Sunday, November 10, 2019

Shot Planning: Full moon rising behind Lick Observatory (Mt Hamilton) Monday November 11?

This is an email thread and some associated notes. The original idea expanded quite a bit.

Email from Chip, 10/28

Fred Larson is putting together some full moon shots up in SF next month, but we have some good opportunities closer to home. I tried a couple of times a few years ago to shoot the full moon rising behind the observatory, but never got the alignment right. Now our tools our better. This shot requires parking on Skyline Rd between Hwy 9 and Castle Rock, and hiking 0.5 mi., partly on the Skyline Trail and some of it off-trail to clear the trees. The parking is legit, though we're supposed to be gone at sunset. Anybody interested?

Here is the setup in TPE:

https://app.photoephemeris.com/?luuid=23e015d8-c545-4509-baf7-25a232c018a6&ll=37.250054,-122.105669&dt=20191111170400-0800&z=11&spn=0.38,0.71&center=37.2933,-121.8473&sll=37.341132,-121.643070

Here is the location and trail in Google Earth:

https://drive.google.com/open?id=1KMH9n2Duov5UA1wp-JPCeGQQGzWCwlJf&usp=sharing

I've hiked this trail before, but not the off-trail part. If anyone is interested, I'll check it out beforehand.

There's an even easier setup the night before, Sunday Nov 10. The shot would be from right off Skyline in the Long Ridge area. The moon is 97.6% lit, and the shot is at 16:35, a half hour before sunset. So the observatory would still be in direct sunlight. Here is the TPE setup:

https://app.photoephemeris.com/?luuid=23e015d8-c545-4509-baf7-25a232c018a6&ll=37.285411,-122.152554&dt=20191110163500-0800&z=14&spn=0.05,0.09&center=37.2871,-122.1497&sll=37.341132,-121.643070

Email from Bill, 10/29

From the Santa Cruz mountains you are pretty far away from Mt. Hamilton. How long a lens are you planning to use? With my 200mm and the crop factor on my Canon 50D, that is still going to be a small image of the observatory. Is that what you are after? Another option is going to Mt Hamilton Road and getting an upwards looking picture from there. (Attached is an example of the shot without the moon.)

Email from Chip 10/31

You have a great point about the tele. I went back and looked at some shots I took a few years ago from the hills above Los Gatos, this one is at 400mm full frame (200 with 2x converter). Yeah it's a lot smaller than I remembered it. I have a 500mm lens now and would use the doubler, but it would still be relatively small. Nice big moon though.




Email from Chip 10/31

Hey Bill,

Yes, there are several more setups along Mt Hamilton Rd and in Grant's Ranch park. Because of the elevation of the shots (we'd be looking up), the shot time is somewhat later, and the moon would probably be less orange because it is higher.

Moon ephemeris and geographic data below are mostly from TPE (photoephemeris.com). Their elevation data does not have as much resolution as I'm giving it credit for, which will lead to some errors in the ideal time and azimuth for the shot. The shot time has to account for the elevation angle between the shooting location and the target (the observatory). The moon moves relatively quickly toward the south as it rises so the shot time affects the azimuth quite a bit. A flat location with room to move north-south will have less error and be easier to correct.

For comparison, my first idea of the shot from the Castle Rock area has an angle of elevation of 0.42 degrees. The moon's radius is 0.26 degrees, so ideally we want the moon to have risen 0.42+0.26 or 0.68 degrees above the horizon to be centered behind the observatory. This occurs a little after 17:03 at an azimuth of 76.0. This distance, as you pointed out, is quite far, 42km. The observatory building is 85m long (the dome plus the flat part). The dome itself is about 24m. At that distance it has an arc of 0.116 degrees if my rusty trig is correct (asin or atan of 85/42000), or only about 20% of the diameter of the moon. I still think that would be cool but my 500mm lens has a horizontal FOV of 4.1 degrees or 2.06 with 2x. The observatory will occupy 5.6% of the horizontal. My 5DIII sensor provides 5760 horizontal pixels so the observatory would be only 324 pixels wide. So maybe not.

I worked on several alternatives. The first two are interesting, the last two don't work.

Grant's Ranch #1 (Hall's Valley Rd).

6.6km from the observatory. Angle of elevation to the observatory is 6.59 degrees. The ideal elevation for the moon is about 6.84 degrees above the horizon, which occurs at 17:35 at an azimuth of 80.8
At this distance the observatory building occupies 0.74 degrees and the dome itself 0.21 degrees. With the 500mm lens, the building occupies 18% of the width of the frame, or 1040 pixels. Usable but not great. With a 2x doubler it's pretty good. But do you only have a 200?
https://app.photoephemeris.com/?luuid=23e015d8-c545-4509-baf7-25a232c018a6&ll=37.331505,-121.716545&dt=20191111173600-0800&z=18&spn=0.00,0.01&center=37.3316,-121.7162&sll=37.341132,-121.643070
This should be a very easy spot to shoot from, just off the road near Hall's Valley campground, with some room to adjust north-south. I would need to check vegetation first.

Grant's Ranch #2 (Canada de Pala Trail).

3.6km from the observatory, angle of elevation 7.85 degrees. Ideal elevation for the moon is 8.1 degrees above the horizon which occurs at 17:42, azimuth 81.8.
The observatory building occupies 1.35 degrees at this distance, the dome 0.38. The 0.52 degree diameter of the moon would be smaller than the building but still larger than the dome. This could be a great shot, but there is more risk that if the azimuth is slightly wrong it won't line up.
With the 500mm lens, the building occupies 33% of the frame width, or about 1900px. I wouldn't use my doubler here. If you shoot Canon I could lend it to you.
https://app.photoephemeris.com/?luuid=23e015d8-c545-4509-baf7-25a232c018a6&ll=37.336619,-121.682245&dt=20191111174200-0800&z=16&spn=0.01,0.03&center=37.3355,-121.6814&sll=37.341132,-121.643070
This is a relatively easy location, accessible via a well maintained trail 1km from a parking area. It is officially closed at sunset (17:00), however, and the parking area does have gates. Parking on the road here may invite a ticket, I don't know. I like this spot so I may inquire whether I can get permission.

