Hi All, I've read a lot of posts on flats here. I went to do some this morning putting the DSLR into the scope pointed through a lighted window and stretched new white T-shirt material over the front of the scope and defocussed.
Now I've tried every exposure from 1/4 sec to 1/8000th sec and every single one has a narrow spike that goes off the scale ...now I know I'm meant to be aiming for 1/3rd pixel saturation. What's wierd is that the frequency of the spike shifts across the entire spectrum range as I change the exposure time. I can't work out what's going on. If I take the t-shirt off I get a normal histogram well spread out and not totally saturating the histogram. I'm stumped. HELP!
Robert can you post a screen capture of the histogram?
What mode and ISO are you shooting in ?
I use P mode at ISO 100 and this usually adjusts the exposure to give me a histogram like the one attached.
I do dawn sky flats.
Quote:
What's wierd is that the frequency of the spike shifts across the entire spectrum range as I change the exposure time. I can't work out what's going on. If I take the t-shirt off I get a normal histogram well spread out and not totally saturating the histogram. I'm stumped. HELP!
This sounds logical, because you are changing the exposure time so the histogram will shift up or down.
And when you take the t-shirt off, it will give you a full histogram, (whatever colours it sees, full spectrum).
Robert can you post a screen capture of the histogram?
What mode and ISO are you shooting in ?
I use P mode at ISO 100 and this usually adjusts the exposure to give me a histogram like the one attached.
Hi RB, posting a histo is stretching my tech capabilities, but yeh it looks like that if I set it to RGB histo, though the width gets narrower as the exposure duration shortens and mine seem to saturate more even at the shortest exposures. I was using Tv, so I'll try P - ISO100 and ISO400. But doesn't the peak of that histo of yours go off the saturation scale too?... I'm confused... I though I was looking for a broader lower (1/3rd height) histograms
Ah now I understand what you mean by saturation-off scale.
No you do not worry about how high the spike is, it's irrelevant.
We only are concerned with the x-axis, this is a scale from 0-255 and is the brightness value of each pixel.
Since there is 8 bits per pixel we have 256 levels and since we have 3 channels this gives 256 x 256 x 256 combinations of levels hence 16.8 million colours.
The height of the scale (y-axis) only represents how many pixels are actually at that brightness level so it could be very high or off scale since it's only a graphical representation and scaled to fit the chart.
When I was using my DSLR camera for astrophotography and before I was going to take flats I just set it to AV(aperture priority) mode and the camera picked optimal exposure length up by itself.
And I guess you are aware that flats should be taken with your camera being at exact position as it was when you took your light frames. If you remove your camera and attach it back again there is a big change that all dust on a camera sensor and a telescope lens/mirror that are expected to be removed after applying flats wont match with the same dust in your light frames.
Thanks Andrew, I tried again with P-setting and ISO 100...but this time with white t-shirt and pointed a blue sky away from the sun.... I now have a blue tint and separation on the blue, green and red channels?
Also it's desirable to have the spike as narrow as possible at the 1/3rd position because ideally it's best not to have a large spread of colours, you only want to record the dust and vignetting in the optical train, colour is not needed or really desirable.
Thanks Andrew, I tried again with P-setting and ISO 100...but this time with white t-shirt and pointed a blue sky away from the sun.... I now have a blue tint and separation on the blue, green and red channels?
cheers,
rob
hmmm I do all my flats at pre-dawn when the sky is still fairly unaffected by the sun's gradient, I have not used the t-shirt method so not sure with your flats.
Do you use IP?
There's a video tut on the disks that show you how to convert the colour flats to mono so they wont affect the colour in the lights.
And I guess you are aware that flats should be taken with your camera being at exact position as it was when you took your light frames. If you remove your camera and attach it back again there is a big change that all dust on a camera sensor and a telescope lens/mirror that are expected to be removed after applying flats wont match with the same dust in your light frames.
Mike
Hi Mike, no I wasn't aware. Now I'm really confused. How can you take flats at night or do you leave the scope and camera out till day or use a light box? I thought flats were to correct for vignetting not dust?
Rob to take an image of your histogram, open your eos viewer, navigate to the folder of flats files and double click on one. A new window will open with an enlarged version of the thumbnail along with a histogram on the right. Hold down your press alt+prtsc keys. Open Photoshop, open a new image (should come up with clipboard as the default size) and paste. save as jpeg and upload.
