Is it worth taking bias frames?

[ianpwilliams]

My thoughts (rightly or wrongly) have always tended to be that the harder the frames are to take, the more useful they are. So flats are arguably the most useful (and are a massive pain to do), then darks (which are easy but inconvenient), and then bias (which are really easy because you can do them quickly and easily indoors). And so I wondered how useful bias frames actually are, and this thread seems to suggest that they might not be worth doing at all:

http://stargazerslounge.com/topic/92031-how-to-take-bias-frames-on-a-dslr/

Any opinions on whether bias frames are worth including? I know they are easy to do, but if they really aren't going to make a noticeable difference then it doesn't seem worth it.

[dph1nm]

If you have darks you don't need bias, as the bias is in the darks. But you will have to take two sets - one for your lights and one for your flats.

NigelM

[steppenwolf]

Bias frames are so easy and quick, it would be rude not to use them surely?      Bias frames can make a huge difference if there are any bad columns or other deficiencies and you will require them to calibrate you valuable Flat frames....

[Waldemar]

 "So flats are arguably the most useful (and are a massive pain to do)"

I use an EL ( Electro Luminance) panel for taking flats... works great! very easy too.

and they work with any type of filter

[MartinB]

Darks contain the bias component but this only works if your dark frames are the same length as your lights. If you don't have temp regulation, e.g DSLRs it can be better to scale your darks so that they can be gathered more quickly ensuring they are taken at a similar temp to lights. In this case you might take a 6 minute light and a 3 minute dark. For this to work you must substract the bias from the dark before (because you don't want to scale the bias!) So you need bias frames to do this. Having done that you will then need to use bias frames to subtract bias from your dark subtracted light.

Another scenario - you've got some nice flats and want to apply them to your dark subtracted lights. You need to remove the bias from the light frames before you do this. You can take flat dark frames to dark subtract your flats or, if you are lazy like me, you can bias subtract them.

One thing to be clear about, it is very worthwhile to remove bias!

[ollypenrice]

Lots of DSLR users are giving up on darks and just using bias instead. The remaining noise is dealt with by dithering at capture. Darks can be quite invasive and sometimes do more harm than good. I'd try 'bias as dark' plus dither. As others have said, it's important to dark subtract your flats (or they'll over-correct) and a bias is a perfectly good dark for this purpose.

Even with set point cooled CCD I no longer use darks. I use a bias-as-dark and a bad pixel map. I find this gives me consistently cleaner stacks.

Olly

[ChrisLX200]

A bias is just zero-length Dark (or as near zero as you can get) so it accounts for all the electronic noise associated with the task of reading out the image data from the sensor. So yes, use them - they're easy and quick to capture, but use a lot of them (maybe 50) to get a good average.

ChrisH

[Vox45]

 "So flats are arguably the most useful (and are a massive pain to do)"

I use an EL ( Electro Luminance) panel for taking flats... works great! very easy too.

and they work with any type of filter

Are you refering to something like this ? If so, It's flexible, spillproof and runs on 12V ... you can even cut it into any shape you want using scissors !

That is a great idea ! Better than a t-shirt and an iPad

[r3i]

I do dithering during image capture and use a bad pixel map rather than darks so the bias frames are worth having in this scenario.

[ianpwilliams]

OK thanks, looks like I should stop being lazy and get some bias frames taken! Apart from anything, I was taking long lights the other night which didn't leave too much time for darks, so maybe some bias frames might help with that.

I certainly want to look into dithering though, once I've gotten a few more sessions done. It's so frustrating taking darks, so it would be great to dither and have all that extra time for lights.

I don't find flats all that troublesome really. I use the laptop with a sheet of paper over the screen, and I've gotten it so that the histogram is always correct. The pain comes from holding the laptop against the telescope, and usually forgetting to put my gloves on first, so my hands end up like ice!

[cjdawson]

Quick answer - yes it's worth taking bias frames.

long answer (it's longer than I was expecting to write)

For a long time I was in the camp of why bother with any of it?   Then I did some research and understood the different image types.   The different types of image all play a part with a complete set you should be able to produce a better final image than having not done them.  My current understanding is that there are the following types of image.

Lights - These are the exposures on the object.

Darks - Exposures of the same length, with the camera in the same condition as for the lights, only with the lens cap on - or more to the point with no light entering the camera.

Flats - An exposure of a flat colour, I do this with the camera set to Aperture Priority.  Pointed at my EL Panel and I let the camera sort out the imaging time (I should experiment to find out what this is)

Dark Bias - These are the same as Dark frames, only the camera is set to the fastest exposure setting that it can do (8000th of a second for my Canon 70D)

Flat Bias - These are the same as Dark bias, but with the camera pointed pointed a the EL Panel.

So, what do these images do?   As the subject of the post is Bias frames, let's start with that one.

Dark bias - These images, show the minimum value that each collector can register. by subtracting this from every other frame, you are smoothing out the inaccuracies in the camera sensor.  You should apply this subtraction to the lights, and Darks.

Flat bias - These images do the same job as the dark bias, but should be applied to the flats only.