Mt Hamilton Rd #1.

835m from observatory, angle of elevation 11.95 degrees, ideal moon elevation 12.21 degrees, occurs at 18:03, azimuth 84.8.

This close, the dome itself occupies 1.64 degrees, the building 5.85, both much larger than the moon. I don't think this works.

https://app.photoephemeris.com/?ll=37.340428,-121.652249&dt=20191111180200-0800&z=18&spn=0.00,0.01&center=37.3404,-121.6532&sll=37.341132,-121.643070

There are turnouts right here, but then we'd be crossing a barbed wire fence to climb a little hill. The fence may be broken right here.

Mt Hamilton Rd #2.

425m from observatory, angle 18.35 degrees, ideal moon elevation 18.61, occurs 18:36, azimuth 89.5.

This is a non-starter because it is too close. The moon will be tiny.

https://app.photoephemeris.com/?ll=37.341093,-121.647628&dt=20191111183600-0800&z=19&spn=0.00,0.00&center=37.3410,-121.6483&sll=37.341132,-121.643070

There is a turnout right here, then we climb a little hill. Doesn't appear to be fenced.


Email from Bill, 11/2

Okay I did a little research today. Attached you will find a couple of doctored up photos showing the observatory today plus old moon images. I wanted to see how big the observatory would be in both places you listed using my equipment (my Canon 50D with 1.6 crop sensor).



The first one, with the fuller moon, is at the Grant Ranch entrance (I didn’t go in because I would have had to pay), and was done with my 300mm lens (a not-very-good 70-300 kit lens). It fills the image well enough. I added an old moon image using the 300mm lens to show the relative sizes.



The second one, at the other parking lot and 15 minutes up the trail, I used my 200mm lens and superimposed an old partial moon also at 200 mm. Again the observatory fills the image well enough, but you can see the moon image is smaller relative to the size of the observatory. There is plenty of room to move left and right to accommodate where the moon might actually rise.

So I guess it depends on what you want to get. The observatory is at the same angle and looks pretty much the same in both photos. So if you are after the orange color, then you might want to go back to your original plan from the Santa Cruz mountains where you will get a lower angle and therefore more color. With your 500mm + doubler, you will get a nice large moon and the observatory might be big enough for you.

I am thinking that for me, I would prefer to use my 200mm lens and therefore want to get closer to the observatory. Also given the sunset time issue (having to leave the park), I may want to go a day or two earlier and get a mostly full moon in more daylight.

Email from Bill 11/4

I have been trying to figure out what to do. I called the park and they said the park does close at sundown although there “may” be a half hour grace period before they lock the upper gate and issue citations. (They also said some nights the rangers don’t actually get there.) Looking at the timing for Sunday night, it seems risky. The moonrise is at 16:26 and sunset is 17:00 so that seems good, but we need to wait until the moon rises a bit, take some photos, then hoof it back to the parking lot. It may be a 20 minute hike back to the parking lot. It doesn’t seem possible to get back before 5:15. If there really is a half hour grace period, then all is good.

So I guess after all my pondering, I think the safest thing for me is to go on Saturday the 9th. Then the moon rises at 3:59. I can easily get my shots and be back at the parking lot before 5:00. Unfortunately I have to hike further, the moon is not as full (93.7%), and it won’t be orange, but I don’t have to worry about closing time.

I think your idea about camping is the way to go, but I don’t want to go overnight.

P.S. Attached is a picture I took on the other side of the valley at Mt. Umunhum with a crescent moon.


Email from Chip 11/4

I appreciate the dilemma for Sunday night. With a little more advance planning we might get a special Park Use Permit which might solve the problem, but I'm doubtful there is enough time now.

If you do want to shoot Monday from Hall's Valley (site #1), you could join me and then head out afterwards -- the main gate only closes for people entering, but the exit lane is still open (with 1-way tire spikes). You'd be welcome to stay of course, but I understand wanting to get back.

Another option is in Hall's Valley on Sunday night, just north of site #1 at 17:07. This would avoid the possible ticket and would get you home sooner. It's a long shot for your 200, but I could let you use my 2x and just shoot with the 500.

https://app.photoephemeris.com/?luuid=23e015d8-c545-4509-baf7-25a232c018a6&ll=37.337813,-121.717795&dt=20191110170700-0800&z=17&spn=0.01,0.01&center=37.3367,-121.7165&sll=37.341132,-121.643070

Email from Bill, 11/10

Bill emailed his results from shooting last night (Saturday) with the following "lessons learned:

  • I wasn’t in the right place. When the moon appears I had to run a ways to get to the spot I wanted. Even then I didn’t quite get the angle I wanted. Be ready to move!
  • The observatory is orange but the moon is white. I guess that’s the smoke being low in the valley.
  • Due to the angle of the shot, it wasn’t until 45 minutes after “moonrise” that I got the shot. So don’t bother to go on the trail tomorrow as you won’t get see the moon before sunset. Speaking of which…
  • I got to meet the ranger at my car since I didn’t get back until 5:15. I didn’t get a ticket, but she might not be as nice next time. (And yes, they do lock the parking lot.)
  • Based on my timeline, tomorrow should be good from your campsite, but Monday may be too late to get any light on the observatory. Perhaps a silhouette only?


Shot Planning Spreadsheet

I made a spreadsheet to help with my shot planning. I was mostly looking for a way to use more precise elevation data than TPE. It also automates some of the other calculations I was doing. 


The primary inputs can be captured using Google Earth, from the URL when centered and zoomed on the points.  The property editor can also be used: use 2D mode, center it on the point, and zoom in.

  • Subject Location and Elevation, 
  • Shooting Location and Elevation
There are optional inputs for subject size, lens focal length and crop factor, and camera sensor width.