Heres one I did earlier This Flat is over exposed slightly. Aim for 1/3 from the left of the histogram edge.
Also take note of what Mike has said about moving your camera, flats must be taken at the same orientation, focus and I believe ISO but that's just me .
Rob flats do a number of things; they help correct for vignetting, they help remove dust motes ....and...my reason for using the same ISO as you lights.....they help correct for non-uniformity of photoelectron production between pixels. While the QE (Quantum Efficiency) of the chip may be rated at 30% or whatever, that is an average across the chip. Each pixel is slightly different. A good flat should take that pixel difference into account and correct for it. By changing ISO when taking flats you are completely changing the pixel responsiveness to photons compared to when you take your lights.
For example (and this is only 1 reason for using the same ISO). IF you take an image at say ISO1600 your camera gain may be 1 ADU (Analog to digital Units), for arguements sake. That means for every electron generated in a pixel well you will get 1 ADU produced for storage, i.e 345 electrons = 345 ADU (I'm leaving out noise generation etc here to make it more straight forward).
But if you change you ISO to say 400 the gain may change to say 10 electrons/adu (its not really, its probably more like 2.5 but this is just an example ). This means that if your pixel well only has 345 electrons in it then you will only produce 34 ADU. the other 5 electrons are not counted. Those 5 electrons are important. They must be taken into consideration when correcting for pixel QE differences.
Hi Mike, no I wasn't aware. Now I'm really confused. How can you take flats at night or do you leave the scope and camera out till day or use a light box? I thought flats were to correct for vignetting not dust?
The same applies to vignetting. A camera sensor is not square and if you rotate it vignetting recorded by sensor will be different than it was before rotation.
No, I didn't use a light box when I was using DSLR. I didn't even bother to build one. I used to bring my mount with my camera left on scope inside and use blinds illuminated by downside lights to take flat frames. It wasn't hard for me to bring the whole setup back into the house because I have a wheel on one leg. Using it I can roll the whole setup by grabbing other two legs.
If you cant bright your setup still assembled it does not mean you have no choice to take proper flats without leaving your scope outside. You can remove your scope from your mount with camera still attached to it and bring everything back into your house. Then it will depend on your method of taking flats. You can wait till the next morning and bring everything back outside and use sky glow and t-shirt for taking flats. Or you can also use my method with blinds/white wall or even build a light box and do it inside your house the same night you take your lights.
Interesting.
I use the T shirt also but aim the scope at a white sheet draped on the wall in front of the scope. I then aim a manual flash at the wall and take a 1/60s exposure. The histogram attached shows that the response to each colour isn't even. The colour balance is set to daylight in the histogram. If I set it to "flash" then the red channel moves further to the right.
On a different not, I think it is probably better not to "defocus" for flats. This will slightly change the distance between the focal plane and any optical surface that is between the T shirt and the focal plane. This will make dust motes slightly incorrect in size & not divide out as well.
I know this thread is almost a month old, but I thought it better to ask a question here as opposed to making a new thread.
I'm just learning to take and use flats. I can't leave my scope out at night so I built a light box to use for taking flats. However, because of this, my flats are yellow.
The program that I use to stack and process darks/flats has an option of converting the flat to monochrome before applying it. But I've noticed that doing that tends to make the flat lighter.
I guess I said all that to ask if it's preferable to change my camera to black and white, then shoot my flats?
Also, as opposed to Darks, where the more the merrier, it's not necessary to take and apply multiply flats is it?
it's not necessary to take and apply multiply flats is it?
Yes it is. Flats contain random noise as well and you want to minimize that as much as possible otherwise you will introduce more noise into your lights.
The best way to do that is to combine multiple images to produce a master flat, the same way you do a master dark. Seeing as your flats would normally only be a few seconds long at most collecting multiples is not a problem. I use a manual bulb cable and just lock it open for a while and collect a couple of dozen of them and then select the ones with the closest histogram spread.
As for the colour, if you really want to go to the extremes you can split your lights into RGB components, split your darks into RGB components, split your flats into RGB components and then process them separately then recombine them, but only if you really like to punish yourself with processing. Using them as greyscale should be fine especially if your software debayers your lights initially anyway (ImagesPlus does this under the automated control, not sure about others).
Just remember if you take flats more than a few seconds long you will should really take flat darks as well.