Re: question on flats


Jun 12, 2006

 


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

#24597 Jun 12, 2006

michaelwwinslow wrote: >

> some of my images have residual marks which show up after calibration,

> and some stretching

>

> is there a way to adjust the strength of flat processing?



Not per se, but adding or subtracting a constant from every pixel will have the

effect of changing the "strength" of the flat-field... to a degree.



More to the point, you might want to make sure that bias/dark is accurately

subtracted from both the flat and the light frame.



I one camera we have here, it looks suspiciously like the bias level changes as

a function of ambient (camera case) temperature. I haven't done careful tests

to prove this, but it really wouldn't be all that surprising. There is no

regulation of the temperature of the readout electronics and A/D converter.

Integrated circuit manufacturers go to great lengths to create voltage

references that are stable over temperature, but not everyone buys the most

expensive kind... and even if they did, other parts might drift over temperature

as well.



Doug



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



Doug George

dgeorge@...



Diffraction Limited

Makers of Cyanogen Imaging Products

www.cyanogen.com



25 Conover Street

Ottawa, Ontario,

Canada, K2G 4C3



Phone: (613) 225-2732

Fax: (613) 225-9688



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



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

#24607 Jun 13, 2006

sorry about the duplicate post.. yahoo was strange yesterday, and

the first post didnt show up..



I'm assuming that Maxim determines an area of uniform response at

the center of the chip, and determines the average value of this

area, then normalizes the individual pixel values against this value.



The resultant is multiplied by the pixel value of the (image - dark)

x,y of the image..



Or close to that anyway..



So this raises a couple questions..



Since I'm using an Mx25, I use the boxcar filter, and that

successfully keeps the color information, by some sort of magic.



(leave that question hanging for a second)



The area of uniform response - The area of greatest illumination

appears to be at the center of the image.



Does Maxim use the center of the image for this? or does it check

for the area of best uniformity or other?



Also, how is the size of the area used for the average determined?

Is it circular or rectangular?



What would be the net effect of having dust/dirt in the critical

area?



The above question leads back to your earlier answer, of adding a

constant value, which leads to the hanging question-



how to offset correct a flat which contains color information?



Thanks Doug!



--- In MaxImDL@yahoogroups.com, "Douglas B. George" dgeorge@...>

wrote: >

> michaelwwinslow wrote:

> >

> > some of my images have residual marks which show up after

calibration, > > and some stretching

> >

> > is there a way to adjust the strength of flat processing?

>

> Not per se, but adding or subtracting a constant from every pixel

will have the > effect of changing the "strength" of the flat-field... to a degree.

>

> More to the point, you might want to make sure that bias/dark is

accurately > subtracted from both the flat and the light frame.

>

> I one camera we have here, it looks suspiciously like the bias

level changes as > a function of ambient (camera case) temperature. I haven't done

careful tests > to prove this, but it really wouldn't be all that surprising.

There is no > regulation of the temperature of the readout electronics and A/D

converter. > Integrated circuit manufacturers go to great lengths to create

voltage > references that are stable over temperature, but not everyone buys

the most > expensive kind... and even if they did, other parts might drift

over temperature > as well.

>

> Doug

>

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

>

> Doug George

> dgeorge@...

>

> Diffraction Limited

> Makers of Cyanogen Imaging Products

> www.cyanogen.com

>

> 25 Conover Street

> Ottawa, Ontario,

> Canada, K2G 4C3

>

> Phone: (613) 225-2732

> Fax: (613) 225-9688

>

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

>



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

#24609 Jun 13, 2006

michaelwwinslow wrote:

> Since I'm using an Mx25, I use the boxcar filter, and that

> successfully keeps the color information, by some sort of magic.



No magic. The boxcar filter removes the Bayer matrix.

> The area of uniform response - The area of greatest illumination

> appears to be at the center of the image.

>

> Does Maxim use the center of the image for this? or does it check

> for the area of best uniformity or other?



It uses the center 1/4 of the array.