Given the locations and elevations, the sheet computes
  • Bearing to subject (degrees)
  • Actual range to subject (kilometers)
  • Angle of elevation (degrees)
  • Apparent subject size (degrees)
  • Lens angle of view (degrees)
  • Relative subject size (percent of frame and pixels)

Final Shot Planning

Sunday Night 2019-11-10

Nominal Moonrise: 16:26 at 81.0 degrees.
I used to TPE to determine a shot location just off Grant Rd near the first bathroom. The TPE parameters are
  • Sunset 17:00
  • Shot time 17:06
  • Moon: 86.7, +7.0
  • shot location + 472 m | 37.3377°N 121.7186°W
  • subject (dome): 86.69, +6.79
Checking the locations using Google Earth I determined:
And my spreadsheet then computes:
-->
Results:
Bearing to subject86.698degrees
Horizontal Range6.690kilometers
Actual Range to subject6.738kilometers
Angle of Elevation6.894degrees
Apparent subject size0.204degrees
Lens Angle of View2.062degrees
Relative subject size:
Percent of frame9.9%
Sensor pixels569.944pixels

Comparing to TPE this is spot on in bearing, as I had hoped and expected. The angle of elevation is 0.1 degree higher. The ideal moon elevation is +7.15 , which makes my ideal shot time a little closer to 17:07. At 17:07 the moon azimuth is 86.8, indicating a small adjustment northward.

Here's my final TPE setup. The location is 37.33782131,-121.71885617,478.36830055.



Monday Night 2019-11-11

In the original TPE setup for this I have
  • Sunset 16:59
  • Shot time 17:35
  • Moon: 80.8, +6.8
  • Location + 509 m | 37.3315°N 121.7165°W
  • Subject (dome): 80.66, +6.59
From Google Earth I'm using:
  • Subject (dome): 37.34110674,-121.64305876,1285.25634598
  • Monday night shot location: 37.33156539,-121.71649342,510.55247654
My spreadsheet computes:

-->
Results:
Bearing to subject80.697degrees
Horizontal Range6.578kilometers
Actual Range to subject6.624kilometers
Angle of Elevation6.716degrees
Apparent subject size0.208degrees
Lens Angle of View2.062degrees
Relative subject size:
Percent of frame10.1%
Sensor pixels579.781pixels

Again, the bearing is fine but TPE gave me an angle of elevation 0.12 lower than my calcs using Earth data. Using the Earth elevation of 6.716 to the dome, my ideal shot elevation is 6.98, pusing the shot time back about a minute to just before 17:36.  Moon azimuth is now 81.0, which is small adjustment northward, and I think smack in the middle of campsite #14. It is a small adjustment. Here is the final TPE setup. The location is
  • 37.33174796,-121.71663604,510.51296615

Other Shots 

I looked into doing a sunrise shoot, but the sunrise Sunday at 6:40 is at azimuth 111, much further south than the moon.  So to get an alignment with the dome one must get much further north.  There is a shot about 4km up the Canada de Pala trail at 7:15 (TPE setup link).  There's only about 40m altitude gain difference from the parking lot, but it involves some up and down. I'm sure it would take me at least an hour to hike carrying my bag, maybe 90 minutes. And there are questions about parking before sunrise. Sunrise Monday is essentially the same.

This would be a good shot for using an electric bike!!

I browsed around on the web, besides weddings, the main shots are landscapes of rolling hills, shots around the lake, and wildlife including bobcats and boar, and the historic Grant home.

References

Grant Ranch

Calculations

These go into my Shot-Planning Spreadsheet.


-->-->

Tuesday, October 29, 2019

Noise Reduction using Median Stack Blending

On October 26 I did another Milky Way shoot at Pigeon Pt Lighthouse. Some of my best material was shot at ISO 6400, which is very noisy on my 5D Mark III.  After exploring alternatives such as the classic Topaz Denoise and their new Denoise AI, I finally resorted to median blending. This worked because I had been taking a more-or-less continuous sequence of photos.  The sequence is marred by various mishaps, so probably not a good basis for a timelapse as I had intended. But it does provide an opportunity to stack multiple frames and use median stack blending to reduce noise.

I knew that this was a thing, but I couldn't figure out how to do it in Photoshop.  I found two good references; there are more I'm sure.

At some point maybe I'll write an expanded post, but for now I just wanted to make sure I had a record of these references.

Milky Way over Pigeon Point Lighthouse, October 26, 2019
Stacked frames at ISO 6400 using median blend
My technique basically followed the work flow in the NatureTTL article, using 5 successive frames.  Stars were manually aligned using Photoshop Warp tool, then combined into Smart Object so that Layer > Smart Objects > Stack Mode > Median could be applied. Note that the layer masks really do need to be deleted, not just disabled.

Lighting on the lighthouse is very inconsistent because of various car headlights, and photographers waving their headlights around. A separate group of frames that had "good" light on the lighthouse was median-stacked as well and blended.

In order to recover the grid in the lighthouse windows I used a single frame from earlier in the evening that was not completely blown out by the light.

For comparison, here is my original effort using a single ISO 3200 frame:

Single frame at ISO 3200
Looking at it now, there are still some aspects of the single frame shot that I prefer.
  • The drama of the deep shadows
  • The multi-frame version now looks somewhat "brown" to me, vs. the pleasant blue of the single frame.
  • The lighthouse lighting on the single frame version is pleasantly shaded. The newer one looks flat.  In fact the lighting overall seems flat.
But the blended one is much more crisp, and there are an astonishing number of stars. So nothing I can't fix.

Sunday, September 22, 2019

Notes on Michael Frye's Method of Processing a Milky Way Photo

I had an opportunity to watch Michael Frye's video on processing a Milky Way photo in Lightroom.  It was so excellent I wanted to take notes on his procedure.  Much of this echoes his general Lightroom methodology, so I also bought his ebook Landscapes in Lightroom for more detail.

Select a raw image and open it in the Develop module.  I'm going to use my PigeonPoint-148 from August 31.
Aug 31, 2019 Pigeon Point (#148) as shot.
30 sec, f/4.0 ISO 3200 16mm, Canon 5D MkIII
Frye has a standard workflow which he uses for processing landscapes. It's explained in detail in his ebook. In the Milky Way tutorial he goes through each step and applies it to his Milky Way example.

