Monthly Archives: August 2015

Moon Shot, take Ⅱ

The quarter moon offers much more contrast than a full moon, and this time I used a telephoto lens: Canon EF-S 55–250mm f/4–5.6 lens on its longest position.  I used the Canon 70D on a tripod, ISO 100, 1/160 at f/5.6 ×3 exposures.

Quarter Moon

It’s difficult to get focused, and about 1/3 of the bursts were out of focus even though I used the touch screen “Live View” to indicate what to focus on.

The lens has optical stabilization, but I suspect it’s turned off automatically if the steadiness of a tripod is detected, as that is a feature listed on some lenses. Unfortunately faster exposure times were subject to blurring anyway. I didn’t bother to use a remote shutter release, but that might help. I don’t know if this is (usual) vibration of the camera lens, or caused by atmospheric turbulence making the image move around. f/5.6 is the fastest aperture at the 250mm end of the zoom, which is also the size where diffraction starts to be noticeable.

Exposure-wise, 1/40 second still (only barely) has no clipping, so would be the optimum for ETTR to capture the most detail and least noise. 1/125 is the good-looking natural exposure, and at 1/640 it shows visible noise.

I improved the contrast by using the “highlight” slider as well as the overall “exposure” slider in Adobe Camera Raw. The difference between the limb and the terminator was still vastly different lighting, and bringing out better the rays of Kepler would darken the terminator and bring the shadow farther in to make more of a crescent than a quarter. So I used “curves”, leaving the darks alone and changing the contrast of the mid-to-bright.

The color is as-shot, with no white balance adjustment or B&W conversions.  The underexposed pictures came out brown, and the bright ones are very neutral.

I merged three exposures as follows:

  • Cancel all adjustments made in Lightroom while evaluating the exposures.  Noise reduction is turned all the way off, but sharpening is left as proper for the basic “developing”.
  • Transfer the burst to Photoshop as layers
  • Double the resolution
  • Auto-align the layers, specifying “reposition only”.  Leaving it on auto can make PS complain that it can’t find anything.
  • Use the stacking mode Median or Mean, whichever works better.  For longer bursts the Median gets rid of outliers, but for low-noise images Mean is more accurate.
  • Rasterize the stack.

The doubling of the resolution allows for alignment to half a pixel in the original images, and combining exposures this way actually increases resolution as well as removes noise.  With non-noisy images, the increase in clarity is due to the “dithering” of different exposures.

After combining the original exposures, then adjust the exposure, apply noise reduction, and so on.  This is easily done using Adobe Camera Raw as a (smart) filter in Photoshop.

Now sharpening is also done in the Adobe Camera Raw filter, at the same time as the other adjustments.  Because the resolution was doubled and the image is not pixel-sharp to begin with, a large radius value is warranted.  I set it to 3, which allows more aggressive processing without introducing fake detail from noise.

Shoot the Moon

A few days ago was the extra full moon in a calendar month.  This is sometimes called a blue moon, as a weaker form of the traditional term. July 2 and July 31 2015 were both full moons. Two full moons in the same calendar month occurred twice in 2012 and will occur twice again in 2018.

I took some pictures using my somewhat-new Sony α-6000 with the Sony 55–210 f/4.5–6.3 zoom lens.  This features image stabilization, and I hoped the auto-focus could lock on it.  I avoided the most extreme aperture setting so set f/5.0 and 1/250 second at the native (best quality) ISO 100.  The shots were hand-held.  Unfortunately,  I seem to have left the lens set to the short end of the range, not the long end, so this was essentially taken with a “normal” (not “telephoto”) focal length.

I also grabbed my Canon 70D, which was mounted on a tripod.  I attached the EF-S 55–250mm f/4–5.6 lens and tried that too.  Again, I backed off from the most extreme aperture setting and used f/4.5, which is brighter than the Sony.  Also again, I used the shortest (not the longest) focal length, so I would have been better off using a high-quality “normal” prime lens.