The analogy that I'm thinking of for the effect of bias frames, is when you look at a lake.  On a windy day, there are lots of ripples on the surface of the water.  If you imaging that the peaks and troughs are the minimum value that each collector in the frame will register, the bottom of a trough is a 0, and the top of a peak is say a 10.  As you take the image, those two pixels will gather light, the one at the bottom of the trough collected 10 photons, so it now reads 10, the one at the peak also gathers 10 photons, but it now registers 20.  The bias frames will sort this out, so that both collectors (or pixels if you prefer) read as 10 on the cleaned up image.    Do this for the entire picture and the quality of the fine detail will be easier to extract later.

Flat frames - when you take these images, they don't look "flat" like the name, actually they look more like a big circle with dark corners.

So, what is a flat frame?  This a picture of an even light source.  That's the important part here, the light that enters the optical train should be even or flat.  As the light passes through, it will be bent, refracted, reflected or manipulated by the optical system.  The result of all that is that more light falls on some parts of the sensor, than others.  Normally the centre is the brightest, and the edges will be dark.  The image will have a gradient between the two.  This represents the amount of light that is captured by that part of the image.   In addition, any imperfections in the optical train will also show up on the image. The imperfections can be dust motes, water marks, etc.  Or more to the point, the image registers the effect of the artifacts.  By taking this image, you can use it as a source to manipulate the light frames in a way to counter the effects of these artifacts.   So it will help to reduce the effects of dust, Vignetting etc.  Again this gives an opportunity to produce a better final image.

Dark frames - oops, I left the lens cap on.

The point of a dark frame is to capture the imperfections in the camera sensor.  Hot pixels, cold pixels, amp glow  things like that.  They are not the same as bias frames at all, as the exposure time is the same as for a light frame.

So why do you need a full set?

So, let's look as the process of processing an image, like a program like Deep Sky Stacker does.

Start with a light frame, now subtract the dark frame.   This removes the hot pixels.

Next job is to subtract the dark bias, is has the effect of smoothing out the ripples in the image.

Now, let's take the flat frame, and remove the flat bias.  again is smooths out the camera ripples

And finally combine the modified flat and light.  This removes the vignetting dust motes etc.

All these steps help to improve the image, and this is why Deep sky stacker and other programs have inputs for all the steps.

[Darkstar_1]

So for flat frames could one use a 20% grey card ( from old days of film cameras) ?

[ollypenrice]

On 21/1/2016 at 23:07, cjdawson said:

Quick answer - yes it's worth taking bias frames.

 

Flat Bias - These are the same as Dark bias, but with the camera pointed pointed a the EL Panel.

 

Flat bias - These images do the same job as the dark bias, but should be applied to the flats only.

 

 

I'm lost. I've never heard the term 'flat bias' and can't see why having the camera pointed at the panel would be anything other than a flat. Maybe this arose from a typo or copy-paste slip up?

 

A flat is an image taken of an evenly illuminated source. Like any digital image it needs to be dark-frame subtracted. However, because flats are from very short exposures there is no significant difference between a dark replicating the precise capture settings of the flats and a bias, so you might as well use a bias as a dark for your flats.

 

46 minutes ago, Darkstar_1 said:

So for flat frames could one use a 20% grey card ( from old days of film cameras) ?

You could, but the problem is illuminating it evenly. Some imagers find that the insides of their observatories give satisfactory flats. Mine don't so I use a panel.

Olly

[cjdawson]

@ollypenriceI'm sure you've heard of a dark frame, and a bias frame.   The Flat bias is to the flat frame what the bias frame is to the dark frame.

 

The difference between a bias frame (proper name is a dark bias frame) and a flat bias frame, is that the flat bias frame is pointing to the flat field rather than the dark field.

 

If you are still confused, let me try to explain this another way.   When you take a dark bias frame, you are taking an image of a dark field of view.  This means that each light collector on your camera sensor registers the minimum value that it can register for a dark image.  To take a Dark or Flat bias frame, you have to set the camera to it's fastest setting, for my Canon 70D that's about one 8000'th of a second.  However, that is still a photo, as the shutter is open and gathering light for that amount of time.  For a dark bias frame this isn't a problem, as the image is black.

A flat bias frame is taken in almost exactly the same way as a dark bias frame.  The only difference is that the 1/8000 exposing is pointing at the flatfield instead.  So this time the light collectors may collect the minimum amount of light from the flat source.  When you process the flat field images, you should use the flat bias instead of the dark bias, in order to properly clean the signal caputred in the flat field images.  The dark bias frame won't do, because they don't capture the minimum amount of light on the collectors that would have been captured in the first 8000th of a second when taking the flat image.  Therefore, a flat processed with a dark bias frame will not output correctly - it will be closer than not using a bias at all, but it won't be as close as when using a flat bias.

[Pompey Monkey]

16 minutes ago, cjdawson said:

@ollypenriceI'm sure you've heard of a dark frame, and a bias frame.   The Flat bias is to the flat frame what the bias frame is to the dark frame.

 

The difference between a bias frame (proper name is a dark bias frame) and a flat bias frame, is that the flat bias frame is pointing to the flat field rather than the dark field.

