dumb questions about flat frames

[troypiggo]

Dumb questions about flat frames I've been wondering for a while.

What if you want to take some at a star party and can't use white light? Can you gel your lighbox or whatever you use red? Do you need the light in all 3 RGB channels, or is one fine? I'm sure I read in one of the stacking software guides something about channels.

Taking flats the next day - does the orientation of the camera on the scope matter? Say you need to rotate your SLR during the night for composition. Can you use the same flats for different orientations? I wouldn't have thought so since the optical tube is round, but thought I'd ask anyway.

[Alchemy]

i use t shirt flats which are noticably blue , i havent seen a problem yet, if you are using a one shot color camera there could be the potential if you only used red ( its a good question).

orientation doesnt make a differece in mine.... not yet and ive pushed them pretty hard.


clive

[Dennis]

Hi Troy

Not really dumb questions – all astro photographers ask these questions at the beginning of their apprenticeship!

A flat field is an image of the “as-used-to-image” optical train so no; you should not change the focus or rotate the camera.

If you change the focus, artifacts such as dust doughnuts will grow/shrink in apparent size.
If you rotate the camera then defects such as dust will now be recorded at a different X-Y position in the flat field image, compared to their X-Y position in the (light) target image.

Then you have other artifacts such as vignetting and gradients, so again, the light image and flat field image need to register precisely. If the flat field is rotated or re-focused compared to the light frame original, when you divide the flat into the light, you will introduce artifacts which will appear as the differences between the rotated sets.

Cheers

Dennis

[troypiggo]

Thanks guys. Dennis - that makes perfect sense about dust etc. No rotation for me.

Still interested in thoughts/advice about the red light.

[Dennis]

Hi Troy

I use white LED’s in my flat field box but I’m sure if I were to analyse the spectrum they produce it would likely have a lot of blue.

I guess that if you are shooting B&W only and the red LED’s emission wavelengths are not blocked by anything in the imaging train, such as IR block filters, then I can’t see why red LED’s could not be used?

But, if you were to migrate to LRGB imaging through RGB filters, then I suspect that white light would be the preferred option due to G&B absorbing the red? You could probably compensate by increasing the exposures for G&B and in post processing, but that makes more work so I suspect it’s better to solve the problem up front with white light flat fields.

I’m just guessing at the above as I really haven’t done enough LRGB imaging to see it working in real life, so I’ll offer my comments as a starting point. Hopefully more experienced and knowledgeable imagers can address your question more fully.

I think that photometry might be the first casualty of using red LED’s if you wanted to measure stars?

Cheers

Dennis

[troypiggo]

I think I get you. But...

If shooting LRGB, wouldn't you just apply the red gelled (or red LED) flats to the L channel rather than one of the colour ones?

In any case, most of my imaging will be with a DSLR, at least for a while. Don't foresee any photometry etc.

[Dennis]

Hi Troy

I assumed that RGB would be through a Filter Wheel with separate RGB filters, so each filter will require a Flat Field due to dust, perhaps differential focus, etc.

Cheers

Dennis

[Paul Haese]

Actually that is not quite true Dennis about camera rotation. The only time this is an issue is if you have a flattener that has dust motes on it. The dust on the camera chip which is the most common dust mote you get will always rotate with the camera. Focus will not matter in this case excepting again when you have dust motes on a flattener or reducer.

Flats can be performed just after sunset too. You don't need a light box, just a flat featureless part of the sky is also fine. The white T shirt is also very good too.

[wysiwyg]

From what I have read, you do need to take separate flats if you rotate your CCD using a particular filter. If there is dust on the filter and you rotate the CCD the dust will move.
If you are taking LRGB images flats are necessary for every filter.
You can stagger the roattion every 5-10 degrees take a flat and then median combine for that filter.
Dont forget to do darks for your flats as well.

Mark

[White Rabbit]

darks for flats as well, good lord does the process never end??? ;)

[Dennis]

Here is a post (http://www.iceinspace.com.au/forum/showpost.php?p=461020&postcount=8) showing some examples of RGB Flats with different artifacts on each Flat Frame.

Cheers

Dennis

[wysiwyg]

A perfect example of when you would be required to take flats again after rotation is if you are imaging through lets say the red filter, and you have to do a meridian flip, which means you have just rotated your camera 180 degrees.

