Flats

[Mike]

I've been having problems with flats. It's not a big problem yet because I have yet to capture lights of a quality that would really benefit from good flats. I seem to be trying to learn all phases of astrophotography at once. My problem with my flats is that I don't seem to be capturing them bright enough. I use a home made light box that has 37 LEDs built into a cylinder that inserts into the dew shield of my TEC 140, so the luminance of the source is not at issue. A friend of mine who is a very smart accomplished imager and processor has used some software that he has and found my flats not bright enough. He says, and I agree because I read the book he learned it from, that flats should be about 50% of the full well capacity of the camera sensor. My flats have been about 33%.

The last time I imaged I took 3 sets of flats; one set using AV flat mode, one set at 1/125 second, and one set at 1/100 second. None of them were bright enough.

I complicate the situation by using an Astronomik CLS clip in filter which skews my RGB histogram with a wide spread between green at the left, red in the middle, and blue at the right. I thought I was compensating by using the Luminance histogram and getting it plus and minus center with the 1/125 and 1/100 second. Still I ended up not bright enough.

Is there a tool in BYE that I have not found yet to get the right setting? Does anyone have any advice to help me get repeatably correct ADU values for my flats?

Thanks in advance for any advice.

Mike

[Mike]

I agree with Guylain too. Though I don't need shorter flats necessarily. What I need is to replace my flat box which uses LEDs with a bluish hue, and possibly my Astronomik CLS filter. That combination is causing the wacky histogram, ruining the ability of stacking software to calibrate properly. So an ELP is getting shipped to me today.

I am going to do a series of flats of many exposure lengths with BYE when I get the new light panel and post the ADU values to help out other folks having these problems.

[s3igell]

Belay that "change my Astronomik CLS filter"...

You need to take your Flats with the Optics Train kept the same (components, focus, camera-to-scope orientation) as your Lights.

Remember, that while "fixing a wacky histogram" is important, so is "ensuring Vignetting and Dust Motes and Internal Reflections are SAME for Flats as for Lights".

I believe your intention (from other Posts / other Forums) is to attempt future Imaging at your semi- Light Polluted site without the use of your CLS Filter.  This is a good experiment - trading Skyglow/Exposure Time for More Natural Color Cast.  But once you commit Images with a specific setup, you need to keep that setup all the way through Flats generation.

I also think that replacing the Light Source in your Lightbox will bring the Blue "Under Control" - although many Electro-Luminescent Panels have a slightly Green Cast (mine does - but it lifts both Blue and Green Histograms - and my Modded 550D is already extra-sensitive in Red - it all kinda balances out).

[Mike]

Thanks S31gell. This set of data is only going to be used for the "things you shouldn't do" file. When I image my plan has always been to take flats before I move my camera IE if I have to adjust angle to reframe a new object, or at the end of a night of imaging. So once I get this whole thing working flats will always be specific to lights. Hopefully that will be happening within a few months. The new flat panel is on the UPS truck.

[Phil Leigh]

This RAW flat file is NOT over exposed - it's not even remotely close to being over exposed. It's no good looking at this type of file in DPP or any "8-bit" software in RGB mode - you need to look at the RAW histogram.

The peak ADU of this file is ~2300 out of a possible 16384 (14-bit camera).

 

[astroman133]

Sorry about that! I based my comments on the understanding that the histogram was a histogram of the RAW data.

I take my flats with an Electroluminescent (EL) Panel.  My exposures range from 0.03 seconds with my DSLR to 24 seconds with my CCD camera with a Hydrogen Alpha filter.  The light from the panel is not pure white nor does it need to be since the software that I use for image calibration uses only the brightness/luminance of the image.  I look at the "at capture" histogram to verify that my exposure is correct.  I also keep a Word document with the historic exposure values for my various setups.

I don't use DPP. Perhaps that is what caused my confusion about the histogram.

Sorry for the misdirection.

 

 

[admin]

Hey guys, I dropped the bomb on this one.

Phil is correct and I'm wrong... and drove Rick on the wrong side of things in the process.  sorry about that,

The histogram in BYE is driven from the jpeg representation of the raw data and Phil is right, the flat is under exposed when looking at the histogram from the raw data.

Sorry for the confusion.

Guylain

[AstroJeff]

This is interesting!

BYE should be displaying the RAW histogram, shouldnt it?  If the histogram is not representative of the RAW data then we are all under exposing everything, right?  I use the histogram to expose to my sky limits, which is about 1/3rd of the way on the histogram.  I used to use a CLS Clip in but I hated it and stopped using it just shooting more images to equal the SNR of the longer subs.  Only LP filter I use is a Narrowband Ha filter.

What is AV Flat for then?  I also use a T-Shirt and elastic band and the sky for flats.

Jeff

[admin]

AstroJeff said:

This is interesting!

BYE should be displaying the RAW histogram, shouldnt it?  If the histogram is not representative of the RAW data then we are all under exposing everything, right?  I use the histogram to expose to my sky limits, which is about 1/3rd of the way on the histogram.  I used to use a CLS Clip in but I hated it and stopped using it just shooting more images to equal the SNR of the longer subs.  Only LP filter I use is a Narrowband Ha filter.

What is AV Flat for then?  I also use a T-Shirt and elastic band and the sky for flats.

Jeff

 

Yes, BYE should probably display the RAW histogram, I don't deny that for 1 second.  I will get to it eventually but for now the embedded jpg is displayed in BYE, as it always did, and that is the histogram display as well.

AV-Flat is just like taking an in-camera AV image... the camera decides the exposure length.

Regards,

Guylain

[admin]

You seem to be doing it right.  Why do you think you flats are underexposed?  Can you please post an image of your flats?

The key with flats is that no pixels falls off either side of the histogram.  If you get the Luminance about half way and then just glance at the R, B, and G individually to make sure none are touching either the left or right edge you are good to go.

I do not calculate the ADU, to be honest I would know where to start... but if someone can provide a concise formula I could implement it for sure.

Guylain

[Mike]

Thanks Guylain, I'll post one of each of my flats later when I am working on the laptop where they are stored. I won't be providing you with a formula, but I may be able to provide you with the software where the values are evaluated. I got the information about half the full well capacity from the book "The Deep Sky Imaging Primer" by Charles Bracken. I am so impressed with your willingness to continually improve your already excellent product. Thank you.

[Mike]

Here's a link to one of my brightest flats. https://drive.google.com/file/d/0BzTNnHsX_guhcjQ0Vi1pbElBcVU/edit?usp=sharing This flat was captured at 1/100 second exposure. My friend analyzed the the image with CCD Stack and came up with 3200 ADU. He recomends flats for a Digic 3 16 bit processor at 7000 to 8000 ADU. I tried using Nebulosity View> Pixel stats, but I apparently don't understand the data it's giving me because it indicates a "Mean" of 22376. I will look more into how this is evaluated.

[admin]

Mike, that flat is over exposed in the blue channel; the blue is falling off the right side (the highlights).

You need a smaller exposure, not a longer IMO.

Guylain

 

 

 

 

[astroman133]

Mike,

Guylain is correct.  The flat that you posted is overexposed because the blue peak is pushed against the right side of the histogram.  To correct this you need to shorten the exposure (which will bring all the peaks back to the left) or dim the light source.  It seems that your LEDs are strongly blue. and depending on your exposure time, it may be that the light source is too bright.

I would think of the pixel values as ADUs.  After all they are the value that the camera has assigned to the pixel after reading the analog photon count from the pixel well.  So having the histogram centered between the left and right sides is closely equivalent to have the sensor well half full of photons.