In the ebook, Step 1 is to set the default settings (mainly exposure -1.00 and contrast -0.33

Step 1 - Set Camera Profile

Frye describes setting the profile in the Camera Calibration panel of an older version of Lightroom (v 6).  His comments:
You may be happy with Adobe Standard.  "What you're looking for is the right hues" in the sky. Don't worry about contrast and saturation, they can be adjusted later.  He does not recommend the Landscape or Vivid profiles ("too contrasty ... difficulty bringing out enough shadow detail") Milky Way.
In Lightroom 8 the profile is set at the very top of the Basic panel, and the options have changed. The current default for me is "Adobe Color", an option that did not appear to exist in LR 6, or at least is not in Michael's menu.  I left it at the default.

Step 2 - Crop and Correct Perspective

By perspective, Frye means for example, "buildings leaning in" and I've certainly got that here.  The  image in his tutorial does not have this problem and therefore does not go through it, but mine does. In the Transform panel I use Guided mode and set one line on the horizon (temporarily boosting the exposure for easier visibility).  A straight line on the left of the building is too extreme so I ease it just a bit. The perspective correction results in cropping of the bottom and right side in particular. This is the result:

After Step 2 - Perspective Correction and Cropping

Step 3 - Clone out Aircraft Trails (Dust Spots, Other Retouching)

Frye demonstrates using the Spot Removal tool to remove jet trails. I have a couple so I practice his methods. The video is really good on this.
  • Open Spot Removal tool. He most commonly uses Heal mode; try Clone mode when Heal does not give good results.
  • Zoom in on the jet trail.
  • Set the brush size.
  • Using the cursor, draw a line along the trail.
  • Press "h" to hide the tool and examine the results. If you want to change it, press h again, then select a different patch source area.
  • To draw a straight line instead of freehand, click at one end of the jet trail, then shift-click at the other end.
Frye video. Spot removal, Heal brush

Step 4 - White Balance

Frye temporarily nudges up the exposure to help see the colors in the sky.  In his case there is a green tinge from "air glow". Mine does not have that problem, but there is a bit of magenta.  My as-shot white balance has the Temp at 3950 and the Tint at -2.

As-Shot: 3950 K, -2 Tint
I take my tint down to -9, but below that it starts to get a little green.

Step 5 - Optional: Convert the Image to Black and White

This part of Frye's normal workflow does not really apply to Milky Way shots, he just skips it.

Step 6 -  Tonal Adjustments

In Frye's general work flow model, this step includes all tonal adjustments using the sliders or point curve. For the Milky Way tutorial he does not use the point curve.

Frye normally starts with exposure at -1 and contrast at -33 because Adobe adds brightness and contrast by default. I notice that this was true as of LR 6 and Adobe's 2012 Process 3, but I'm now running LR Classic 8.4 with Process 5, so I don't know if this still applies.  I'm going to assume that it does and, to stay consistent with his tutorial I start there too.

Here are Frye's tonal adjustments steps "for low-contrast images":
  1. Set the white and black clipping points.  He does this by holding Alt (or Option) to make the clipping visible, and moving the White and Black sliders in turn.
  2. White Clipping: Pressing Alt key while moving the White slider.
    White: The idea is to let the brightest stars clip (i.e. be maximum brightness).  In this image, the light on the lighthouse itself also clips.

    Black: I set it so that the only pure black areas are shadows.  The ocean has a little brightness, as does the green embankment.
    Step 6a: White clipping +70, Black clipping -10

  3. Tweak the Exposure and Contrast.
    In the oldest versions of Lightroom, changing the Exposure slider would move the white and black points, and I still thought it worked that way.  But it has changed!  In his ebook Frye explains that, starting with process 3 in LR 4, the Exposure slider is primarily just moving the midpoint "unless you push it really far". The ebook goes into more detail.
    Contrast behavior remains about the same. More contrast means brightening the highlights and darkening the shadows, i.e. making the curve more S-shaped.
    Because the lighting on the lighthouse is relatively strong in my image, I determine two completely different settings. For the stars I end at -.90 exposure, contrast +55. But the lighthouse is far too bright and contrasty at that setting. I prefer the lighthouse at exposure -3.50, contrast -100.  I decide to set the overall image for the stars (-0.90/+55) and deal with the lighthouse with a brush later.
  4. Step 6b: Exposure -0.90, Contrast +55
  5. Consider pushing up the Shadow slider a bit.
    When Fry does this he retrieves some shadow detail from some mountains. In my case it helps with the fence and embankment, but also considerably brightens the hazy portion of the sky.  Reducing it darkens the sky, but loses the color in the sky. In the end, I settle on +25, but may add some more to the fence and embankment with a brush.
  6. Step 6c: Shadows +25

Step 7 - Presence Panel 

As of LR Classic 8.4, the Presence panel includes
  • Texture
  • Clarity
  • Dehaze
  • Vibrance
  • Saturation
Frye's tutorial uses LR6, which is missing "Texture", but he discusses the others.
  1. Clarity. Clarity increases local contrast, which brightens the stars nicely. But it is very harsh on the lighthouse and cliff.  Frye adds +15, but plans to add more later using a brush on the sky.
  2. Dehaze. Also good for the stars, but harsh on the shadows of the terrestrial features.  Again, Frye decides to add this later using a brush on the sky.
  3. Vibrance. Good for bringing out some of the colors in the Milky Way. Frye puts this at +20.
I found that I liked Clarity +20 and Dehaze +25 for the sky, but it is awful on the lighthouse and shadowy earth features.  So I'll follow his lead and apply them later using a brush. I think this is better than adding them globally then trying to reverse them on the land.  So no change yet in those.

Vibrance somewhere between +22 and +32 looks fine for both earth and sky, so I set that here, choosing +26.

Step 8 - Adjust Individual Colors with the HSL Panel

In Frye's sample image he is trying to get rid of a green tinge in the sky. He does this by pushing up the Aqua Hue slider, and then easing off the Tint to reduce the amount of purple/magenta. He also tweaked the Temp a bit.

My image also shows some green in the sky, but not as much. Pushing the Aqua Hue is very effective in removing it.  I did also move my Tint back up to the original -2, but found that I like reducing the color Temp to 3850 from 3950.