In both cases, in full manual mode, I set a variety of shutter speeds and shot bursts with each.  The in-camera metering and histogram is not useful because the black background overwhelms the data and the moon is actually very small in the frame.

Blue Moon #1

Blue Moon #2

I did notice that the underexposure warning overlay was standing off from the disk of the moon, and in Lightroom it was even clearer. The sky around the moon was not black but hazy. Presumably the atmosphere scatters light, enough to show in the exposure, even if it appears very clear by eye.

The histogram showed two distinct distributions, with the haze a few stops below the bright portion. I wondered if the dark shades of the haze were only in the dark sky, or if the dark areas on the moon also included these values. That would determine whether I truncated these darks and to to what extent. So I used a gradient map filter in Photoshop, setting different colors for the darkest darks, the next darks, and others. By moving the slider of the gradient editing tool I could set the split so that the background was “darkest” but any more would start showing the other color; then see if the dark shades on the moon were tagged as “darkest” or the other key color.

000453I found that a tiny amount in the Sea of Crises was in the same range as the background.  The Levels tool controls in the screenshot inset shows what I decided on. The “haze” is the lobe to the left, and it’s significantly below the main subject’s detail.  Only shadow areas that essentially go to black — don’t show any detail anyway — extend farther left than where the right lobe rises visibly from the zero value.

Even with the contrast and exposure levels set nicely, it’s hard to make out detail in the colorless image.  The earlier use of gradient mapping to find details inspired me to use the same tool again, for artistic purposes.  I used midnight blue for the darks and a warm-light color for the lights, but set yellow for the brightest values and adjusted the slider so the split between them was showing color contrast on the rays coming from Tycho Crater.

 Exposures

In both photos presented above, I stacked a burst of 3 shots and took the median of each pixel, to reduce noise.  Since this was ISO 100 with brightness in normal midtone range, I don’t think it did anything useful for the main subject, though it did improve the pixelization of the “haze” background (which is truncated from the final image anyway).  Before aligning the exposures I doubled the image size, as this will help with sub-pixel alignment differences.  The disk of the moon is only 250 to 275 pixels even after doubling: as I said, very small in the frame or about 3% of the height of the exposure.

Since the range of values (see histogram inset above) is a small portion of the exposure’s latitude, there is a large latitude in useful exposures possible.  I chose 1/250 f/5 for the Sony and 1/400 f/4.5 for the Canon, and applied the identical adjustments to each, so you can see they are the same EV.  A brighter exposure on the Canon, which is not overflowing the high end of the available capture range, was not nearly as good even though you would expect it to be less noisy and more nuanced.  Oddly, the size of the moon is different from exposure to exposure, so the zoom lens was not held fixed but moved.  Markedly more chromatic aberration and lower contrast (even after adjusting for the difference in exposure) could be due to the performance of the lens at different zoom settings?

Of the shots I took, the Canon (second picture above) is the clearest and sharpest.  If the lenses are equally good, the lower resolution and longer focal length of the Canon would be better, even though the two cancel out to give the same number of pixels.   A larger pixel pitch and larger projected image to match, covering the same number of pixels, should give better quality.  However I think any differences are dominated by lens performance.  It’s hard to tell exactly if it’s really “better” though because the best exposure in the Canon was actually a little smaller than the Sony, so it would look sharper due to that.

Conclusions

  1. shoot lots of exposures, even more than you would think you need.
  2. vary the settings, even beyond what you think is the right range and also change things you don’t think would matter.
  3. know your lenses.  Not only how the sharpness varies with f-stop, but are there any sweet spots or sour spots in the zoom?
  4. know how to operate everything by feel in the dark.  That includes reviewing images and zooming to inspect what you just shot with the camera’s screen.

This is also a striking example of how a good camera trounces what a cellphone can do.  People might be used to “selfies” and social media shots that are of very poor technical quality, and not really notice when a picture is better in that respect.  But when it comes to pictures that could not be taken at all with a smartphone camera, it’s a difference between getting the shot and having nothing useful at all.

It’s amazing that a small fist-sized compact camera can get hand-held pictures with the kind of close-up you have using binoculars.