> Also, how is the size of the area used for the average determined?

> Is it circular or rectangular?



Take a rectangle half the image dimensions, and center it.

> What would be the net effect of having dust/dirt in the critical

> area?



Not much. First of all, the effect would be inconsequential because it would

contribute only a small number of pixels. Also dust donuts are never black;

they're just slightly lower than the surroundings. Only a contrast stretch

makes them look so big and ugly. Even if they did cause a very slight

underestimation of the average brightness, this would only increase the overall

brightness of the image by a trivial amount.

> The above question leads back to your earlier answer, of adding a

> constant value, which leads to the hanging question-



No, that's an entirely different problem. A slightly different scaling factor

will not change the effect of the flat field; it will just make the whole image

very slightly brighter or dimmer.

> how to offset correct a flat which contains color information?



Subtract bias/dark.



Doug



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



Doug George

dgeorge@...



Diffraction Limited

Makers of Cyanogen Imaging Products

www.cyanogen.com



25 Conover Street

Ottawa, Ontario,

Canada, K2G 4C3



Phone: (613) 225-2732

Fax: (613) 225-9688



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







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

#24611 Jun 13, 2006

Thanks, I'll play a bit..



The center 1/4th might be the problem, since I see some tapering off

of the illumination in a circular area, but the gradient appears to

be shy of a 1/4th rectangle..



I'm not sure if there is value in asking you to allow a user tweak

for the area used for uniformity.. I wonder if the idea is patent

worthy? I havent heard it discussed in public before (that just

changed.. :) )



Also, a circular area would make more sense considering the shape of

most optics.



I might be able to achieve the right results with the offset as

suggested.



I do appreciate your help as always!



--- In MaxImDL@yahoogroups.com, "Douglas B. George" dgeorge@...>

wrote: >

> michaelwwinslow wrote:

>

> > Since I'm using an Mx25, I use the boxcar filter, and that

> > successfully keeps the color information, by some sort of magic.

>

> No magic. The boxcar filter removes the Bayer matrix.

>

> > The area of uniform response - The area of greatest illumination

> > appears to be at the center of the image.

> >

> > Does Maxim use the center of the image for this? or does it

check > > for the area of best uniformity or other?

>

> It uses the center 1/4 of the array.

>

> > Also, how is the size of the area used for the average

determined? > > Is it circular or rectangular?

>

> Take a rectangle half the image dimensions, and center it.

>

> > What would be the net effect of having dust/dirt in the critical

> > area?

>

> Not much. First of all, the effect would be inconsequential

because it would > contribute only a small number of pixels. Also dust donuts are

never black; > they're just slightly lower than the surroundings. Only a

contrast stretch > makes them look so big and ugly. Even if they did cause a very

slight > underestimation of the average brightness, this would only

increase the overall > brightness of the image by a trivial amount.

>

> > The above question leads back to your earlier answer, of adding

a > > constant value, which leads to the hanging question-

>

> No, that's an entirely different problem. A slightly different

scaling factor > will not change the effect of the flat field; it will just make

the whole image > very slightly brighter or dimmer.

>

> > how to offset correct a flat which contains color information?

>

> Subtract bias/dark.

>

> Doug

>

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

>

> Doug George

> dgeorge@...

>

> Diffraction Limited

> Makers of Cyanogen Imaging Products

> www.cyanogen.com

>

> 25 Conover Street

> Ottawa, Ontario,

> Canada, K2G 4C3

>

> Phone: (613) 225-2732

> Fax: (613) 225-9688

>

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

>



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

#24612 Jun 13, 2006

michaelwwinslow wrote: >

> The center 1/4th might be the problem, since I see some tapering off

> of the illumination in a circular area, but the gradient appears to

> be shy of a 1/4th rectangle..



As I stated previously, it DOES NOT MATTER. All that is going to do is make the

final image very slightly brighter overall. It does NOT affect whether dust

donuts or vignetting "leak through" the flat fielding.