Here is the image now, including the Vibrance adjustment from Step 7 and the color tweaks:

Steps 7 and 8: Vibrance +26, Aqua Hue +100, Hue -2, Temp 3850

Step 9 - Sharpening and Noise Reduction

Frye's standard workflow would do local adjustments (e.g. brushes) first. But for this image he wants to do some NR as part of those brushes, so it makes sense to do the global NR first.

Frye notes that we tend to zoom in too close in judging noise. What final print size are we going for?  At 300 dpi, the 22MP 3840x5760 images from my 5D MkIII would print 1:1 at 12.8" x 19.2".  But my 4k 2160 x 3840 32" monitor (screen is 28" wide) is about 137 dpi, and the same 1:1 image would be 28" x 42".  Now imagine that you've printed it at that size and are scrutinizing it from 20" away -- it's just unrealistic. Frye likes using 1:2 for looking at sharpening and NR.

I assumed that we would jump right into the using the Luminance slider to adjust noise. But Frye has a better idea. He believes the most effective NR in Lightroom is in the Masking slider under Sharpening.

The Masking slider is under Sharpening in the Detail panel.
First of all, notice that default sharpening of 40 is being applied to the whole image. The Masking slider controls "Edge Masking" and allows us to sharpen only edges, not the smooth textured areas. It's vital to understand that the mask is not adding sharpening, rather it removes sharpening from the "smooth" parts of the image.  In a noisy image like this one, sharpening is accentuating noise. So removing some sharpening is in effect noise reduction.

Hold down the Alt (or Option) key while moving the slider to see the mask.  It starts out at zero all-white. Then moving to the right, the detected smoother areas become black or gray.  The white areas will be sharpened, the black areas are not sharpened

Masking at about 42. Textured areas in the tower will get some sharpening.
The idea is to avoid sharpening smooth areas, like sky.  In Frye's example, he pushes the mask all the way to 100.

Masking = 100. Edges of the roof, chimney and foreground picket fence have disappeared and will not be sharpened.

But in my image, 100 causes some of the edges to disappear. They would therefore not be sharpened.  I need to bring it down so that the roof line, chimney and picket fence get some sharpening.

Masking = 85. Roof and picket fence get some sharpening
In the end I used 80 rather than 85, in order to pick up more sharpening in the chimney.

The Detail slider controls sharpening of fine detail which, in this image, is mostly noise. The Adobe default value is 25, but Frye's default is 50. He brings his down to 35; I just leave mine at 25.

Under Noise Reduction, the Luminance slider would allow us to get rid of virtually all noise, at the expense of turning the image to (Frye's words) "mush".

Noise Reduction: Luminance = 100.  "Mush"

Because of the high ISO (3200), there is a lot of luminance noise. It is especially noticeable in dark areas such as the sky and the shadows. At Luminance = 10 we see a meaningful improvement in the sky, but without losing all the texture.  Since so much of the "light" in the sky is noise, this noticeably darkens the sky, increasing the relative contrast with the stars.

For color noise, the default setting of 25 is fine.



Step 9: Masking = 80, Luminance = 10

Frye's sample image has water, and he notes that he will be going back to remove more noise there using a brush -- it's ok for the water to be more smooth.  I suppose the same argument could be made for the sky ...

Step 10 - Lens Corrections

My Canon 16-35 EF shows quite a lot of barrel distortion at 16mm, which is corrected by enabling the profile correction. Given the strong lines of the lighthouse, this is a big win for the image.

The profile corrects both the distortion and the vignetting of the lens. I like retrieving more brightness at the top and bottom, but it's not an improvement on the right and left.  In particular it reveals a lot of sky brightness on the left, evidently from the city of Santa Cruz.

Step 10: Lens profile corrections for distortion and vignetting

Chromatic aberration and fringing is not really an issue for this image -- Frye spends quite a bit of time on it in his tutorial and demonstrates various methods, including how to fix coma.

Step 11 - Adjustment Brush

We've saved up several adjustments for the brush:
  • Recall that I prefer the lighthouse at exposure -3.50, contrast -100, but set the overall image for the stars  at -0.90/+55.  I'll need to take down the lighthouse exposure and contrast quite a bit.  The light on the lighthouse needs to be taken down enough so the glazing between the panels shows.
  • The color of the lighthouse is too strong and too ugly. It may need further adjustment after the exposure.
  • Depending on which monitor I use, the ocean and the embankment may be completely lost.  In a print they will very likely be black.  So this needs some lightening.
  • I'd like to add Clarity +20 and Dehaze +25 for the sky.
  • Local noise reduction?  Maybe on the water.
Here it as after a first pass over all these items, using 11 separate adjustments brushes.

Step 11: Adjustment Brushes
And again after going back to tweak the global HSL (the sky was too aqua), as well the main lighthouse brush (it still had too much orange color on some monitors):

Tweaks to HSL and lighthouse
Well, it's hard to see much difference between the two.

Step 12 - Vignetting

This step is in Frye's Lightroom book; he does not mention it in the Milky Way video so I'm just including it for reference to his full workflow.  The Effects panel includes the Vignetting controls.  As I noted earlier, my lens in this case actually quite a lot of natural vignetting. This was useful on the sides of this particular image, but I was less pleased with the effect on the top and bottom.  Compare the images for Step 9 and Step 10.

I could have added vignetting by simply dialing down the vignetting correction in the Lens Profile correction stage and accepting the natural result.  Or I could have used the Effects panel.

In this image, instead of doing either of those I chose to add back a darkening brush over the city light bleed on the left.  Some darkening on the right of the lighthouse might also be nice, but I haven't done it yet.

Friday, August 16, 2019

Chip's Notes on Making Still Frame Animations and Timelapse Movies

About 10 years ago I made a bunch of animations (GIFs and timelapse video) from sequences of still frames.  Now I want to do that again, but have pretty much forgotten how.  So now as I rediscover the procedures, these are my notes.

There are multiple ways to do this. I used Vegas Video the last few times. I still have a current license for Vegas Pro, so I could do that.  Another way is through Photoshop.  Since I use Photoshop more often (but not for animation) in my regular workflow I'm going to start out try it that way.

