Re: Calculation from scratch; practical measurement
1
Seems like we have converged fairly well to a nice benchmark. Consistent with your and my experience.
OutsideTheMatrix wrote:
So I did a test run tonight and found that 10 sec is perfectly fine, even with the LCD 14x magnification. 13 sec is questionable; in some frames I could clearly see two separate stars, while in others it looked like a single streak. 15 sec and above at 14mm was definitely not acceptable on the E-PL6, by my eyes anyway.
The formula we acquired earlier to find the shutter speed equivalent to the E-520 was SQRT(1.6) I believe, which comes out to 1.265, so 15/1.265=11.86 sec. So my pinpoint 15 sec exposure at 14mm with the E-520 would be 11.86 sec on the E-PL6, in terms of star trail equivalency anyway. Sounds like the "eye test" of the tipping point being somewhere between 10 sec and 13 sec proved to be correct. How much of a star trail do you get with 13 sec exposure at 14mm? I think I'll keep it to 10 sec just to stay on the right side of tolerance.
Sorry, haven't done this specific test, and it's not too likely that I will in the near future. (I have two lenses that I use for night sky shots. My "old favorite" is my 20 mm Panasonic 1.7. My 12-40 Olympus f/2.8 (at 12 mm) is about 3/4 of a stop less efficient (needs more ISO), so I seldom use it.But, I'm confident enough in the "theory" now that I'd go by the benchmark we've come to.
I found the same 7x equivalent to my LCD for 100% viewing, but just to be safe, I also view at 14x; it seems like the 10 sec exposure holds well at either magnification.
Great.
Is ISO 1600 a bit too low for the Milky Way? Try ISO 2500, it keeps the star colors. ISO 3200 may limit the DR too much and stars go a bit pale. Are you using Auto White Balance?
My standard exposure for crystal clear skies (tried out in the Kalahari desert) is:
20 mm, f/2, 30 secs, ISO 200
I was shooting to combine multiple images into star trails, so I did not worry this will leave some visible trail. Turns out, pixel peeping, this looked pretty exactly at the point that essentially no stars were blown, so preserved all the color that's there. I haven't shot the Milky way much, but articles I've seen put good exposure at least 3-4 stops more than this, obviously blowing out the bright stars (which, unless you're shooting for the color you just can't see with your eye, is not a concern). It was just barely possible to get decent Milky Way visibility with the shots I did exposing for colored star trails if I used a lot of post processing. But, too noisy. The night sky has a huge dynamic range.
Pinpoint stars (7.5 secs, with my 20 mm), exposing with +3 stops for the Milky Way puts ISO at 6400. Probably hopeless. But, I imagine 3200, with some post processing might be OK. Oh, wait, I could open my 20 mm another half stop, so 3200 doesn't seem so bad. But, I haven't tried. I did conclude that I need a "better" (shorter and/or brighter) lens for night sky shooting, if it becomes more important for me.
I used 4000 K manual white balance. Read that somewhere.
150/fl is good! Based on the 11.86 sec we acquired earlier for 14mm on a 16mp m4/3 sensor, 150/fl is a really good approximation.
About noise, it seems like our sensors have leveled off since about 2012 or so, noise hasn't gone down much in the newer cameras versus the E-M5/E-PL5 generation.
Yeah, sounds right. May get a little bit better if I go for the E-M1 Mark II. Clearly not worth the cost just for the supposedly marginally better noise level. We're into the weeds with non-negotiable shot noise, so sensor technology cannot get much better without some amazing breakthrough.
So, my current benchmarks.
- Exposure: 20 mm, f/2, ISO 200 -- for non-blown bright stars.
- Exposure: +3 to 4 EV for Milky way.
- Duration: 150 / f for pinpoint stars on large prints; 300 / f for smallish screen display.
Here is one of my star trails, followed by some Milky Way I managed to pull out of the same careful exposure (PP to compensate for several stops underexposure)


It's been great chatting. My night shooting should be on a stronger footing, now.