Lights, Darks, Flats & Bias.

[Moonshed]

Okay, I know what lights are. I also know what darks are. I know what flats are but do not know how to take them. Bias, I have absolutely no idea. Any advice would be most welcome. Thanks. Keith.

[geordie85]

 

Hope this helps. 

Doug has alot of good astrophotography photos

[wxsatuser]

Bias, like a dark but the shortest exposure your camera does.

[ollypenrice]

1 hour ago, wxsatuser said:

Bias, like a dark but the shortest exposure your camera does.

...and they are useful as darks for your flats. They also allow some stacking software programmes to scale your darks to your lights if there is a temperature mismatch between the two.

Olly

[Ouroboros]

I place a white cloth over my telescope tube and use an A4 light screen to produce an even illumination. I then take about 20 images with the camera on auto so it decides the appropriate exposure.   I usually do this at the time I image so the camera is at focus and in the same orientation as when the images were taken. 

In the beginning when trying to learn this imaging game I didn't bother with flats. But I found that flats make the most significant noticeable difference to the quality of my images. Those and darks. 

[Zakalwe]

Darks and Bias calibrate for the sensor. As long as you take darks at the same temperature and duration as the lights they will be fine (hence the big advantage of setpoint cooling- you can create a library of darks to use over and over- redo them about once per year as the sensor characteristics change over time).

Bias are the shortest possible exposures that your camera can take- they calibrate for read-out noise. Again, you can create a library, though if you are using darks then you don't need Bias, as the read-out noise is present in the darks (bias are handy if you are not using darks, or are using Bad Pixel Map).

Flats have to be done per filter WITHOUT moving the filter. However, most people only use flats for the Luminance files- the thinking being that the Lum files contain the structure and fine detail. RGB doesn't need flats, as you can blur the heck out of colour data as long as you have decent, calibrated Lum files.
 

[Moonshed]

Thank you guys, you have all been most helpful. Reading through some of the posts here I have come to understand the importance of Flats and Bias etc. but as I said didn't know how to take them. Now I do. Thanks so much. Keith.

[cjdawson]

For the flats you just need an even light - colour doesn't really matter too much (but I do prefer white).

I made a light frame using an EL Panel.  This works great.

Here's the panel that I used.

http://elwirecraft.co.uk/el-products/el-panel-categories/white-el-panel-in-many-sizes-a6-a5-a4-a3-a2-and-a1/

Just put it into a picture frame, and you are done.  For others, might be a good idea to throw some clother over it to block the stray light.

[Moonshed]

6 hours ago, cjdawson said:

For the flats you just need an even light - colour doesn't really matter too much (but I do prefer white).

I made a light frame using an EL Panel.  This works great.

Here's the panel that I used.

http://elwirecraft.co.uk/el-products/el-panel-categories/white-el-panel-in-many-sizes-a6-a5-a4-a3-a2-and-a1/

Just put it into a picture frame, and you are done.  For others, might be a good idea to throw some clother over it to block the stray light.

Thanks for that, looks like a good way of getting an even consistent light.

[cjdawson]

You might also want to look at led boxes.  People have success with those too.

[alacant]

Hi. Try a stack without the dark frames. I too have a Canon and you'll probably find that you have less noise that way. Bias and flat, yes. Dark, no. There's a techie reason for it involving internal electronics on Canon dslrs which explains this. That linked to the impossibility of being able to match dark sensor temperatures with those of the light frames is probably why dark frames make things worse. Oh, and dither in between frames. It's a one click option in APT and really improves your snaps. HTH 

[MattJenko]

On ‎06‎/‎09‎/‎2016 at 08:55, Zakalwe said:

 RGB doesn't need flats, as you can blur the heck out of colour data as long as you have decent, calibrated Lum files.
 

Why would you blur the heck out of the RGB channels, even if you are going to apply a Lum over the top? It makes no sense to me.

[Zakalwe]

7 minutes ago, MattJenko said:

Why would you blur the heck out of the RGB channels, even if you are going to apply a Lum over the top? It makes no sense to me.

You don't have to, but you can without affecting the final result. Colour information is remarkably vague.

A certain amount of blur in the RGB channels can smooth out any noise which can help to reduce the overall noise in the final image.

[Zakalwe]

To illustrate the above:

Here's an M42 image:

Here's the Luminance information:

and the RGB information

 

Here's another version of the final image:

after a massive Gaussian blur has been added to the RGB data:

 

 

[MattJenko]

Thanks Stephen - pictures do indeed speak a thousand words. I guess star clusters would benefit from tighter RGB channels, but I can clearly see the point you are making with those images.

[Zakalwe]

28 minutes ago, MattJenko said:

Thanks Stephen - pictures do indeed speak a thousand words. I guess star clusters would benefit from tighter RGB channels, but I can clearly see the point you are making with those images.

I think the point is that you don't need that much sharpness or detail in the RGB part of an image. So with L-RGB imaging you can quickly collect the colour data using binning, which drops the resolution and shortens the capture time. This low sharpness information can then be added to high-quality luminance data that you've spent a lot longer collecting. The end result is the same, or better, than an equivalent length of time capturing just RGB data.

[MattJenko]

I wonder then when I tried binning RGB in an LRGB image, I perversely don't think I "blurred" the RGB enough, in that I got slightly blocky stars as a result of having the RGB layer too crisp...

Apologies @Moonshed for slightly hijacking this thread. I'll let it run without my interference from now on!

[Zakalwe]

20 minutes ago, MattJenko said:

I wonder then when I tried binning RGB in an LRGB image, I perversely don't think I "blurred" the RGB enough, in that I got slightly blocky stars as a result of having the RGB layer too crisp...

Apologies @Moonshed for slightly hijacking this thread. I'll let it run without my interference from now on!

That could be down to the sampling ratio of camera to telescope? You could also try restacking the Luminance data and selecting x2 Drizzle in Deep Sky Stacker.

http://deepskystacker.free.fr/english/technical.htm#Drizzle

http://www.lightvortexastronomy.com/stacking-with-drizzle.html

[Moonshed]

6 hours ago, MattJenko said:

 

Apologies @Moonshed for slightly hijacking this thread. I'll let it run without my interference from now on!

 No need to apologise, I love it when a thread expands, I learn new things that way.