DSLR Exposure Time

Problems with an exposure that is too long:
  1  Sky Glow saturates sensor
  2  Exceeding guidance stability, oval stars
  3  Blowing-out the bright parts of some brighter DSO's
  4  Lost time if a sub fails for any reason (random mount wiggle, cloud, plane…
  5  Not having enough subs for effective dithering
  6  Not having enough frames for Winsorized Sigma Clipping during integration

Problems with an exposure that is too short
  1 Too many subs create excessive processing time
 2  Exposure too short to accurately capture faint DSO objects or features
 3  Excessive disk space consumption
 4   Lost time from delays between subs

CS

Thank you both!

I think I’ll shoot 3 minute exposures next and see how it goes.

The answer depends on a lot of factors. But at the end it can be quite simple. Single subexposure time matters virtually nothing, as long as you stay within reasonable limits, and you are in the read-noise dominated regime. Than only _total integration time matters_.

To determine subexposure time, it depends on:

• Sensor read noise, which depends on ISO (and camera model, of course.) The HIGHER the ISO, the LOWER the read noise. That is an area where you can find misleading or plainly wrong information online, especially coming from daylight photographers. Can sound crazy, but that is the simple truth. Check photonstophotos RN charts. The lower the read noise, the shorter the minimum subexposure. Each camera has some specific "recommended ISO range". For older Canons, 800-3200. For newer Nikon/Sony sensors, about 400. I suggest you with this recommendations.[/*]
• Light pollution - the brighter the sky, the shorter the minimum subexposure.[/*]
• Focal-ratio (f/) The faster the scope, the shorter the minumum subexposure. [/*]
• Filter: if you use narrowband filters, the minimum subexposure time should be a lot longer (and broadband LP filters can also affect the minimum subexposure, albeit much less than narrowband filters).[/*]
• Optical throughput and efficiencies (QE, transmissivity...) - less impactful than other sources.[/*]

The minimum subexposure should "swamp" the read noise, making the shot noise (from object and light pollution especially) (or dark current) the dominant noise term.

For DSLRs, the best rule of thumb is to use the back of camera histogram (the one you see when displaying the image on the LCD). The peak should be somewhat around 25% from the left. Much more to the right (brighter), you are gaining very little benefit, and you might start saturating too many stars.

Once you get to the minimum exposure time - let's say for you it is 1 minute. Then it *does not matter* if you do 60x1min, 30x2min, 20x3min, 6x10min (practically speaking. Longer subexposures will always have a slight, diminishing, advantage). Provided you can track accurately for any amount of time. Losing one 10min frame can be very detrimental, much more than simply using shorter exposures. Of course you should also avoid saturation - you do not want to clip all the stars (but the bright ones will almost inevitably clip), and definitely not clip the DSObject you're shooting.

Usually, most systems under most skies, this number should be between 1min (fast optics, or light-polluted skies) and 5 minutes. For slow systems under dark skies, it can be as long as 10-20min. Add in narrowband filters (with which DSLRs are quite inefficient), and the "optimal" time will probably be 10-60min, which may be very impractical, and you should use shorter, 5-10min exposures and live with the small SNR loss.

At the end it is a tradeoff. Once you find your optimal subexposure lenght, then the integration is what matters.

Best regards,
Gabriel

PS: unless you have very warm nights (20ºC or more), thermal noise should not be a major concern, especially in the beginning. If your night temperatures are below 5ºC, than dark current becomes really negligible. I would recommend against "cooling" the sensor, especially if it means capturing less photons. Integration time is the most important!

https://youtu.be/3RH93UvP358

Let's try that again…..

Gabriel is correct though I think he unintentionally made a mistake when he said that sub exposure time should be in the read noise dominated regime; your subexposure time should be such that the read noise is swamped by the background signal.

Thank you very much for pointing that out! I'm sorry, I definitely misspelled things there - should have read the thing more carefully. I cannot edit the post, so, to be clear:

The subexposure time should *NOT* be in the read-noise limited regime! RN should be swamped by the other noise sources, and you should be in the *SKYFOG* limited regime.

Another errata: "I suggest you start with this recommendations."

Oh, and the emoji was automatically added because of the ).

I used to go with the "30% from the left" histogram rule.
Nowadays I rely on mean ADU values of the RGB channels. It's faster and more reliable then checking an histogram

How long to expose does not only depend on how many stars you're saturating, but also on how many subs you're planning to stack and store at the end of the session.

When I was "newbier" than today (Nov 2020), I shot ~750 subs at 75s at a dark nebulae. Trust me, it wasn't fun (>12h with Xeon 8 core, 32gb ddr4, nvram ssd) to stack them all also storage can become a problem quite fast, if you keep shooting tons of short subs.

In response to that question, Scott, it really depends on the camera.  There is a sweet spot somewhere in the ISO /exposure time dilemma that we all face as DSLR users.  I have not had any banding issues with my 6D and have found that ISO 3200 is the best value for me, allowing three minutes exposures with my L- enhance. I have actually seen my DR increase with these longer exposures and higher ISO, producing cleaner subs for me. It is just so different for each model though.

Scott - I think the reason that banding is ISO dependent has to do with the way Canon sensors are constructed. There are two components to read noise - the noise from the sensor itself and the noise introduced during analog to digital conversion. The latter is quite significant. The higher the ISO, the more the signal from the sensor is amplified before the A-D conversion is done, and hence the lower the contribution from the A-D conversion noise. Generally for Canon sensors ISO1600 and up seem to be sweet spots, though you lose one stop of DR for every doubling of ISO beyond 1600 for older Canon cameras and 400 for the Mark IV.

I have used and continue to use on occasion, a Canon 5D Mark IV (unmodified). I would always get one or two very strong bands, regardless of ISO. Using darks without biases helps get rid of them.

One note note: I use a Canon EOS Ra so I inherently have a higher sensitivity to IR / Ha eith my DSLR over an unmodified camera.