9

In an answer to question "What lens characteristics are important in astrophotography?" there was mentioned "If the temperature changes drastically, you may need to refocus as different materials in the lens will expand and contract at different rates." But the accepted answer to "How does one focus in very dark conditions?" suggests a tape marking on lens, and wouldn't that turn out to be useless when temperature drops at night?

Is it possible to estimate the focus shift beforehand, when focus was first found in 5 Celsius degrees in the evening and then at night the temperature drops to -10 Celsius degrees? How would I calculate, or estimate, the focus shift between these temperatures?

On the other hand, is it likely that 15 Celsius degrees (27 F) change in temperature really does shift focus so much that I'd need to re-focus? How drastic change in temperature makes refocusing typically necessary?

In my case it is a small size dslr-like camera with a 14mm lens that has somewhat large front element and the lens weighs more than the camera body. And focus would be at infinity for stars.

Community
  • 1
Esa Paulasto
  • 5,984
  • 7
  • 37
  • 83
  • 1
    Each lens design will react to temperature changes differently. I generally check focus several times during an astrophotography session as much to be sure that I haven't accidentally changed focus as for any other reason. Once you are close, as you would be with a change in temperature, it is relatively easy to turn on LV, zoom in on a medium star and check the focus and correct slightly if necessary. A lot depends on just how in focus you want your stars to be. Sometimes slightly OOF looks more realistic as the brighter stars look larger to our eyes. – Michael C Dec 09 '13 at 19:19
  • @MichaelClark - I feel you have enough already in your two comments to post an answer too, no? The accepted answer to How does one focus in the dark? recommends a tape marking on lens, and actually that is why I'm asking this one here. – Esa Paulasto Dec 11 '13 at 09:43
  • 1
    A tape marking gets you close enough to start the process. When focusing on dim point sources of light such as stars, nothing shows up in the viewfinder or even via Live View until focus is almost achieved because the miniscule amount of light is spread out too far to be detectable when the lens is too far out of focus. Until you are fairly close to infinity focus you can pan the entire night sky and not see anything to focus on (unless the moon is visible). – Michael C Dec 13 '13 at 17:24

4 Answers4

7

From my reading, the amount materials within a "handheld" lens expand / contract over, for example, 15C is so minimal it's not worth thinking about.

However this really becomes an issue on large telescopes, both refracting and reflecting (even more so).

Why?

As a (rough) example, let's imagine a large reflecting telescope which has a body length of 3m, and is constructed from tubular aluminium. Pure aluminium pure has a linear thermal expansion rate of 0.0000222m/m/K, which means it gets longer / shorter by 0.0222mm per degree Kelvin (or C) per meter of its length.

Therefore the telescope would shorten by 0.0222mm × 3m × 15C = 0.999mm as the temperature drops 15C. This coupled with the magnification at the secondary mirror leads to a dramatic focus shift.

Digital Lightcraft
  • 9,060
  • 3
  • 31
  • 69
  • 3
    You are correct that the physical amounts of contraction/expansion are very small. But the effect those amounts have can be significant. If a camera's mounting flange is off by even as little as 50µm (50 microns or 0.05mm) it will make the camera unusable for any type of critical focus work at large apertures. But the effects can be seen with as little as 20µm of error. Astrophotography, where you have extremely small point sources of light across the entire field of view is the most focus critical work most photographers ever do. – Michael C Dec 10 '13 at 14:04
  • Your telescope example ignores the effect of temperature on the shape of the mirror and other optical materials as well as the length of the tube or trusses supporting them. – Michael C Dec 13 '13 at 17:26
  • @MichaelClark - yes 2 reasons: I have no way of calculating that, and secondly there is a huve variation in materials, shape etc of the optical elements. – Digital Lightcraft Dec 16 '13 at 09:21
  • Isn't there also a huge variation in the materials, shape, etc. of the optical elements as well as the bodies housing them when speaking of camera lenses? – Michael C Dec 16 '13 at 20:26
  • yes... so its basically impossible to calculate without an extremely advanced computer model (which will probably be wrong anyway when you factor in grease/lube, tolerances, wear etc) – Digital Lightcraft Dec 17 '13 at 10:26
  • So since you can't calculate something, then it must not exist? That seems to be what you are saying. – Michael C Dec 17 '13 at 17:21
  • no... what im saying is, it cant be calculated, so im not going to try. – Digital Lightcraft Dec 18 '13 at 09:19
  • If you can't calculate it, how can you be certain it is "...so minimal it's not worth thinking about?" – Michael C Dec 18 '13 at 09:20
  • Thats like saying "if you cant PROVE there isnt a god, there must be" - I'll leave it to you to calculate and somehow visualise the effect of a 15C temperature variation on a "14mm lens" - the linear thermal expansion ratio of hard glass is between 4 and 5.9 x 10^-6, and aluminium 22.2 x 10^-6 - :-) – Digital Lightcraft Dec 18 '13 at 09:29
  • No, it is not like your "if you can prove.." quote which relies on a rhetorical argument without any chance of observing an entity that by definition transcends the observable universe. Because even if you aren't capable of predicting the exact amount of change using calculations based on linear specifications and thermal expansion ratios, etc. you CAN observe the effects on an actual lens under the conditions specified. Although the OP does mention a 14mm lens "in my case", the general nature of the way we treat questions here would also allow for a much longer lens. – Michael C Dec 18 '13 at 09:36
2

