Camera and Enfuse settings for the Moon

Options: Exp 1.0; Cont 0; Sat 0.1; Mu 0.45; Sig 0.13; Grayscale Proj: l-star

For a wider range of subject that includes foreground and sky, I have landed at Exp 1.0; Cont 0; Sat 0.0; Mu 0.6; Sig 0.1; Grayscale Proj: l-star

Thank you

I could not have done this without the help of the generous folk out there!

• Chris Spiel (cspiel) for his insight and recommendations on the hugin-support group
• The gang who have contributed to Hugin, Enblend, Enfuse (panorama and HDR stitching/blending): sourceforge
• Ingmar Bergmark for EnfuseGUI

Trade-offs

Unless you are running 2 or 3 cameras, high-end lenses, professional image processing and time-lapse software, you will have to choose your settings carefully to avoid critical problems:

• to get minimal trailing, you must accept noisier images from the higher ISO settings
• time-lapse or still: you can't suddenly zoom or change the composition in mid time-lapse then return - it's too jarring
• basic, "package" lenses tend to create images that are a little soft, so increase the f/ratio and bump up the ISO
• stills give more leeway for the Moon to move between exposures, you can remove the background and align in post-processing
• If your foreground is not changing, and there are no clouds, you can shoot the Moon and foreground separately, then composite in post.
• Wider angles allow longer exposures/lower ISO settings and longer intervals, but the Moon is smaller
• Close-ups can be spectacular, but time-lapses demand shorter intervals and individual frames shorter exposures to prevent trailing. Some cameras will not allow a set of 6 images in 5 seconds with an interval of 6 seconds and still preview the last image before the next set is triggered.
• Some cameras may not allow rapid fire bracketing while in direct view (mirror up), so you can't track how things are going
• Ideal HDR enfuse settings before sunset are going to be different than twilight and darkness - unless you are a sucker for punishment, accept "all-round" settings to cover the entire holy grail time-lapse.
• Fewer images in a bracket will force you to be more careful and use ramping. A 7-bracket set covers you from day to night!

Camera settings: a 6-7 bracket set at 2EV change, f/6.3, ISO800 from 1/2000 to 1.6s every 6 seconds. Refine as needed based on trade-offs.

Once the full Moon is more than 10 degrees above the horizon it is close to maximum brightness/whiteness. Classic recommendation is ISO800 1/500 at f/16, which is equivalent to ISO800 f/6.3 1/3000; Some extinction in the lower atmosphere allows for the slightly longer exposure of 1/2000.

This adds up to 5-15Gig of images for a 2-hour session! Oh, and at the moment I shoot 5184x3456 in JPG, because the camera cannot download RAW fast enough.

In a pinch, one could probably do a bracket of 5 at 3EV, but then the Moon at some point is 1.5 EV from an ideal exposure.

Enfuse settings: exposure 1; saturation 0, contrast 0, mu 0.2 to 0.25, sigma 0.2 (for full Moon daylight to darkness) . The mu = 0.25 works if you have a really short exposure.

This is a tad dark, but in post you can gamma 1.3, contrast 5, and light 3 to render nicely the dark images, and barely affects the light images.

This also works well : Exp 1.0; Cont 0; Sat 0.0; Mu 0.6; Sig 0.1; Grayscale Proj: l-star; The smaller sigma minimizes the use of the bright end when mu is medium

It would be nice to have some form of gamma built in, to minimize the whites blowing out while bringing up the darks. Perhaps this is a particularity with astronomical subjects.

Over the Canada-Day long weekend I spent a considerable amount of time playing with the enfuse parameters using Erik Krause's droplets. I had badly wanted to prevent the overexposed parts from contributing to the final enfuse, so I spent a lot of time varying one parameter while holding the rest constant, starting off in a minimalist way. A big part of the problem is that the lunar highlands are the brightest things in the image, apart from city lights, so when I use exposure-cutoff, I am up at 240 out of 255 anyway. Then I discovered what Chris (see below) had told me from the start - I need to include shorter exposures! When I turned the mu down to 0.2 and sigma at 0.2, I am effectively pushing the weight of the brightest pixels to very low because they are in the tail of the bell curve of the distribution, so it does not matter whether they are cut off or not, they don't get used - which explains why I couldn't see any difference despite thinking a cutoff should help.

I have put in a suggestion for a gamma weighting to bump up the faint end while keeping the top end under control.

If EnfuseGUI is giving you an error, your machine may be missing a Visual C++ library.

Download and install the following from Microsoft: http://www.microsoft.com/en-us/download/details.aspx?id=29

De-flickering: generally not a problem thanks to the inherent smoothing built into the enfuse process

Recovering from overexposure (partially):

if you have previous events that did not have a short exposure, or you were in a rush and did not set your camera correctly, and notice you are "missing" a short exposure, you can fake one! Mostly. How to fake a short exposure.

Discussion from Chris Spiel:

"To improve on the brightest highlights try e.g. 99% or 95% as upper cutoff values. You may want to set the saturation weight

to zero to get a clean response, while playing with the cutoff. BTW, option `--save-masks' gives you access to the blend masks and can be

used to check the effectiveness of the weights, cutoffs, etc. A 16-bit channel width of the input images probably will help, too."

"exposure cutoff cannot help much, because you did not feed Enfuse an image where the moon is tightly exposed, say e.g. Zone 4 to 6, to dig up Ansel Adams.

Enfuse never makes up pixels _outside_ of the luminance range of all relevant input pixels. See the fundamental weighting equation

in the Enfuse manual. So, just feeding Enfuse "overexposed" input pixels invariably leads to an "overexposed" output pixel.

However, you can try to tilt Enfuse's notion of optimum exposure and make it prefer darker values instead of exactly the middle of the

normalized luminance interval by passing option `--exposure-optimum' with an argument less than .5. Consult the manual for the meaning

of "normalized luminance", if it sounds all greek to you."