Flats for all of LRGB or just L?

[SkyHiker]

Howdy, last night my object was NGC2170 in LRGB. I normally take flats and biases in L only. That usually works but this time my stacked colors looked rather horrible except B. It looks like there was a nasty gradient running through the image. The way I take my flats is against a white wall lit up by my headlamp. The histogram was at about 20% of the well depth, a bit low but fine otherwise. I did use bias files in L, as usual. 20xL flats, 20xbiases, 10x5 min LRGB. Gain=210, offset=24.

Do any of you have advice about taking flats separately for LRGB, and if so what is the reasoning? I can't think of a good reason to do this andd I don't want to spend more time than I should.

I have to say that the fact that B looks good and that B was the last color I took, made me wonder if I had reset the filter to L. But when I was taking flats I did look at the filter, and Ekos created a folder of that color for it so it must have been right - that can't be the reason.

What else?? Well this is not too far away from the Alnitak bros so it may have caused a gradient. But why not for B. Another one, the OTA started out in the direction of Santa Barbara where most of the light pollution is. So it gradually got less LP as time moved on.

One more option, I did a manual flip after the 8th G (the sequence was LRGB). Are the flats somehow oriented flipped 180 degrees? I checked how those images are indeed stacked upside down, and they turned out OK.

PS When I say upside down, I mean that DSS shows them upside down but of course it corrects for the orientation, as you can tell in the images below (produced by a quick DSS edit).

Aargh now I have to correct it in Gimp with extra artificial flats.

L

R

G

B

[yobbo89]

I can't help you . I use 1xflat taken in ha used for ha,nii,oiii,sii and 1 x flat in lum for lrgb,or get rid of any minor problems in post, i'm extremely lazy with flats , if there is a dust mote, i will pull apart the camera and get rid of it, it is a much viable option then taking flats every night .

what colour is your led ? it's strange that one side is dark and one side is brighter in your lrg frames , light leakage comming in ?

strange blue should be the worst for Alnitak , do you have an image of your flat frame to upload ?

[SkyHiker]

I can't help you . I use 1xflat taken in ha used for ha,nii,oiii,sii and 1 x flat in lum for lrgb,or get rid of any minor problems in post, i'm extremely lazy with flats , if there is a dust mote, i will pull apart the camera and get rid of it, it is a much viable option then taking flats every night .

what colour is your led ? it's strange that one side is dark and one side is brighter in your lrg frames , light leakage comming in ?

strange blue should be the worst for Alnitak

Thank goodness that you do the same, I thought that it was just common sense to do it that way.

The color of my LED headlamp is white as in cheap consumer headlamp LEDs.

I now doubt if it was Alnitak, I am now wondering how DSS determines how to orient the flats - it doesn't know about the flip, and a flip does turn all images the other way doesn't it. Since B worked out fine, and the flip happened right before that. How in general does DSS know how to orient the flats?

[SkyHiker]

Oops actually in retrospect you do take flats for each color, just once. Is one flat enough? You must have a pretty highly lit set if one flat is enough.

Let me ping the DSS group.

[yobbo89]

Oops actually in retrospect you do take flats for each color, just once. Is one flat enough? You must have a pretty highly lit set if one flat is enough.

Let me ping the DSS group.

The illumination is pretty decent on my scope, it's not perfect even with the 2.5'' focuser and larger secondary, the variance from using one flat frame on all filters is adjusted with frame adaption and background extraction in post processing , i would say my crude methode alone is not enough without the post processing but it dose work better then no flats.

[yobbo89]

I can't help you . I use 1xflat taken in ha used for ha,nii,oiii,sii and 1 x flat in lum for lrgb,or get rid of any minor problems in post, i'm extremely lazy with flats , if there is a dust mote, i will pull apart the camera and get rid of it, it is a much viable option then taking flats every night .

what colour is your led ? it's strange that one side is dark and one side is brighter in your lrg frames , light leakage comming in ?

strange blue should be the worst for Alnitak

Thank goodness that you do the same, I thought that it was just common sense to do it that way.