Any bias offset will cause this "leak through", as will any light leaks you had

while taking either the flat-field or light image. A light leak could result in

an apparent bias offset; i.e. non-focused light spread over the entire detector.

It's a common problem when trying to do flats because, unless you're doing sky

flats, you have a bright light source in the observatory.

> I'm not sure if there is value in asking you to allow a user tweak

> for the area used for uniformity.. I wonder if the idea is patent

> worthy? I havent heard it discussed in public before (that just

> changed.. :) )



There is no advantage whatsoever to doing that. If your scaling factor comes

out 1% too high, all that happens is that every single pixel in the image gets

multiplied by 1.01. It's just as if you did a very slight screen stretch on the

image.



Doug



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



Doug George

dgeorge@...



Diffraction Limited

Makers of Cyanogen Imaging Products

www.cyanogen.com



25 Conover Street

Ottawa, Ontario,

Canada, K2G 4C3



Phone: (613) 225-2732

Fax: (613) 225-9688



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



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

#24618 Jun 13, 2006

FSQ, and an SX-M25..



Bias is now confusing me I guess.



If I look at bias, flat and image I get on the order of



1100 as the minimum pixel level for Bias

28,000 for a .5 second flat

and 19,000 on a 20 minute exposure



Does that mean anything or am I getting myself confused?!



I did like the results of (flat-bias) +750 as a flat a bit better

than the 'normal' calibration.. It didnt do too much for the 'glob"

of dirt which apparently above 778,934 (seems to be on the

protective cover on the chip)

--- In MaxImDL@yahoogroups.com, "Chris Peterson" clp@...> wrote:

>

> I've observed this when imaging with an SCT that has any dust on

its

> corrector during times when moonlight can fall on that corrector.

The dust

> scatters a small amount of light into the OTA and onto the sensor,

creating

> an artificial bias on the science images that doesn't match the

flats. The

> result is a slightly flawed calibration, with some vignetting and

dust

> shadows that don't fully disappear. I've also gotten bad twilight

flats for

> the same reason- scatter off the corrector.

>

> I didn't notice what equipment Michael is using, but it is

important to be

> scrupulous about keeping an SCT corrector clean when imaging. Even

a tiny

> bit of dust is enough to be problematic.

>

> Chris

>

> *****************************************

> Chris L Peterson

> Cloudbait Observatory

> www.cloudbait.com

>

>

> ----- Original Message -----

> From: "Douglas B. George" dgeorge@...>

> To: MaxImDL@yahoogroups.com>

> Sent: Tuesday, June 13, 2006 11:05 AM

> Subject: Re: [MaxImDL] Re: question on flats

>

>

> Any bias offset will cause this "leak through", as will any light

leaks you

> had

> while taking either the flat-field or light image. A light leak

could

> result in

> an apparent bias offset; i.e. non-focused light spread over the

entire

> detector.

> It's a common problem when trying to do flats because, unless

you're doing

> sky

> flats, you have a bright light source in the observatory.

>







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

#24623 Jun 13, 2006

Doug, I have to say that you are amongst the best at supporting your

product.. Definitly, cutting edge, and I have astro-art, AIP4WinV2

as well as PS-CSII as tools :)



I had the time to play a bit with this today, I appreciate your help

greatly.



so I did some comparisons



a) 1 flat, 3 bias 4 20 minute subs SD mask combined

b) 1 flat, bias subtracted, offset added - 4 20 sd

c) 3 flat 3 bias 3 dark - 4 20 SD

d) 3 flat 3 bias no dark 4 20 SD



B is better than A



C and D are better than B, but they are indestinguishable from each

other..



yeah.. clean the chip is a good idea.. scary, no "revert"

or "healing brush" with physical things..



As you said, playing with an offset gives me some remedy when I cant

go back and do more flats..



I really dont see any significant difference with darks on this

camera.. ( I sense that the non-SX users have a hard time believing

this)



Contact Us
This Site's Privacy Policy
Google's privacy policies

S
e
n
i
o
r
T
u
b
e
.
o
r
g