Hacker News Reader

Comment by dperfect | Hacker News Reader

http://www.pbase.com/flying_dutchman/image/157113241

I could be wrong, but I think it would be almost impossible to capture an 8-second exposure while flying and somehow manage to keep the stars from becoming light trails - at least not without some very serious camera stabilization equipment.

Since the photographer didn't seem to mention anything special used for taking the photos, I'm inclined to say they've been 'shopped.

View on HN

Have you ever looked at the moon while driving? It stays in the same position relative to the car, if the car is not turning.

Plane on autopilot + camera steadied by plane body + fisheye lens (which he was using) makes this very believable.

EDIT: not to mention, we are all hurtling around the Earth at around 1,000 miles per hour AS WE SPEAK. If that doesn't produce trails (it doesn't, see my math below), a few extra hundred MPH in a plane doesn't change jack squat.

Sorry, I still don't believe it. Even with a fisheye lens and an incredibly stable aircraft, you're still going to see trails. If you don't believe me, I invite you to try it.

EDIT: I assume all the downvoters can provide us with links to their libraries of long-exposure, night aerial photography without using gyros?

See my edit. The only trails you will see are due to the rotation of the earth; not the movement of the plane.

EDIT: MATH: Assume the horizontal FOV is ~90 degrees (conservative according to Wikipedia). 30 (time) seconds around the Earth is (90/360)/(24x60x60/30) = 1/720 of that field. Meaning, you'd need a 720 px image to even see one PIXEL of blur due to the rotation of the Earth. The additional velocity of the plane contributes MUCH LESS than this; hence it is not visible.

Running some rough numbers:

If a 10mm lens can capture ~120° angle of view, then the image at 1024 px (low quality) represents 0.117° per pixel. In the 30s exposure, there is < 3 px represented by many stars, so that's 0.351° of tolerance.

How often have you flown in an airplane with less than 1/3 of a degree of drift on any axis over 30 seconds?

Yes, I'm quite familiar with the rotation of the earth and its effects on long exposures (I do some astrophotography myself). I'm not referring to that here, but rather to the normal instability of an airplane.

EDIT: I'm not just referring to pitch, but also to roll, yaw and vibration.

Seeing as 1/3 degree drift in a plane traveling 700 MPH translates into nearly 300 feet of elevation gain/loss; and that the seat-back altimeter readouts on commercial airlines I've flown in never seem to deviate more than 10 feet from the set altitude, I'm gonna guess every auto-piloted flight I've been on drifts less than that.

Not to mention that these are not the first pictures of stars this guy has taken from a plane: http://www.pbase.com/flying_dutchman/image/155755548

Alright, you've convinced me. To be fair, most of this guy's other night aerial photos either have shorter exposure times or exhibit some obvious trails in the stars (even the one you posted shows considerably more streaking in the stars).

Taken out of context, the few photos included in this post did seem more suspicious, but it looks like a combination of very good/stable conditions and a lack of resolution to notice more streaking.

I thought it was worth letting you know that I upvoted this comment for conceding the debate when presented with evidence. This is a commendable skill, one which many of us lack, and which all of us occasionally forget to exercise.

Yes, I also upvoted that comment and the one he made before. This is how a debate should look like - both sides present their reasoning using (however rough) numbers and facts, and settle on the result evidence suggests. Thank you for reminding us how adults should talk, and - if you forgive going meta - thanks @nitrogen for reminding us to remind this ourselves :).

And I upvoted his still-grayed-out original post. This is what figuring things out looks like. It was a reasonable thing to bring up when examining something this bizarre.

The vibration is still going to kill you, though. You don't get pinpoint stars like that in a 30 sec exposure on a tripod on a vibrating surface. Unless it's doing active image correction, I find it implausible.

1. Have you ever been on a commercial airliner? There's not really much vibration. (Yes, there's vibration, but the amplitude is not more than like a millimeter.)

2. Mechanical image stabilization exists and is common. Not sure whether fisheye lenses have it, but:

3. Fisheye lenses GREATLY reduce parallax error. Do the math out.

4. The vibration would NEED to be rotational, not lateral, for all the same reasons discussed above (stars are too far for lateral motion to change their apparent position). However little lateral vibration there is in an airplane, I guarantee there's even less rotational vibration. Sound/vibration simply doesn't work that way.

