Active Topics
-
Is there a section to talk about Java ME? Apps, etc. (6)
to General by Kalatti - 1 day, 10 hrs ago -
Full linux distros on Sailfish OS (253)
to SailfishOS by qoh - 4 days, 4 hrs ago - more...
jetmarc@hotmail.com |
2016-03-07
, 00:54
|
|
Posts: 11 |
Thanked: 50 times |
Joined on Dec 2009
|
#2771
|
Originally Posted by joerg_rw
Today I made a measurement of the flash firing in "photo mode". Target and tools are the same as before, distance approx 2cm (but not identical to yesterday), and the command used was spotread -f
Code:
Code:
Code:
Code:
|
|
2016-03-07
, 03:31
|
|
Posts: 11 |
Thanked: 50 times |
Joined on Dec 2009
|
#2772
|
Originally Posted by wpwrak
(I'm not a light expert either)
I realize that there is a lot to know about light that I didn't even know I didn't know ;-) Of these three, would "CCT" (approx. 5700 K) be the one that corresponds to the CCT parameter of the LED ?
To really compare two light sources, you need to look at the spectrum. But the spectrum is difficult to interpret because of the complex intrinsics of human perception. CCT simplifies this by mapping a spectrum to a single point on a curve, representing daylight. It allows you to compare two spectrums with reasonable confidence (in the context of the curve).
This curve covers only a small subset of all possible colors. Of course, mapping to it introduces an error (the original color is replaced by the most similar daylight color). But since photography flash modules try to imitate daylight anyway, the error is small. Within these limitations, CCT numbers seem to be an easy way to compare flash color.
Don't let the color issue distract you too much. In the end, color is a subjective topic. There should be harmony between camera, flash, driver and usage pattern. The blue decoration on the N900 backcover results in a blue cast on photos and shows that the original design was not perfect either. Imitating it may not be as important as you think. You could pick any popular camera reference design with good reviews and call it done.
|
|
2016-03-08
, 01:41
|
|
Posts: 79 |
Thanked: 719 times |
Joined on May 2014
@ Buenos Aires, Argentina
|
#2773
|
Originally Posted by jetmarc@hotmail.com
Wow, it even measures the flash duration !
Code:Apparent flash duration = 0.347616 seconds
The parameters look quite consistent with the results from "torch" mode, as expected. And yes, these LEDs have manufacturing tolerances in the order of 100s of Kelvin, even after binning, so trying to find a "perfect match" would be futile for this reason alone.
Thanks a lot for the great measurements !
- Werner
 |
|
2016-03-08
, 09:54
|
|
Posts: 6,445 |
Thanked: 20,981 times |
Joined on Sep 2012
@ UK
|
#2774
|
Originally Posted by wpwrak
Indeed. Is it just me or is that figure two orders of magnitude higher than expected?
Wow, it even measures the flash duration !
__________________
Русский военный корабль, иди нахуй!
Русский военный корабль, иди нахуй!
|
|
2016-03-08
, 13:33
|
|
Posts: 11 |
Thanked: 50 times |
Joined on Dec 2009
|
#2775
|
Originally Posted by pichlo
Dont trust the duration figure. I recall that the N900 flash lighted up twice, although I had the shutter button depressed half-way through before starting the measurement and despite having selected "always" rather than "red-eye" setting.
Indeed. Is it just me or is that figure two orders of magnitude higher than expected?
The documentation sais that argyll records many individual measurements for as long as the button is pressed. You are supposed to press the button, then fire the flash, and then release the button. Afterwards the software looks through the recorded data to find the flash, and discards the rest.
In normal patch-read mode (for printer profiles), the instrument can do 200 samples per second. For lack of better information, I guess that a similar rate was used.
I dont know how argyll handles two light peaks. It could have measured the whole time covered by the two events, or concatenated them, or discarded one of them, or something else. I used version 1.8.3 if you want to dig into the source code and find out.
You could also video a N900 flash and analize the time sequence with virtualdub. I guess that it is easy to figure out where this 300+ ms duration comes from, and maybe also how to avoid it for future measurements.
|
|
2016-03-08
, 14:26
|
|
Posts: 6,445 |
Thanked: 20,981 times |
Joined on Sep 2012
@ UK
|
#2776
|
The simplest method might be what I use for long exposure photos to avoid handshake interference (in your case any manual interference). Set a 2 or even 10 seconds shutter delay, press the button, release, let go.
__________________
Русский военный корабль, иди нахуй!
Русский военный корабль, иди нахуй!
|
|
2016-03-08
, 16:02
|
|
Posts: 72 |
Thanked: 184 times |
Joined on Apr 2011
@ Germany
|
#2777
|
Originally Posted by pichlo
No, they values quoted on the Wikipedia page are for conventional xenon/gas discharge flashes. LED flashes have a much longer duration - try photographing a moving subject, you will see that they're basically unable to freeze motion and are only sufficient for illumination purposes.
Indeed. Is it just me or is that figure two orders of magnitude higher than expected?
EDIT: I just took a video at 240fps of an LED flash (sony xperia, as my n900 is sitting somewhere in a drawer). I counted ~36 frames, so ~150ms for the main flash duration.
Last edited by Oblomow; 2016-03-08 at 16:44.
|
|
2016-03-08
, 16:38
|
|
Posts: 72 |
Thanked: 184 times |
Joined on Apr 2011
@ Germany
|
#2779
|
Originally Posted by pichlo
Well, they're just working fundamentally different. In short, you would destroy the LED with the very high current needed for such a short pulse (100x shorter flash at same integral intensity means 100x the current). In a Xenon flash, the light-emitting medium is the gas, so the energy density (per volume) is much lower than in an LED, and so heat dissipation is less of a problem.
Any idea why is that?
 |
|
2016-03-11
, 03:32
|
|
Posts: 2,222 |
Thanked: 12,651 times |
Joined on Mar 2010
@ SOL 3
|
#2780
|
Originally Posted by wpwrak
https://www.youtube.com/watch?v=4qgPhNHJPB8
I did a quick search for material supporting that LED current would directly affect color, but didn't find anything. However, I did find that junction temperature has a significant effect on color:
https://www.youtube.com/watch?v=OZCm-NMC9M4
https://youtu.be/FcG7ZkV12Sw?t=59
Obviously the opposite effect (heating up a 100K junction temperature) also applies during high loads. With opposite effects on color temperature.
I've seen green LEDs turning orange during burnout from overload.
Last edited by joerg_rw; 2016-03-11 at 03:48.
| The Following 5 Users Say Thank You to joerg_rw For This Useful Post: | ||
