Dark frames with IQ backs

[gerald.d]

So I've been doing a lot of shooting with the IQ250, IQ180, and Achro+, and with all due respect to what Doug has said in an earlier thread, it is absolutely not the case that a dark frame is shot for every exposure, equal to the length of the exposure.

Two observations strike me as odd. Firstly, if you're shooting a long exposure on a tech camera with a manual wake-up, the dark frame exposure does not equate to the exposure time. It is the exposure time plus the time between wake-up and firing the shutter.

So for example, if I shoot a 1 second exposure, but take 3 seconds between waking up the back and taking the shot, the dark frame is exposed for 4 seconds.

Is that correct/expected behaviour? Is it because the back is actually "live" as soon as it is woken up, and so my 1 second exposure is not 1 second at all, but 4 seconds?

If that is the case, and it makes sense to subtract the noise from the dark frame over the full 4 seconds, then why - if my shutter speed is, say, 1/500th, and I still take 3 seconds between wake up and exposure, is there not a 3.002 second dark frame shot for subtraction?

Second observation.

I have shot thousands of frames now with the IQ250. Even on 5 second exposures, I would estimate that perhaps less than 1 in 20 exposures are accompanied by a dark frame.

This goes totally against what has been explained before. Is it an IQ250 thing?

Given my reasons for having the IQ250, I really would appreciate an explanation/understanding of what the logic is for firing off a dark frame.

Kind regards,

Gerald.

 

[Pemihan]

I'm interested in this as well. I some times shoot with at RZ67 Pro II D with mirror lock up and cable release to the lens. The behavior is exactly as Gerald describes it when using manual wake-up..

Peter

 

[kdphotography]

I'm still trying to digest that you have three MFDBs. I mean, three really really nice MFDBs.

You don't happen to also drive a Land Rover do you?

Sorry, Graham---couldn't resist.

:ROTFL:

 

[gerald.d]

I'm still trying to digest that you have three MFDBs. I mean, three really really nice MFDBs.

You don't happen to also drive a Land Rover do you?

Sorry, Graham---couldn't resist.

:ROTFL:

The IQ180 and Achro+ are mine, IQ250 work's, but pretty much exclusively for my use on a long-term project.

Every morning when I wake up, I remind myself I am a very, very, lucky guy

 

[Pemihan]

BTW as I understand it, if you shot more than one exposure with the same shutter speed the back only make a dark frame for the first shot and then use that for the other exposures as well.

Peter

 

[dchew]

I swear my 180 acts differently since firmware 5.xx. I don't think it used to include the second shot wake up delay in the dark frame, but now it does exactly as you describe. If I have a 1 sec exposure but take 3 seconds to hit the shutter it does a dark frame for 4 seconds.

Dave

 

[Steve Hendrix]

As long as you don't change your shutter speed, the dark frame will only be needed for that first capture. You can test this easily - just set up a 5 second capture, take the shot, watch it count down, release again, no dark frame. Change the shutter speed, dark frame. Which totally makes sense. Just needs to know how long the sensor is exposing for.

As far as the time delay between a wake up and a release, that is normal. Perhaps there is a limit to what the dark frame counts to in terms of .0002, etc. The question is does it make any difference that the dark frame might be 5 seconds for a 1 second exposure, and probably not. If you take a 4 minute exposure with a wake up and wait 4 seconds after the wake up, the extra 4 second dark frame probably doesn't make any difference. Even if it is a larger ratio - say 1 second and the dark frame is 4 seconds plus 1 second for 5 seconds, it may not make any tangible difference. Easy enough to check - wake up your back, take a quick shot right away, then do the same but wait the 4 seconds. See if you can see any difference.


Steve Hendrix
Capture Integration

 

[ondebanks]

So I take it that, since Phase One DB owners have to ask these questions publically, none of this DB behaviour is documented in any sort of user manual?

Ray

 

[dougpeterson]

As long as you don't change your shutter speed, the dark frame will only be needed for that first capture.

This is not always true.

The back monitors the temperature of the sensor and will do a new dark frame if either the exposure is judged to be problematically different in length or too different in temperature than the most recently captured dark frame. Since the iq250 sensor generates less heat than previous phase one sensors it less often has to do a new dark frame for repeated exposures of the same shutte speed.

The exact behavior is frequently tweaked in firmware updates based on continued research and development.

 

[stephengilbert]

User manual? What's a user manual?

 

[Steve Hendrix]

This is not always true.

The back monitors the temperature of the sensor and will do a new dark frame if either the exposure is judged to be problematically different in length or too different in temperature than the most recently captured dark frame. Since the iq250 sensor generates less heat than previous phase one sensors it less often has to do a new dark frame for repeated exposures of the same shutte speed.

The exact behavior is frequently tweaked in firmware updates based on continued research and development.

Thanks Doug for expanding on my answer and providing the additional information.