You are not incorrect about anything. The only exception to the challenge you are making could be in that assumption that the images were intentionally doctored with to disguise/obfuscate/alter reality. In face: They were intentionally doctored, there has been an editor on these files. The originals will no doubt provide further math.

EDIT: Note that JPC seems to be preparing for a naming-of-feature challenge, which in itself is an interesting aspect of the whole story! Go for it, I say!

Skepticism is mandatory on HN, except when it's forbidden.

People have taken to down voting things that they don't agree with, it's sad because it stops debate and contrary opinions.

Down vote trolls, down vote personal attacks, down vote propaganda - sure.

> People have taken to down voting things that they don't agree with, it's sad because it stops debate and contrary opinions.

Downvoting to express disagreement was explicitly sanctioned by pg back in February 2008.

https://news.ycombinator.com/item?id=117171

It's actually very possible to capture that with some crude stabilization considering that his aircraft was moving towards the area he was photographing instead of laterally. (You would get a much-more pronounced motion blur of the foreground otherwise)

Just to be sure, I ran the watermarked "original" through FotoForensics:

http://fotoforensics.com/analysis.php?id=847c810d887372384b2...

There's some artifacts around the area but the overall pattern of the noise seems to check out. The EXIF data is also still intact (albeit processed through Photoshop), he'd have to be somewhat committed to forge that as well.

Not entirely impossible. A fixed mount for the camera (such as a tripod) would be able to get such an image, assuming steady flying conditions. In 8 seconds, stars would not leave much of a trail on a fish-eye lens (since the field of view is large).

The star trails are not going to be very visible with a 10mm lens and the photos are low res. Also have to consider where he's pointing too.

I still see trails on the outer edges of some pics though.

Or 30 seconds for that matter.

http://www.pbase.com/flying_dutchman/image/157113240

The hubble moves significantly faster than an aircraft and takes exposures significantly longer than 8seconds.

The Hubble also cost thousands of times more than many private aircraft.

It's less a matter of cost and more of its situation: orbiting, not flying, through vacuum, not air, with careful attention to stabilization and vibration elimination.

It costs a lot of money to get that large of an object into that ideal location, though. In the context of my original comment, "cost" could be seen as a proxy for "difficulty" of any kind. Perhaps it was too short and a bit flippant, but it seemed appropriate to respond in kind to the original comparison between Hubble and an airplane.

There are also aircraft-based observatories. Principally for exploring specific wavelengths of light absorbed in the lower atmosphere. And, incidentally, rather less expensive than orbital observatories.

Good guidance, getting above turbulence, and having specific compensation for movement/motion all helps.

My points stand: the characteristics of Hubble are not directly related to cost, and attributing the distinction to that alone is a poor explanation.

The Hubble isn't flying through the atmosphere, and is specifically engineered to take very long-duration exposures.

You've seen the Hubble Deep Field image, right? The one where Hubble's operators found an entirely empty region of sky and stared at it for over 134 hours over ten days and 342 exposures (mostly separated to keep individual exposures from being degraded by cosmic ray strikes).

That's really not comparable with an aircraft, moving through the atmosphere, with turbulence, engine vibration, and other factors contributing to deviations from a steady trajectory. Though the image does appear to be fairly plausible from others' comments.

http://en.wikipedia.org/wiki/Hubble_deep_field

The question was about why there weren't streaks from the long exposure photo. My point is that the hubble also moves very fast and takes long exposures of tiny points of light without streaks. Streaking has very little to do with the atmosphere and more to do with moving the target around the sensor during exposure.

The hubble is above the atmosphere to reduce other kinds of optical interference, but streaking isn't one of them.

Any vibration or relative movement of either the camera or aircraft will also cause movement trails, though not the ones typically associated with long exposures and star trails centered on the North Star.

Most such long exposures span at least several minutes, though.

>You've seen the Hubble Deep Field image, right?

This naming convention they are using doesn't seem scalable.

Hubble Deep Field

Hubble Ultra-Deep Field

Hubble Extreme Deep Field

Reply on news.ycombinator.com