The color of my LED headlamp is white as in cheap consumer headlamp LEDs.

I now doubt if it was Alnitak, I am now wondering how DSS determines how to orient the flats - it doesn't know about the flip, and a flip does turn all images the other way doesn't it. Since B worked out fine, and the flip happened right before that. How in general does DSS know how to orient the flats?

DSS should auto flip frames. the info should be in the fits/file header of the image among many other things,exposure time ect , it knows where pixel 1 is on all frames .

i've only ever encountered a bug in pixinsight where i have had to manualy rotate a frame that would not combine in a mosiac generally pi flips and rotates the images when alighning with another frame.

[KathyNS]

You need a set of flats for each filter. Multiple flats for each. I would take 15-20 flats for each filter. Every filter has its own dust bunnies and each one can have its own gradients, so lack of flats will make itself known eventually. Bias frames can be saved in a library. Flats can be saved only if the optical train does not get disassembled between sessions.

[STEVE333]

Hi Henk -

First, ideally one has flats for each filter to take care of any artifacts specific to each filter (dust motes, scratches, etc.). I don't change my camera orientation and I always leave the camera attached to the filter wheel, so, one set of flats lasts me for many months.

Second, when creating each master flat I usually combine 30 flat frames to reduce the noise (reduces noise in master flat by 1/sqrt[30]). This way the flats add minimal noise to the final stacked image.

Third, when you do image calibration, it doesn't matter if the telescope was flipped or not when an image was captured. Image calibration (including flat frames) are correcting the camera image NOT the telescope orientation. Different image orientations (caused by the mount flipping at the meridian) would only be an issue when stacking the images, and, your processing software might even flip them for you automatically during stacking. If not you can always flip (rotate?) them manually before stacking.

Hope this makes sense.

Steve

[dcrowson]

I'd second the last couple comments. You want flats for every filter used. One is better than none but the SNR is the same as lights so two are twice as good as one. People tend to throw out random numbers of exposures but it is the same as lights in that once you get to 24, it takes going up to 48 to the next meaningful increase. You can definitely get away with much less. These are always short exposures - I would take at least 7-8.

You mention bias files. I couldn't find what camera you use in in your vast list of equipment in your sig. If you're using a CCD, subtracting a bias from the flats is probably fine. If you're using a CMOS camera, subtract darks taken at the same settings and duration. Not doing this (or using bias) can result in over or under correcting.

Lastly, I always like to mention that if you find a way to keep your imaging train connected, you can use the same flats for a long time. I made a small wooden carrier for my refractor (the dovetail fits into it). I move this from my garage or even transport it to a remote site. I can and have used my flats for months or even longer. Differences in focus tend to not matter much.

These are topics where you'll ask 50 people and get 60 answers so I'd try and see what kind of results you get. Mine are based on a lot of imaging.

Dan

[SkyHiker]

DSS should auto flip frames. the info should be in the fits/file header of the image among many other things,exposure time ect , it knows where pixel 1 is on all frames .

Ah of course, DSS knows which light subs are flipped but the base reference is always the same for all calibration frames, so it knows...

[SkyHiker]

You need a set of flats for each filter. Multiple flats for each. I would take 15-20 flats for each filter. Every filter has its own dust bunnies and each one can have its own gradients, so lack of flats will make itself known eventually. Bias frames can be saved in a library. Flats can be saved only if the optical train does not get disassembled between sessions.

I'm curious why it would make that much difference though.

When you talk about gradients, it must be about the gradients in the flats. Usually, we talk about gradients in light subs because of light pollution from the sky, and sometimes light that enters the OTA from a streetlamp. Those are not present I think when I take flats against a wall. And if they are present, I would think it would be pretty even across the light bands, would they not?

About the dust motes, you must be referring to the possibility that the light that goes through dust motes to the sensor is spectrum dependent. I was hoping that this spectrum dependence would not be that bad. So far, I have not been able to see that - white light in L always took good care of them.

