Le 19/01/2019 à 22.21, jys a écrit :
On Sat, Jan 19, 2019, at 12:55, Bernhard wrote:
If I expose a bit darker (e.g. by reducing the flash output), then white
becomes too dark (<255) without the lowest values moving significantly
closer to black (=0).
Exposing darker by dimming the flash probably isn't what you want. Maybe try stopping
down the lens for the desired black point, and then increasing flash power if needed for
the white? Even then, that setup may simply not be able to create the needed range. You
could try pointing the camera so that one edge of the frame shows the edge of the table
and a truly dark space beyond it. The best way to keep light spill off of the
"black" area is for there to be nothing there for light to hit. ;)
A cheap and easy way to get an a good dark region in a scene is to take
an archive cardboard box with one hole. I can't find a reference now so
I write here my own analysis of why it works so well.
The object needed
Here's a random example of a cardboard box with a hole :
https://www.amazon.co.uk/32-Archive-Storage-A4-8-5-Cardboard-Archivio/dp/B079Y1RVFP
The hole is simply intended to ease drawing the box out of a shelf where
all boxes make a flat surface without any option to grip the box.
Application to your setup
Ensure that the hole is visible somewhere in the scene from the camera.
The RGB data provided by the sensor on the location of the hole is a
better black that most easily producible black pixels.
In practice, since we want the box to be not too big, it is best if it
is physically closed to the camera. Which is probably okay in you setup
because you already adjust focus very close to the camera (so that the
edge on the screen table is blurred).
In your example you would put the box so that the hole is visible on the
rightmost quarter of the image (where bright white already fades).
Why it works
The reason why it works is related to solid angles. Most surfaces are
locally flat and so a point on a surface receives light from roughly
half of all the directions, which makes a solid angle of 2 pi steradians
<https://en.wikipedia.org/wiki/Steradian> (vs 4 pi steradians for the
full sphere).
From a point inside the box, direct light comes only from the hole.
(Indirect light is negligible if the inside of the box is painted black,
but even with unpainted cardboard the result is surprisingly good.)
Viewed from the point inside, that hole is a light source covering up
only a small solid angle.
That small solid angle is a small _fraction_ of the half sphere from
which the light comes when outside the box. Unless you send a light
source directly to the box through the hole, that fraction directly
affects the amount of light received on that point.
Back to photograph's point of view, from outside the box, assuming the
edge of the hole is focused enough to not blur the measurement, measured
RGB values on the sensor, on the area looking at the hole, are
multiplied by that small fraction. Hence, cheap, easy, good black.
I'd be glad to learn how well it worked.
--
Stéphane Gourichon
____________________________________________________________________________
darktable user mailing list
to unsubscribe send a mail to [email protected]
