Need help: "correcting" exposures in time-lapse imagesets

Hi, my son and I are doing a science project in which we're measuring dye
transport
in plant leaves over time. Using a Canon Powershot A40 and Canon's
RemoteCapture app to
snap images every 60 seconds.

Unfortunately, through RemoteCapture you can set film speed, but shutter
speed
remains in Auto mode. In order to digitally compare each image with the
initial
reference image (to determine the amount of dye absorbed), the exposures
need
to be the same each time.

Is there a way to compensate for exposure time in a relatively accurate way,
say,
see each image as if it were shot in 1/20 sec? I wish I could use the CHDK
package,
but the A40 doesn't seem to be supported.

Regards,
kinata

Quoting kinata <forums@gimpusers.com>:

> Hi, my son and I are doing a science project in which we're measuring dye
> transport in plant leaves over time. Using a Canon Powershot A40 and
> Canon's RemoteCapture app to snap images every 60 seconds.
>
> Unfortunately, through RemoteCapture you can set film speed, but shutter
> speed remains in Auto mode. In order to digitally compare each image
> with the initial reference image (to determine the amount of dye
> absorbed), the exposures need to be the same each time.

Is flash mode available in remote capture? If so, perhaps one could be
used to obtain a consistent exposure (no AE, and the "shutter speed"
effectively is determined by the flash).

> Is there a way to compensate for exposure time in a relatively accurate way,
> say, see each image as if it were shot in 1/20 sec? I wish I could use
> the CHDK package, but the A40 doesn't seem to be supported.

Perhaps you could include a test pattern in the frame (ie, place a
picture of a test pattern next to the plant) and be able to match
exposure settings by adjusting the levels of the images such that the
two test patterns match (if this works, it could likely be automated).

Saul, thank you, sir...replies follow.

>Is flash mode available in remote capture? If so, perhaps one could be
>used to obtain a consistent exposure (no AE, and the "shutter speed"
>effectively is determined by the flash).

Well, our brainstorm was to use a fluorescent dye,
without using a flash. The exposure times range
from 1 sec to 1/20 sec, ISO 200, f2.8. This
wouldn't represent a showstopper, as we're
concentrating on the area of each image that
contains pixels having that color of dye. So I
just need to iterate through the pixels and count
those having a certain threshold of a defined RGB
value to get the area of the image containing the
dye.

>> Is there a way to compensate for exposure time in a relatively accurate way,
>> say, see each image as if it were shot in 1/20 sec? I wish I could use
>> the CHDK package, but the A40 doesn't seem to be supported.
>
>Perhaps you could include a test pattern in the frame (ie, place a
>picture of a test pattern next to the plant) and be able to match
>exposure settings by adjusting the levels of the images such that the
>two test patterns match (if this works, it could likely be automated).

Now I reveal my true ignorance. I was thinking
that I could duplicate a layer containing a source
image, play with its opacity, and then merge
the layer back into the original source image to
arrive at a target "exposure". Might that work?

In a related issue posted elsewhere in this forum,
I'm wanting to isolate the dye part of the image:
-- convert the RGB triple for the dye color
(approximately a cyan) to HSB (HSV?)
colorspace,
-- rotate it to (say) green, and then
-- extract the green layer of the image.
This last is probably off-topic...just being
conversational.

Regards,
Kinata