely on a mathematical basis and without having actually tried
>> these, it would appear that:
>>  - Using an extended flash (e.g. an umbrella reflector), the light should
>>      fall off less rapidly than with a small flash source. [Inverse
>>      square is the model for a point source.]
>>  - If the flash is powerful enough, moving it back further away from
>>      the subjects should help reduce uneven illumination (provided
>>      the shadow of the camera, etc. doesn't fall on the subject!).

>In effect, the flash source for the umbrella is a point source.  If you
>include the distance from the source to the umbrellas surface in your
>calculations, then the illumination level will obey square law.

>However, the umbrella in effect is one way to achieve the secong method
>of evening the levels of illumination over the scene, use a brigher
>flash placed farther away.

I think that assumes the umbrella is a mirror-like (specular) reflector
(and probably also assumes that the surface of the umbrella is flat).
But all the umbrella reflectors I recall seeing appear to have a grainy
or diffuse surface, to spread out the light, so that it's more or less
accurate to model each point on the surface as a separate point source,
and the entire surface as a diffuse source.

In principle, the brightness from a point source should drop off with
the square of the distance (k / d^2), the brightness from an infinitely
long diffuse line source should be inversely proportional to the distance
(k/d), and the brightness from a diffuse infinite plane source should
not vary with distance (k). A diffuse umbrella reflector is not an infinite
plane, and a some distance it will start showing an inverse square
dropoff, but the "near field" region where this is not a good approximation
should extend out a considerable distance (certainly within the range
of a typical flash).

John R


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