The spectrum dependent light intensity for vignetting should not matter for flats because flats are multiplicative. I have never seen any bad effects of taking flats against the blue sky (something that is common practice) or working with white light in L. Aside from a filter dependent constant (that does not matter because different intensities can be corrected in the editor) that too should be spectrum independent.

The differences that I see in the images are way beyond what I would expect are reasonable spectrum dependent differences. Maybe I don't get the whole picture yet, not sure wat I'm missing.

[SkyHiker]

I would take at least 7-8.

I usually take 20-30.

You mention bias files. I couldn't find what camera you use in in your vast list of equipment in your sig.

Sorry it's the ASI2600MM.

If you're using a CCD, subtracting a bias from the flats is probably fine. If you're using a CMOS camera, subtract darks taken at the same settings and duration. Not doing this (or using bias) can result in over or under correcting.

I don't do this, the ASI2600 has negligible dark current.

Lastly, I always like to mention that if you find a way to keep your imaging train connected, you can use the same flats for a long time. I made a small wooden carrier for my refractor (the dovetail fits into it). I move this from my garage or even transport it to a remote site. I can and have used my flats for months or even longer. Differences in focus tend to not matter much.

Re-collimating the Newt will have some effect I'm afraid. I don't mind taking flats and biases, it's pretty quick and I have to get up anyway to take the equipment in.

[KathyNS]

You need a set of flats for each filter. Multiple flats for each. I would take 15-20 flats for each filter. Every filter has its own dust bunnies and each one can have its own gradients, so lack of flats will make itself known eventually. Bias frames can be saved in a library. Flats can be saved only if the optical train does not get disassembled between sessions.

I'm curious why it would make that much difference though.

When you talk about gradients, it must be about the gradients in the flats. Usually, we talk about gradients in light subs because of light pollution from the sky, and sometimes light that enters the OTA from a streetlamp. Those are not present I think when I take flats against a wall. And if they are present, I would think it would be pretty even across the light bands, would they not?

About the dust motes, you must be referring to the possibility that the light that goes through dust motes to the sensor is spectrum dependent. I was hoping that this spectrum dependence would not be that bad. So far, I have not been able to see that - white light in L always took good care of them.

The spectrum dependent light intensity for vignetting should not matter for flats because flats are multiplicative. I have never seen any bad effects of taking flats against the blue sky (something that is common practice) or working with white light in L. Aside from a filter dependent constant (that does not matter because different intensities can be corrected in the editor) that too should be spectrum independent.

The differences that I see in the images are way beyond what I would expect are reasonable spectrum dependent differences. Maybe I don't get the whole picture yet, not sure wat I'm missing.

I think you are reading a lot into my comments that was not there.

Light doesn’t generally go through dust motes. They create shadows. The dust motes on the red filter are not the same as the dust motes on the green or the blue filter. Each filter introduces its own dust shadows, and the only way to eliminate them is to take separate flats through each filter. If you only take flats through the luminance filter, you will eliminate the dust shadows in the luminance frames. But you will be subtracting shadows that don’t exist on the red, green and blue filters, plus you will miss the shadows that are on those filters.

[SkyHiker]

I think you are reading a lot into my comments that was not there.

Light doesn’t generally go through dust motes. They create shadows. The dust motes on the red filter are not the same as the dust motes on the green or the blue filter. Each filter introduces its own dust shadows, and the only way to eliminate them is to take separate flats through each filter. If you only take flats through the luminance filter, you will eliminate the dust shadows in the luminance frames. But you will be subtracting shadows that don’t exist on the red, green and blue filters, plus you will miss the shadows that are on those filters.

Thanks, and very true, I was thinking about dust motes on the sensor, but you are right each filter of course can have their own dust motes. I'm still unclear about gradients you were referring which is why I elaborated but it doesn't matter.

Dust motes are not the main problem with the images that I showed though - the main issue is about light being blocked or reduced by obstacles such as the CC, filter holes, incident angle of light on the sensor. These effects are largely independent of color. Yet they are huge in this case, and I don't know what I am missing.

