My first somewhat decent shot

I’m still new to this whole night sky photography thing. I snapped this one last night from my back deck, facing roughly SSW. It is 7 images stacked with DSS and some PP in CS3. I’m going to try some of the other free stacking programs to see how they compare. On a side note my camera seems to have a lot of hot pixels so I had to de-saturate the blue and magenta to get rid of those.

If memory serves correct..... 17mm, 2.8, ISO 800, 7 x 10 second exposures

I didn't understand the whole dark, flat, bias images in DSS so only uploaded the photos I took. I guess those would be the light images. I'll have to look into those other types of images and how to make them.

shane_c wrote in post #18136467
I didn't understand the whole dark, flat, bias images in DSS so only uploaded the photos I took. I guess those would be the light images. I'll have to look into those other types of images and how to make them.

When you shoot a digital image, some of what shows up on your sensor is "signal" (real information from objects that were visible to the camera) and some of what shows up on your sensor is "noise" (not real data). The noise is sometimes referred to as "Poisson noise".

Each image has a "signal to noise ratio" (SnR). What you really WANT is more signal and less noise.

The whole idea of stacking software is to help improve the SnR to result in much cleaner images. Many astrophotographers will refer to the number of "hours" spent capturing data and can generalize what sort of quality they expect to have based on the number of hours they spent collecting data (they are specifically talking about the "light" frames -- not necessarily the extra time to capture all the other non-light frames.) e.g. you make get decent results if you spend at least 1 hour capturing data. But you'll do better if you spend 2 hours. You'll do much better still if can spend 4-6 hours capturing data.

The more data you capture, the better the SnR of your resulting stacked image.

As for the other types of frames...

Dark frames - these are images shot at the same time, same camera, same true outside temperature, etc. except the lens cap is left on the lens. It allows the camera to gather up "noise". The processing software can use this to determine how much noise is present in a normal image (when the lens cap isn't on the camera) and also to detect "pattern noise").

Bias frames - these are frames to work out the true "bias" level of the sensor. In order for the sensor to work at all, it has to be powered up. If you think about this, powering up a sensor, and then simply doing a read-out (without actually taking an exposure) would actually read values for each pixel (even though technically you'd think every pixel should read "0" ... it wont.) This ISO should be set to the same setting as your "light" exposure, but the shutter duration should simply be the shortest possible exposure and the lens cap should be on. In software, you can think of this as the determine the "floor" (lowest possible values) for your real exposures (your "light frames").

Flat frames - these are "light" exposures shot of a perfectly evenly lit surface. The exposure settings do not need to be the same as your "light" exposures but they do need to be shot with the same lens, same focal length, and the same aperture setting (but the ISO and shutter exposure can be different.) Every lens + focal length + aperture combination will produce some vignetting in an image. You might look at an image and think your lens doesn't have any vignetting at all... but technically every lens has some (it's just that it's not enough to notice.) It turns out, however, that most of the data for an astrophotography image (if you look at a histogram of a properly exposed image) is down in the lower 1/3rd to lower 1/4 of the graph. You will then "stretch" the image exposure to tease out all the details. As soon as you start "stretching" the exposure (using curves or levels, etc.) you will also start exaggerating the vignetting... so suddenly a lens that didn't seem to have a vignetting issue now has a horrible vignetting issue. The stacking software can profile the vignetting pattern of your instruments (camera, lens, telescopes, etc.) and use this to "subtract" the vignetting from the frame.

When you stack, DSS will create a profile of your flats, bias, and darks. Lots of flats will create a master flat. Lots of darks will make a master dark, etc.

The software will first calibrate each single "light" frame against these master dark/flat/bias frames to produce a set of "calibrated light" frames.
The software will then perform a star alignment using the stars in each "calibrated light" frame to produce "registered" frames (essentially these are "registered calibrated lights"). Finally, the software will then merge (using some algorithm) these registered + calibrated lights to produce what you can consider a "master light" frame. That single image you get as the output is the image you will then start to with in other software.

Thank you for the detailed reply.

Did I summarize this okay?

Dark frames – leave lens cap on. Take a few shots before star shots, use the same settings as star shots.

Bias frames – leave lens cap on. Take a few shots before star shots, same ISO, fastest possible shutter speed.

Flat frames – same lens, same focal length, and the same aperture setting (but the ISO and shutter exposure can be different).


Do the flat frames need to be taken at the same time? Can I just shoot a piece of white Bristol board in the daylight?
Do I need to shoot the Dark, Bias and Flat frames each night I go out? Or can I reuse them for future shooting?

shane_c wrote in post #18136772
Thank you for the detailed reply.

Did I summarize this okay?

Dark frames – leave lens cap on. Take a few shots before star shots, use the same settings as star shots.

Bias frames – leave lens cap on. Take a few shots before star shots, same ISO, fastest possible shutter speed.

Flat frames – same lens, same focal length, and the same aperture setting (but the ISO and shutter exposure can be different).


Do the flat frames need to be taken at the same time? Can I just shoot a piece of white Bristol board in the daylight?
Do I need to shoot the Dark, Bias and Flat frames each night I go out? Or can I reuse them for future shooting?

I would take the dark frames "after" taking the lights (some people intersperse them). This is because the long exposure images will heat-up the sensor. The warmer the sensor, the more "noise" you'll get. There is a direct relationship between physical temperature of the imaging chip and the amount of noise you get. So if you take the darks "first" (with a camera that isn't yet "warmed up") then you wont get an accurate representation of the amount of noise likely to be present in most images.

Yes, your definitions are correct.

As for flat frames being needed every time....

I sometimes image with an SCT telescope (big 14" mirror). The slightly little shift in the mirror will alter the pattern of the vignetting. Also, depending on the object I want to image, I may rotate the camera. Lastly... I may also use a focal length multiplier or focal length reducer in the telescope and that will alter the vignetting as well.

If you're imaging with a camera lens, then the orientation of that lens to the camera body is the same every time you attach it.

So when imaging with a "telescope" it's recommend that you do re-take the flats to match the camera orientation of your "lights".
When imaging with a camera lens, it's probably not necessary to re-shoot flats every time because they probably don't change much.

You can shoot "flats" by holding up an iPad in front of the telescope or lens, setting it to a plain "white" screen, and then shoot with an exposure to render the iPad screen (which needs to completely fill the frame of the camera) as a "middle" exposure (you want it to appear gray -- not white). You only need a handful of these and they're fast exposures.

You can also shoot the plain blue sky... but there are some tricks to this. The suggestion is to wrap a piece of white fabric (like a pristinely clean t-shirt) over the front of the telescope and stretch the fabric to make sure it has no wrinkles. As the sun is setting in the "west", point the scope to the "east" (so it's facing away from the sun). The backscattering light onto your fabric will make a very evenly lit surface of the t-shirt fabric. The lens doesn't not need to be focused (you wouldn't be able to focus on the fabric when it's on the front of the lens anyway.)

This is an rather reliable "poor man's" flat-field generator. There are expensive light-panels that guarantee perfectly flat illumination that some imagers use.

There are people who try to shoot a plain white "board" illuminated by lights. But the trick is that it's VERY hard to make sure that every square inch of that board receives precisely the same amount of light (remember, once you start "stretching" the histogram to enhance an image, those very subtle variations will show up in very obvious ways.) So there are guys who build their own flat-field generators and and use techniques of lighting the surface that look like a geodesic dome.

For however as many "lights" you choose to shoot, shoot roughly half that many "darks". It's helpful if the number is actually an "odd" number. E.g. if you shoot 25 "lights" of an object, then shoot about 13 darks.