Steve Hendrix
Capture Integration

 

[gerald.d]

As far as the time delay between a wake up and a release, that is normal. Perhaps there is a limit to what the dark frame counts to in terms of .0002, etc. The question is does it make any difference that the dark frame might be 5 seconds for a 1 second exposure, and probably not. If you take a 4 minute exposure with a wake up and wait 4 seconds after the wake up, the extra 4 second dark frame probably doesn't make any difference. Even if it is a larger ratio - say 1 second and the dark frame is 4 seconds plus 1 second for 5 seconds, it may not make any tangible difference. Easy enough to check - wake up your back, take a quick shot right away, then do the same but wait the 4 seconds. See if you can see any difference.


Steve Hendrix
Capture Integration

This is absolutely key though to the logic/methodology of the dark frame.

Logically, it makes sense to me that a 1 second exposure would require a 1 second dark frame so that the noise for the 1 second can be subtracted.

So far so good.

If I shoot with a manual wake-up, and wait 4 seconds before firing the shutter for a 1 second exposure, then the back WILL shoot a 5 second dark frame. (note - this is assuming a dark frame is required - Doug's explanation that the back decides whether or not one is needed does make sense).

That behaviour makes sense if it is correct that during the period between wake-up and exposure, the back is "live" and collecting noise.

I can understand the logic of the behaviour so far.

What I cannot understand is that IF the above is correct, then why when shooting a short exposure - again with a 4 second delay between wake-up and exposure - does the back not need a 4 second dark frame?

Both behaviours cannot lead to correct dark frame noise subtraction. Only one can be right.

Either noise is being generated during the wake-up to exposure time, or it is not. Therefore, either there is noise from that period that needs to be subtracted from the exposed image, or there is not.

It's a real shame that this behaviour cannot be manipulated by the end user. If you think about it logically, there's no reason why you couldn't create a library of dark frame shots in exactly the same way you can create a library of LCC shots.

OK, so perhaps things are a little more complicated than just what temperature the sensor is at, and what the exposure time is, but given the apparent inherent "inaccuracy" of the existing methodology, I would have thought that a "sensor temperature = x, shutter speed = y, therefore use dark frame library shot z" would be good enough?

Kind regards,

Gerald.

 

[tjv]

Do the IQ backs record operating temp in the EXIF? If they do, and by Doug's explanation it seems possible, then one should be able to generate a library of approximate dark frames, as Gerald suggests, surely? Seems far better to give users the option and state the limitations than limit end use.

 

[Paul2660]

On the Dark Frame for Phase One, I have always noticed the lag that Gerald mentioned since I started using the Tech camera solution.

I don't know if the temp is recorded, but I have exiftool and will look at a few files. I don't think it shows in the metadata in Capture One.

As for what the Dark Frame does, I thought I knew, (mainly map out stuck pixels and excessive noise) but after looking at the files from the DT Library shoot, I am not as sure. At least the IQ260 shots had considerable stuck pixels in them and they were I believe only 6 seconds long. In C1 the "single pixel" noise slider will take most of the stuck pixels out, without compromising too much detail. However if they are not being taken out in the algorithm from the dark frame, not sure what it's doing.

I have not taken my 260 to anything longer than around 6 seconds, I kept planning on getting a 45 minute shot over the winter in the cooler temps with less humidity, but never got around to it. I have around 100 or so P45+ images between 30 min and 1 hour and they all loaded clean, in C1, no stuck pixels, so I naturally assumed that the Dark frame was working to remove them on the P45+ and now the IQ260.

In the world of 35mm DSLR's I have taken thousands of night exposures. I can attest to both Canon and Nikon's implementation of the "long exposure noise reduction", which is basically a dark frame for the same length of the previous exposure. On both of these cameras, the stuck pixels are mapped out very well, I would say close to 95% on Canon and more like 90% on Nikon, this is true in raw's or in camera jpg. Since I stack, I shoot 100% of the time with "long exposure noise reduction" off. Canon allows the dark frame to run in the buffer in the background, but eventually you fill the buffer and are locked out for a while, while Nikon works like Phase One, shoot and then the complementary dark frame locks you out of shooting until complete.

On the need for the dark frame after each shot, for stuck pixels, I can see the need. In my experience, the same pixels tend to stay stuck, until the camera is turned off for a period of time. So for brief periods it might be OK, but longer periods, not.

It would be very interesting to see an IQ250 shot for say 2 minutes without dark frame work. I am curious as to how many stuck pixels are there vs noise.

Stuck pixels seem to get worse over time, i.e. as the camera ages. They also tend to show up after the back or 35mm has been on for about an hour (in winter) and around 15 minutes in warmer weather.

Paul

 

[GrahamWelland]

As your camera ages and/or you travel a lot by air, the CCD/CMOS of the sensor and various chips can get affected by cosmic rays etc. Its one contributor to pixel behavior over time. (Not to mention what it does to you sitting in an aluminum tube at 30,000ft for hours at a time).

Btw, before someone asks, don't bother wrapping your MFDB in foil ... :ROTFL:

 

[jlm]

you need about 10' of concrete

google cosmic rays; evil buggers