In the example given in the previous post, the large donut on the green filter would shift 180 degrees also, hence flats are necessary if you rotate.

Mark

[troypiggo]

You lost me on the "meridian flip" logic. When your mount goes through a meridian flip, doesn't the relationship between the scope, camera, and filters still stay the same though? Isn't that what the flats are for?

[wysiwyg]

After a meridian flip, if you do not rotate your camera, your image will be upsidedown. Hence you need to rotate the DSLR or CCD 180 degrees.
So the relationship between the camera and everything else changes depending on what kind of setup you have.

Mark

[Paul Haese]

That is all true but one thing to remember is that if the whole rear system rotates including filters and reducer then you don't need to take new flats as the whole system has not changed.

[wysiwyg]

I don't agree Paul, the system has changed.
In an average refractor imaging setup has potentially 7 surfaces that can attract dust in the imaging train. Front optics (2), reducer/flattener (2), filter (2) and CCD chip (1).
If you rotate any of these independently you have changed the system.
Hence flats are still required.

The only way you can avoid taking flats for a rotation, is if you rotate the entire OTA.

Mark

[jase]

Mark is correct, but it should be made clear as to why you'd need to rotate the camera post a meridian flip as the comment that your subs end up upside down is irrelevant. In fact, I'd go as far as saying that subs taken on the east, then the west side of a mount can actually benefit the imager as it will place chip anomalies (dead pixels, column defects) on different parts of the image making them easier to remove through data rejection algorithms.

The primary reason why you rotate the imaging train post flip is to pick up the same guide star when guiding on axis. Rather important when you're operating at long focal lengths. For the regular imager, particularly one using a wide field instrument this is a mute point as guide stars are typically plentiful for on axis guiding. So in this case, you wouldn't bother rotating your imaging train.

Automated data acquisition tools such as ACP, CCDAutoPilot etc will take this into account;
"With rotator support, if the telescope is a GEM, half of the flats in each set are taken with the rotator at the specified PA and the other half are taken 180 degrees from the specified PA. This allows making flats for the common case where the target crosses the meridian, invokes ACP's auto-flip feature, and thus has a 180 degree rotation in the middle of acquiring its images. The west and east images require flats with rotations 180 degrees apart."

I have heard of people rotating the flat manually to address this. We should note that we're talking quite an advanced technique here and it can generate considerable workload...
"Note that use of a rotator for multi-target imaging can quickly lead to needing far more flats than you can produce. This is the main disadvantage of using an internal guider and rotator. Careful planning of multi-target runs can help. For example, if you are going after three targets over three nights, you can take one set of flats for each target on each of the three nights at dusk and dawn. Take all of the flats for a particular sky position angle (for one target) in one dusk or dawn run to eliminate rotator slew time detracting from the time available for flat acquisition."

No one said producing flats was straight forward. Once you've done it a few times, familiarity kicks in and it becomes routine. Start off with lightbox or tshirt flats. Skyflats need further consideration as the sky is dynamic in nature. At dusk its constantly getting darker, at dawn, brighter. This needs to be factored in if you're trying to hit the target ADU count for each flat sub illumination (20%-30% well depth).

Flat Darks Addition: Don't forget those with cooled CCD cameras that can take bias frames have the ability to scale a dark frame to match a light frame (or in this instance a flat frame). In essence, subtracting a bias frame from a dark frame provides a thermal frame to match the light frame exposure. The act of scaling darks can work exceptionally well if you've got flat frames with varying exposure times. For example a narrowband Ha filter does not pass as much energy as a green filter, hence would require greater exposure time to reach the target ADU. Dark scaling would allow you to take a set duration dark say 300s and scale it to match the light frame providing its under the 300s duration.

Cheers

[wysiwyg]

Jase,

You touched on an interesting point here. Even in widefield I would have thought that rotation after a flip is a must, especially if you are going to plate solve. I could be wrong, but how can you achive a plate solve with the image 180 degrees.
My understanding of plate solving is still very limited, so forgive me if I am missing something.

Mark

[jase]

Mark,
No, plate solving doesn't care about the rotation. All its doing is matching stars in the FOV and referencing a catalogue to look them up. Based on reference information, plate solving will then advise of the PA (position angle) of the camera's rotation - zero (0) being true north, 180 true south.

Cheers

[wysiwyg]

I knew I missed something.

Thanks Jase.

Mark