Start in Lightroom

Either way, my normal still-frame workflow always starts in Lightroom.  Here I would first make whatever image adjustments I want.

I am shooting RAW 5760x3840 images on the 5D-III, which result in 17 MB DNG files.  For several hundred frames, these are too large to conveniently animate in Photoshop. So instead I export them as JPEG or PNG.  This reduces the file size, and I could reduce it even more by reducing the resolution.  Standard 4k video is 3840x2160.  To export as JPEG at full size and 100% quality produces a 10 MB file

If I do this, it is best to put them in a separate folder.

Optionally, I can then add them to the Lightroom catalog, and send them to Photoshop this way:
  • Within Lightroom, select the entire set of frames that you want to use.
  • Right click to open the context menu, then "Edit In > Open as Layers in Photoshop...".
But in order to minimize Lightroom clutter, I prefer to leave the JPEGs out of the catalog.  It's still easy to open them as layers in Photoshop using the Photoshop "File > Scripts > Load Files Into Stack" command.

Still Frame Animation in Photoshop

If you have a lot of frames, Photoshop will take a lot of time, but after a while  you should have a single file with as many layers as you have frames.  Now open the Timeline window (Windows > Timeline) and select "Create Frame Animation" from the dropdown menu in the center of the window.

Note that after you select it from the dropdown menu, you have to actually click "Create Frame Animation". This stumped me, solved thanks to Lindsay Kolowich's tutorial.

Select all layers (Select > All Layers).

Using the hamburger menu at the top right of the Timeline window, first select "Create new layer for each new frame".  This is just a checkbox, it won't do anything yet.

Again with the Timeline hamburger menu, click "Make Frames From Layers". This makes each selected layer a frame of the animation.

For me, the order of the frames was the opposite of what I intended.  My layers were ordered in sequence from top to bottom, but for whatever reason the frames were ordered from bottom to top. Fortunately there is a "Reverse Frames" item on the hamburger menu. Solved.

Photoshop allows you to set the display duration for each frame. A smooth animation needs at least 20 fps. I set the duration to 0.0333, for 30 fps. Select all the frames, then click on and set the duration.

In my Las Vegas animation, the first 20 frames are 4 minutes apart. The rest are 1 minute apart. In order to use these in the animation I tried to create 3 tweens between each of the first 20 frames to simulate the same 1 fpm frame rate.  I could not find a way to do this all at once so I individually selected each of the first 19 frames and tweened it with the next frame.  This creates opacity layers for blending, but I couldn't figure out how to make them work correctly, finally deleted them.

If I wanted a GIF animation I could export using File > Export > Save for Web (Legacy).  But I want a video. So I clicked the "Convert to Timeline" button, which was below the playback controls in the Timeline window.  This worked ok, but the result is confusing -- it looks like I have a separate video clip for each frame and they are all overlaid.  Trying to poke into this led to Photoshop bogging down and becoming unresponsive, apparently while trying to save the file. I finally just killed it.

I don't yet understand how Photoshop's tweening really works. If I did, I could do fades and so forth. But I am more inclined to try this in Vegas.

Still Frame Animation in Vegas

I have a copy of Sony's Vegas Pro 13.0 from 2014, now owned by MAGIX Software. After re-registering it, I used File > Import > Media to load all of my still images.

After a little poking around, I just dragged all the stills onto the timeline.  Now I have a video sequence, but by default each frame is displayed for 5 seconds. This can be changed using Options > Preferences. Select the Editing tab, then set New Still Image Length.  After failing to recall a way to change the duration in the timeline after the fact, I just started over, first setting New Still Image Length to 0.033 (1 frame at 30fps).

It was straightforward, using the Time Stretch/Compress tool and Ripple editing, to extend the first 20 frames so that they were each 4 frames long.  This adjusts the overall duration. Then, turning off time stretch and ripple, I dragged the first 20 events so that they overlapped for a cross-fade. This was all done one at a time.

I also tweaked the last frame for a fade-out. Then File > Render As and I was done.

Results

Here is the resulting YouTube video


References

Tuesday, July 30, 2019

Chip's Notes on Star Trails

2019-07-28, Sisters OR.   I have just finished nearly 5 days (3 days of shooting) with Aperture Academy's new Central Oregon workshop, led by ApCad veterans Brian Rueb and Mike Wardynski.  This was my 28th ApCad workshop, but comes after a personal workshop hiatus of nearly two years. Among many other subjects, we shot star trails from the Ray Atkeson Trail at Sparks Lake on the last night of the workshop.

July 27,2019. Sparks Lake Oregon.
15 frames, 240 sec f/5.6 ISO 800, Canon EF16-35 at 16mm, 5DIII 
I have shot star trails before in two other ApCad workshops:
May 8, 2015. Bear Gulch Reservoir at the Pinnacles. Light-painted cliff.
96 frames, 30 sec f/4.0 ISO 2500, Canon EF24-70 at 25mm, 5DIII
June 19, 2016. Tenaya Lake, moonlight
24 frames, 120 sec f/4.5 ISO 100, Sigma 15mm EX DG fisheye, 5DIII

June 18, 2016. Tuolumne Meadow, moonlight.
Lens correction flattens the horizon, but distorts the star trails into an ellipsoid.
19 frames, 102 sec f/4.5 ISO 100. Sigma 15mm EX DG fisheye, 5DIII
But it has been 3 years since that last time at Yosemite, and the passage of three years has aged more than the batteries in my intervalometer. I remembered very little of the star trail production process.  So this time, as I re-learned the procedure, I've created these notes. Besides helping me to remember next time, they may be of use to someone else.

Basics

With a film camera, star trails are produced by opening the shutter for a long time, say an hour, closing it when you are done.  But a digital camera sensor heats up over time and increases noise on longer exposures.  So star trails with a digital camera are commonly produced from a sequence of much shorter time exposures, each containing a portion of the trail.  These images are later "stacked" in Photoshop or a similar app to produce the full trail.

Longer exposures result in fewer frames for same overall duration, which simplifies processing. But also consider that each frame will capture more ambient light, if there is any.

