Photon Transfer Curves

[ngc2403]

That sounds like some good software,

i used 16bit fits that were the the converted raws from the camera

i think that it would be very interesting to see the difference between the colours, though as some of the data you will be using is inferred from the other pixels to make a complete red blue or green image there maybe some complications

[narrowbandpaul]

cant imagine there not being complications...I used luminance

[ngc2403]

Just though i would post my latest work with the camera and Matlab,

the following are graphs from which i have taken the gain, read noise and tried to get the PRNU?

(click to enlarge)

that is the gain on my camera i think the Constant i used here was 0.4?

giving a gain of ~6

using that i got this graph

(click to enlarge)

now i though about the linearity of the chip so here it is

(click to enlarge)

ally

[narrowbandpaul]

they are looking better.... i asume you meant the constant was as in Cx^0.5....C=0.4...then Kadc= 1/0.4^2= 6.25

the going from DN to e- you multiply by e/DN....whixh is the gain. So multiply the DN signal and DN noise by 6.25

what you can do is a gain variance PTC (how gain varies with signal)...ach point that you think shows a good shot noise performance is used to calculate a value of the gain...so in your plot in the upper right hand corner, there are 15 points taht lie close to your straight line.

To find the gain of each point.....gain=signal (DN)/ Noise^2 (DN)

plot all these 15 points vs signal....the points with large sugnal should give a good value for gain...

those nearer the read noise bit willl be a bit wild. But from the good points...take an average and error.....Gain Done!