 

If you are still confused, let me try to explain this another way.   When you take a dark bias frame, you are taking an image of a dark field of view.  This means that each light collector on your camera sensor registers the minimum value that it can register for a dark image.  To take a Dark or Flat bias frame, you have to set the camera to it's fastest setting, for my Canon 70D that's about one 8000'th of a second.  However, that is still a photo, as the shutter is open and gathering light for that amount of time.  For a dark bias frame this isn't a problem, as the image is black.

A flat bias frame is taken in almost exactly the same way as a dark bias frame.  The only difference is that the 1/8000 exposing is pointing at the flatfield instead.  So this time the light collectors may collect the minimum amount of light from the flat source.  When you process the flat field images, you should use the flat bias instead of the dark bias, in order to properly clean the signal caputred in the flat field images.  The dark bias frame won't do, because they don't capture the minimum amount of light on the collectors that would have been captured in the first 8000th of a second when taking the flat image.  Therefore, a flat processed with a dark bias frame will not output correctly - it will be closer than not using a bias at all, but it won't be as close as when using a flat bias.

So are you saying that the bias is affected by the light on the sensor?

If that's the case, then you would surely need to take a set of bias frames for every single light frame, as these will be affected by the captured image, making the other bias frames invalid???

Bias (in AP) is the noise induced by the camera on the image due to the process of transferring the signal from the sensor to the memory. The amount of light falling on the sensor has nothing to do with it.

[ollypenrice]

I'm intrigued to read about these flat bias frames because I've never heard of them, ever, on any forum or in any book. The standard advice is to take a set of flats exposed to about a third of the chip's full brightness range and calibrate them by subtracting either a dedicated dark taken with the same settings or, more easily, a master bias (since in short flats exposures there is no build up of thermal noise.)

 

When teaching AP here I often demonstrate the effectiveness of  flats by taking one master flat and dividing it by another master flat just as the software will divide a light by a flat. When I do this I get a perfectly flat image, just as I should. I see no evidence to suggest that there can be anything wrong with my flats.  They seem to work fine on my lights, as well, but the test of 'flattening a flat' is more exacting and quantifiable.

 

So if I take an image of a flat panel at my shortest exposure time how bright should the panel be? This has to matter because if the panel were bright enough it would saturate the chip! Could you supply a link to your source of information regarding what you call flat bias? According to Wiki,  A flat-field consists of two numbers for each pixel, the pixel's gain and its dark current (or dark frame This agrees with the method I describe. The gain is the signal from the panel and the dark current is what will be subtracted by the dedicated dark for flats or the almost indistinguishable bias. I'm at a loss to understand your third number, the one generated by taking a short exposure flat.

 

Olly

 

 

[cjdawson]

Hang on guys.   Olly has contacted me in a PM with a query about this. Looks like I got a mixed up a little.  From what I've read so far, I think it's safe to say that the "Flat bias" that I was talking about in previous posts is actually no different from an "Bias" frame.  So forget that I mentioned that.

 

As for Bias frames making a different, I still do believe that they do make a difference to the overall image.  I stand by what I said about Bias frames. A bias frame shows the minimum value of each collector (or pixel if you light) in your sensor. So by removing this value, or offset from each pixel, you'll end up with an image that is more consistent.

[ianpwilliams]

This link here:

 

http://dslr-astrophotography.com/shoot-bias-frames-dslr/

 

suggests that you need at least 50 bias frames to make it worthwhile (certainly from looking at the graph), so I plan on doing that.

 

Obviously I plan on using my camera remote timer to take the bias frames for me. But what kind of gap should I leave between each frame? I believe someone mentioned leaving 20 seconds between light frames in order to let the sensor cool. Does 20 seconds sound reasonable for gaps between both light and bias frames (and darks too for that matter)?

 

[ollypenrice]

My understanding is that bias frames are not temperature sensitive. Perhaps one of the electronics buffs would confirm or deny this? It might need a separate question in the Imaging Discussion section. 

Olly

 

 

[ianpwilliams]

Good point. You can do them indoors at any time after all (that's what I've always been told). Which would suggest that you wouldn't need a gap. Or at least not a large one.

[cjdawson]

I think from a purist point of view it will make a difference.  But in practical terms it's probably ok to take them anytime.

[ianpwilliams]

OK thanks, will go for 50 bias with a small gap (maybe a couple of seconds)

[ollypenrice]

I've asked the man who knows and will update you with his reply.

Olly

[johnrt]

I have always read that bias frames are not temperature sensitive and so no real need to leave a gap between them, and especially as they are such short duration subs. I think the target of aiming for 50 minimum is a good plan, I normally make 100 and stack to form a master. You can do them at your leisure, so why not take the most you can?!

[Scott]

I'm getting a little confused now so I'll go through dss as I don't want to be doing anything silly

so open picture file is my subs...straight forward

dark files ... an image of same time and temp but with the lens capped

flat files...a file using (in my case a light panel) very short

dark flat file...?????????????????????  I don't use these...should I?

offset/ bias....an image with lens capped at fastest possible speed.

so my question is, a) whats the difference between a dark flat file and an offset/bias file?

Oh, I should mention I use a ccd camera