[yobbo89]

if you have a clean sensor without dust motes you can kind of get away with using one set flats on all filters for the vignetting, mainly in two separate sets ie narrowband and full spectrum,

generaly , variant gradiants are from outside the scope, in my experience fixed lopsided gradiants are from vignetting and collimation, not having your secondary centered can give you an offset gradiant , if your scope stays in collimation the vignetting equal or lopsided will be permanent and flat corrections should be repeatable through all filters for vignetting except for defects like motes and scratches , keeping a collimated scope can have 1 set of flats all year round unless dust shifts around.

[SkyHiker]

if you have a clean sensor without dust motes you can kind of get away with using one set flats on all filters for the vignetting, mainly in two separate sets ie narrowband and full spectrum,

generaly , variant gradiants are from outside the scope, in my experience fixed lopsided gradiants are from vignetting and collimation, not having your secondary centered can give you an offset gradiant , if your scope stays in collimation the vignetting equal or lopsided will be permanent and flat corrections should be repeatable through all filters for vignetting except for defects like motes and scratches , keeping a collimated scope can have 1 set of flats all year round unless dust shifts around.

Thanks Rob, that's what I thought.

I checked everything and still have no idea why this happened. I will take some flats before it gets dark and make them a bit more saturated. I will take them against the wall and against the sky, for all filters and see if it makes a difference. I install my rig pretty much the same way each time and after I focus it should be accurate enough. When I take them against the wall, I have to tilt the OTA up a bit or else I get the azaleas in the flats - don't know if that was issue, I think not.

Ditto about the collimation. It's a good thing that you have a 2.5" CC, you can see the significant vignetting that I have so taking good flats becomes more of an issue.

[XCalRocketMan]

if you have a clean sensor without dust motes you can kind of get away with using one set flats on all filters for the vignetting, mainly in two separate sets ie narrowband and full spectrum,

generaly , variant gradiants are from outside the scope, in my experience fixed lopsided gradiants are from vignetting and collimation, not having your secondary centered can give you an offset gradiant , if your scope stays in collimation the vignetting equal or lopsided will be permanent and flat corrections should be repeatable through all filters for vignetting except for defects like motes and scratches , keeping a collimated scope can have 1 set of flats all year round unless dust shifts around.

Since I'm usually moving the camera and filter wheel from one OTA to another I need to take flats for all filters as there is no way to prevent the dust motes from ruining a good imaging run.

[SkyHiker]

After last night imaging M81 where I took fine flats, I again got strong circular gradients. I think the cause is the supernova across the street. The neighbors love having a net of LED's around their tree and a string of LEDs around the roof line the whole year through. This is why the gradient is worst at L then R then G then B - that correlates with the angle at which the LP from the neighbors comes in, since I image in the sequence LRGB. G and B hardly have gradients, whereas R and especially L have gradients much stronger than the nebulosity. I can recognize it from the shape of the luminosity curves in DSS. I will get a yoga mat, fashion it into a light shield and try again tonight.

[XCalRocketMan]

After last night imaging M81 where I took fine flats, I again got strong circular gradients. I think the cause is the supernova across the street. The neighbors love having a net of LED's around their tree and a string of LEDs around the roof line the whole year through. This is why the gradient is worst at L then R then G then B - that correlates with the angle at which the LP from the neighbors comes in, since I image in the sequence LRGB. G and B hardly have gradients, whereas R and especially L have gradients much stronger than the nebulosity. I can recognize it from the shape of the luminosity curves in DSS. I will get a yoga mat, fashion it into a light shield and try again tonight.

Yeah, that seems to be a problem for a lot of folks. I'm lucky in that I don't have light issues from my neighbors, but every now and then I need to deal with smoke in the winter months from the neighbor up my extended driveway who uses an outdoor wood burning heating system to heat his house. Not a big problem as the smoke usually dissipates rapidly, but ever now and then it just hangs there in the southern sky.

[Stuart]

I'm going to put another wrench into this. Your method of casting a headlamp on a wall may be causing more problems. Flat calibration frames have to be completely uniform in illumination. The problem with a headlamp is that the center will be brighter than the edges and that will screw up the flat frame subs. That's why many of us have illuminated flatfield panels.