I think you can do some experiment on your own. I would take several snapshots (with camera,settings, focus and light blocked to the same values) and with different temperatures of a slanted edge standard and measure the MTF with ImageJ and MTF plugin or with Imatest. Then you can plot the MTF graph with different temperatures and see the result.

I think the different expansion coefficients aren't the only causes of this change, materials have different refractive index with different temperatures and I think you should take in account that if your lens become very cold condensation could be formed in the optics because inside your camera is warmer than outside.

G M
  • 413
  • 2
  • 15
2

Focal length rises about 0.7 promille in the temperature range -10 to 20C, and the day vs night / temperature focus position in Hubble is carefully modeled (and they are not simple relations "focus vs Temp") and is in the range of 5-7 microns. But still the deviation of those models are pretty large. One thing is temperature, another is temperature changes, and the spreading of that change into the components over time, and then there's extended exposure to those temperatures.

check this out and this and this

Michael Nielsen
  • 10,644
  • 35
  • 48
1

Each lens design will react to temperature changes differently. I generally check focus several times during an astrophotography session as much to be sure that I haven't accidentally changed focus as for any other reason.

A tape marking gets you close enough to start the process. When focusing on dim point sources of light such as stars, nothing shows up in the viewfinder or even via Live View until focus is almost achieved because the miniscule amount of light is spread out too far to be detectable when the lens is too far out of focus. Until you are fairly close to infinity focus you can pan the entire night sky and not see anything to focus on (unless the moon is visible). Once you are close, as you would be if you were properly focused before a change in temperature, it is relatively easy to turn on LV, zoom in on a medium star, check the focus, and correct slightly if necessary. A lot depends on just how in focus you want your stars to be. Sometimes slightly out-of-focus looks more realistic as the brighter stars look larger to our eyes.

The physical amounts of contraction/expansion in the lens due to temperature changes within the range of your question are very small. But the effect those amounts have can be significant. If a camera's mounting flange is off by even as little as 50µm (50 microns or 0.05mm) from one side to the other it will make the camera unusable for any type of critical focus work at large apertures. And the effects can be seen with as little as 20µm of error. Astrophotography, where you have extremely small point sources of light across the entire field of view is often the most focus critical work many photographers ever do.

Michael C
  • 175,039
  • 10
  • 209
  • 561
  • This also makes me start considering AA filter Vs none... – Digital Lightcraft Dec 12 '13 at 10:38
  • Cameras designed with an AA filter that is subsequently removed need a spacer to keep the sensor-to-flange distance the same as before the removal of the AA filter, if that's what you are getting at. – Michael C Dec 12 '13 at 17:45
  • no, actually i was thinking about the quality of focus of point sources (stars), AA is there to slightly blur the image, so is non AA preferred for astro? – Digital Lightcraft Dec 13 '13 at 15:46
  • It depends. For most photographers using most cameras probably yes. But if you are using certain Nikons or similar cameras if a point source is small enough (only lights up a single pixel?) the A/D conversion will discard it as noise. This why the D300S and D7000 are known as 'star eaters', even with an AA filter in place. – Michael C Dec 13 '13 at 16:24
  • It seems that one reason the Nikon sensors consistently outscore the Canons at DxO Mark is because Nikon applies some NR on the sensor itself prior to saving the RAW file whereas Canon leaves all NR until after the point in the process where the RAW file has been recorded. – Michael C Dec 13 '13 at 16:27
  • Related to your "star eater" comment above: 4 years on, and it looks like companies still manage to create "star eater" situations. In late 2016, Sony pushed out a firmware update to their a7SII and a7RII cameras (The Star Eater Issue: Why I No Longer Recommend Sony Cameras for Astrophotography at petapixel.com), that included a more aggressive (and slightly brain-dead) noise reduction and hot-pixel removing spatial filter. It wound up turning otherwise great cameras into star eaters. – scottbb Sep 26 '17 at 01:28
  • 1
    @scottbb It will be interesting to see how Canon's new sensors, which seem to use more on die NR than in the past (because apparently that is what boosts a score at DxO) will affect their astro capabilities. In the past Canon sensors were regarded as the ones least likely to eliminate weak stars as noise. – Michael C Sep 26 '17 at 01:38