One can add foreground interest to the shot by creating a base or "plate" image of the terrestrial elements from the same camera position, then layering it into the final scene.  The simplest way is to set up before dark and take a shot during twilight. Or after you have enough star-trail shots, using the same set up, elements can be light-painted.

Chip's Nutshell Guide to Shooting Star Trails

In a nutshell, do this:
  1. Which way is North? Where is the North Star?  Remember that the stars will seem to rotate around the North Star, so consider where that central point will fall in your composition (if at all).
  2. Get your camera and tripod all set for the composition you want.  Make sure it is secure because you don't want the setup moving for the next few hours. Make sure you have ready access to all the controls without moving the rig. Remember, it will be dark!
    [At Sparks Lake I set up my camera in such a way that I had to balance on some steep lava rocks to adjust things. It was fine in the twilight. But later, while checking the intervalometer during the very first star trail frame, I became a little unbalanced and rocked the tripod slightly. It was enough to throw off the alignment and the first frame had to be tossed.]
  3. While you still have lots of light, also set up your intervalometer.  See below.
  4. Before it gets too dark, shoot a few "base" or "plate" images while you can still make out the foreground and other terrestrial objects.  Later you may want to blend one with the star trails. Otherwise your foreground may be nothing but dark shapes against the night sky.  These should use the same f-stop as the star trails.
  5. When stars become visible, test your focus. Then set it to Manual Focus and consider taping the focus ring with gaffer tape.
  6. Depending on how the camera is set, it may not release the shutter if it cannot auto-focus.
  7. When it's really dark, consider taking some "star point" shots. See below for exposure suggestions. Avoid changing the f-stop. This will give you another option for the finished product, instead of just star trails.
  8. Long-exposure noise reduction: turn if OFF. The delay will interfere with the successive star trail shots.
  9. Now let's make star trails!
    -- Set your camera's shutter mode to B (Bulb).
    -- Set your ISO and aperture on the camera. At Sparks Lake we used ISO 800 at f/5.6 (I don't recall why I didn't open all the way, f/4 on this lens). The duration is controlled by the intervalometer.  We used 4 minutes. Shorter durations will pickup less ambient light in each frame and may be preferred.
    -- Trigger your intervalometer. Let it run for one exposure, stop it and check the results.
    -- Test exposure good?  Restart the intervalometer. Now relax, and avoid shining a light anywhere toward the camera(s).
  10. Some time later ...  stop the intervalometer: you're done with the star trail part! But before you move your camera consider whether a light-painted foreground is called for. If so, then simply open the shutter (you should still be on Bulb), paint the light, and close it.  Check the results (without moving the setup!) and repeat as needed.  Try to avoid adjusting the f-stop.
  11. Finally, before you completely pack up: if it's a nice clear night, point your camera completely into the sky, with no terrestrial elements at all, and grab a few more star point shots.  These could be layered into this or other shots as needed. Good to have.

Intervalometer Settings

You don't actually need an intervalometer -- a stop watch and a cable release with a hold button will do.  But it automates the task of shooting all the star trail frames -- besides convenience it is more reliable than a bored, cold human.  My Satechi WTR-A unit is functionally similar to the much more expensive Canon TC-80N3, with addition of wireless. But it is not currently found anywhere on the Satechi web site, so maybe they've quit making it.

Here are the basic settings on the Satechi:
  • Delay - Initial delay before the sequence starts. For our purposes this can be zero, or perhaps a few seconds to minimize vibration as you step away.
  • Long - Length of each exposure, assuming your camera is set to Bulb.  For this event I used 4 minute exposures as recommended by Mike and Brian, so set it to 04':00". [For some of my earlier star trails I used shorter exposures which of course provide a darker sky. The star trail itself is just as bright, so shorter may be preferred. But it means more frames to wrangle afterwards. Below 1 minute the small gap between each frame may become annoying.]
  • Intvl - Interval between exposures. For Star Trails, this should be as short as possible. The Satechi minimum is 1 sec.
  • N - Number of Frames.  For Star Trails, this would be the total time divided by the interval.  So if you want to shoot 4-minute frames for an hour, then N is 15.  But unless you want it stop on its own, I prefer to either make it a much bigger number, or set it to [--] and just stop it manually when I'm ready to stop shooting.
  • Audible beep - You'll quickly get tired of this and your shooting buddies will tease you. Turn it off.
The Satechi supports wireless control, that just adds several more things to go wrong. For star trails it is better to plug the control unit directly into the camera.

Shooting Star Points

Don't want a star trail?  Just want nice crisp stars?  A few things need to change.  This applies to time lapse stars as well.

Focus

If you haven't already set the focus for stars at infinity, do it now. Check the results and then set focus to manual to avoid changing it inadvertently.  This is also where gaffer's tape would be handy to secure the focus ring.

Long Exposure Noise Reduction

Consider whether you want to turn it back on.  This will cause a delay after each shot, equal to the exposure time.  If shooting timelapse, there may or may not be enough time between frames. I normally keep it turned off.

Exposure

If we don't want the stars to be motion blurred from the earth's rotation, then we can't use too long an exposure.  The venerable "500 rule" supposedly tells us the maxium shutter speed for a given lens. The "500" applies to full-frame cameras, the equivalent is "330" for APS-C crop sensors.
500 / focal length = longest shutter speed (seconds)
So with a 24 mm lens: 500/24 = 21 sec
My widest lens is 16mm.  500/16 = 31 sec 
These are maximum times, but they are too long for pixel peepers like me. For example, here's a 1:1 detail from a shot using 16mm for 30 seconds:


The motion blur is pronounced. It would be usable when displayed at smaller sizes but not at full size.  But in the detail below from the same 16mm setup at 10 seconds the motion blur is nearly invisible at 1:1:


So maybe use "200" or "160" instead of "500" in the rule.  200/16 = 12.5 seconds

Use the widest aperture available on your lens, and you'll need to crank up the ISO compared to star trails. I've used 6400 at f/4, lower is better if you can of course.

Pro Tips and Bonus Points!

  • Canon DSLRs show a bright red light on the back when the shutter is open. It's annoying; consider putting tape on it. You can put a pinhole in it if you want.
  • My intervalometer also has a red light. Same thing -- tape.
  • Bring something to sit on or lie on.
  • Infinity focus: set it while there is still light.  Consider putting gaffer tape on the focus ring so it doesn't budge.
  • Turn down your LCD monitor display. Brenda Petralla suggests also getting red transparent sticky film, cut to size of LCD display, stick it over LCD to preserve even more night vision.
  • Don't carry a big roll of tape? Another Petrella tip: wrap strips of gaffer tape around an old credit card to keep in your bag.

Post-processing

The general sequence of post-processing actions to produce star trails is:
  1. Process the entire sequence of star trail images in, say, Lightroom to adjust color, light dynamics, sharpness, and so forth (see "Lightroom Adjustments" below).  The star trail images should all be processed the same way (use Sync).
  2. Then select all of the star trail images and your desired base layer image in Lightroom.  Right-click and select "Open as layers in Photoshop".
  3. In Photoshop it is convenient to put any base layers at the bottom of the layers list, then group all the star trail layers together. Select all of the star trail layers in the Layers panel and set the "Blend Mode" to "Lighten" ("Screen" can be tried as well).
The night sky should fill with star trails as if by magic, and Bob's your uncle!  A few things to watch out for:
  • The trails should be smooth continuous curves (except for very tiny gaps every 4 minutes).  If there is any discontinuity, then something is wrong: did you include an extra frame or leave one out?  Did the camera move at all during the shooting? One way or another you'll need to fix this.
  • Are there lights from automobiles or other human activities? You may want to mask these out.
Finally, to complete the project:
  1. Merge your star trails layers into a single layer.
  2. Optional: Use standard blending techniques to add the base layer image.
    I have one example where the Blend Mode of the base layer was set to "Overlay", but I don't recall why. 

Lightroom Adjustments

I kept my star trail images "as shot", including a white balance of 4850, but cranked the DeHaze up to 100.  I added Lens Profile adjustments, Sharpening and Noise Reduction. Remember to Sync to all the other star trail images.

Depending on personal preference, you can get rid of unwanted car headlights and so forth here in Lightroom, or later in Photoshop.  I find it easier to do in Photoshop.

Example

Basic Star Trails

Here are the basic trails from the Sparks Lake shoot. There are 15 frames of 4 minutes each, representing 1 hour of time. There would have been 16 frames, but I wiggled the tripod during the very first one.  They have been adjusted in Lightroom as described above, and automobile headlights from a road in the center have been masked out in Photoshop.

Ambient light is a factor. The first frame was shot about 10:30pm, but the moon, 22% illuminated, did not set until 11:17pm, so it would contribute some ambient light in the sky. There is also the glow of Portland, 100 miles to the northwest.

Step 1: July 27, 2019. Sparks Lake, Oregon.

"Shooting Star" Effect

Next I create a simple opacity gradient so that the trails appear to "shoot" through the sky from east to west (counterclockwise looking north).  I want to give the 15 frames opacities from 100% down to about 10%.  A linear gradient is fine. So each step will be about (100 - 10) / 14 = 6.4. For simplicity I just made it 6 and stepped from 100% down to 16%. 

In order for the trails to shoot in the "correct" direction, the frame at the bottom of the sequence (i.e. the very last shot) remains at opacity 100%. Starting with the next to last frame, subtract 6 from the previous opacity value: 100% - 6% == 94%, and so forth.

Step 2: "Shooting Star" effect produced by opacity gradient. The most annoying aircraft trail is also removed.
Not everyone is a fan of this effect. I like it.

Adding the "Base" or "Plate" Layer (No Masking)

The mountains and lake are completely lost in the images shown above. The easiest way to add the "base" or "plate" layer simply puts it below the star trails group, and sets the star trails group blending mode to "Pass Through".



The image below uses a plate taken about 30 minutes before the star trail sequence starts. Exposure 30 sec f/4, ISO 1600.

Step 3a: Adds the "base" or "plate" image. Star trail group blend mode is Pass Through.
You'll notice of course that we've acquired a blue sky, against which the star trails themselves appear less vivid. Also we've acquired individual star points. But the blue sky goes with the lake. It works, I think. By the way, in this case, "Pass Through" and "Lighten" look the same (to me).

An equally simple variant uses "Screen" blending mode on the star trail group. This lightens somewhat more, and certainly shows a lot more of the star trails.

Step 3b: Star trail blend mode is "Screen".
I especially like that this mode expresses more star trails in the lake reflection.

The prime alternative to this approach (that I'm aware of) is to do some masking so that the mountains appear against the black sky.  This would also get rid of the star points. But then what to do about the lake?  This (below) just doesn't work:

Step 3c using masking to "clean up" the sky.
My next version instead uses two adjustment layers to tone down the plate, with a layer mask so that it only affects the plate's sky.



The result is this:

Step 3d uses adjustment layers to tone down the plate's sky, but leaves the lake and mountains alone.

This looked pretty good to me at first. But the next day, I realized there is a lot of fringing around the skyline.  In fact, it's worse than that:

Detail from Step 3d, showing misaligned plate.
Remember that above in Step 2 of the "Shooting" section, I mentioned that I had rocked the tripod during the first star trail frame and had to toss it due to the shift in alignment?  Well of course that also meant that every image before that point, i.e. my plate shots, had the same shift relative to the later images. That's what we're seeing here.  More work required ...

"Final" Version

Are we ever really finished?  Hence the quotes around "final".

My "final" version masks out the mountain and land from the star trails and makes separate adjustments to the land and sky+water.  I applied a Median filter (8px) to the sky+water plate image, which effectively removes the point stars.  That's the main reason why there is a separate land image, it was needed to preserve the details.

Layers used in final image
The mask on the Star Trails group is the inverse of the mask on the plate Land image. The plate Sky and Water subgroup does not need any masking.

This "final" version avoids most of the fringing around the mountain and provides a better blend overall, as well as removing the extraneous point stars.  However it does not address the brighter portion of the lake just below the mountains, which really should have been toned down like the foreground lake.

Step 4: Final image
Also, after looking at it for a while, I'd prefer a darker sky, and hence a darker lake.

References

These are just a few references.