[Regicide]

2 1/2 hours of a piece of garlic bread being sent to about 35 kilometers up, taken with a nonfisheye camera and with a box in the foreground for flat reference. Enjoy!


Take some screenshots and draw flat lines on. I tried to do it myself, but the upload size was too big.

[TomInAustin]

2 1/2 hours of a piece of garlic bread being sent to about 35 kilometers up, taken with a nonfisheye camera and with a box in the foreground for flat reference. Enjoy!

Take some screenshots and draw flat lines on. I tried to do it myself, but the upload size was too big.

Somewhere near the high point.  Pretty curvy for sure.

[Pete Svarrior]

nonfisheye camera

Barrel distortion is not exclusive to fish-eye lenses. It is a largely unavoidable consequence of optics.

and with a box in the foreground for flat reference

Indeed, it does! Try drawing straight lines across the styrofoam box to see just how "flat" it appears.



Not particularly helpful or useful, even if you ignore the FAQ's clarification on photographic/video evidence.

[TomInAustin]

Barrel distortion is not exclusive to fish-eye lenses. It is a largely unavoidable consequence of optics.

Very true.  The cloudy day they picked makes it impossible to compare the low altitude frames to the high altitude.

[DuncanDoenitz]

1.  At 35 km, the camera is about 0.5% further from the centre of the earth than it is at the surface.  How much better perspective is this supposed to give us? 

2.  Any lens with a focal length <infinity will distort the image to some extent.  And what is a "non-fisheye camera"?  Focal length?  Can I buy one?

3.  Styrofoam is a well proven straight-line reference medium.  Oh, I'm sorry, I'm thinking of stainless steel.  Apparently, styrofoam is NOT a well proven straight line reference medium. 

4.  Garlic?  Bread? 

[J-Man]

Thanks for this, while I don't see curvature, I do see what looks like the molten glass dome covering earth.  Science ends up proving things you weren't looking for !!!

[AATW]

Thanks for this, while I don't see curvature, I do see what looks like the molten glass dome covering earth.  Science ends up proving things you weren't looking for !!!

Can you show what you mean by this?
And can you explain how a glass dome of this size wouldn’t just fall apart under its own weight?

[Bikini Polaris]

The clip shows the sun for the entire ascension of the balloon, but isn't the spotlight effect supposed to create a light to dark moment? Is there a computation on the height where this transition should happen? As far as I understand at some point a weather balloon going up should stop seeing a bright sun.

[Pete Svarrior]

The clip shows the sun for the entire ascension of the balloon, but isn't the spotlight effect supposed to create a light to dark moment?

No. The only way you'd be obscured from sunlight is if the Earth itself obstructed it. In other words - you'd have to be launching at night or entering the night during the flight.

As far as I understand at some point a weather balloon going up should stop seeing a bright sun.

I'm not sure where you got that from, but it's not what we predict, and, as you just saw, it's not what actually happens.

[Regicide]

Had to get a new account, but here is my continued argument:

Once the camera is above the clouds, look at the horizon. It should appear flat.

Once the camera reaches its full height, look at the horizon. It should appear curved.

In both perspectives, look at the box in which the garlic bread is. It should appear identical in both.

The curvature observed at the full height is not a product of the camera. Otherwise, the horizon would appear equally curved at a low altitude.

[Pete Svarrior]

Had to get a new account

You could have just not deleted it in the first place.

Once the camera is above the clouds, look at the horizon. It should appear flat.

Once the camera reaches its full height, look at the horizon. It should appear curved.

Neither of those are necessarily correct, given that we've already established that your video is subject to very obvious barrel distortion.

[Regicide]

I’m going to address some complaints.

First of all, extreme barreleye? I’m willing to accept that a camera will always have some barreleye, but this is hardly a matter of extreme distortion. Before it enters the cloud layer, draw lines along straight objects like the truck bed, table, roads, etc. They all seem fairly straight. This is hardly a fisheye lens.

Secondly, I am aware that this was done on a cloudy day. The creator who launched this lives in England, where it is cloudy for 320 days of the year. It’s going to be difficult to launch this on a day with clear skies.

Thirdly, I don’t understand how you are refuting my request for you to make an observation, I guess I worded it weirdly. Once the camera is above the clouds, draw a line next to the horizon, and you can see it appears fairly flat. Once it reaches its apogee, draw a line next to the horizon and you can see it appears curved. How can an observation of the horizon that changes over time be attributed to camera distortion?

Finally, I did not make this video. I have neither the time nor the resources to put together a weather ballon observation launch, but if I did I would do several things differently. I would use a camera that is as close to zero distortion as possible, even though this one is fairly close. I would also use a ruler for reference. I know the video isn’t perfect, but unlike many other videos it doesn’t have a jump from launch to apogee. In this one, you can observe the curvature of the horizon slowly increasing, relative to the box. There’s a bit of distortion at the edges of the frame, but if you look at the middle, there is hardly any. Note I said hardly, there will always be some.

P.S. Just wondering, what’s the official FE explanation for the sky fading to black from blue. Is it still a product of atmospheric scattering, or is there some alternate explanation.

[Longtitube]

Pete claims "very obvious" barrel distortion, but I wonder if he has thought this one through. There is very mild distortion (tray seam line dips down a little in middle) consistent with barrel distortion at the bottom edge of the camera view, the sort that most photographers would tolerate. This distortion would be reversed at the top of the camera view (straight line would rise up a little in middle) and be of a similar amount.

However, nobody has demonstrated distortion across the middle of the camera view, where the horizon is shown in the screen capture displayed in the OP. Barrel distortion would not distort the view there, perhaps Pete is thinking of a bottle bottom lens?  http://cameramaker.se/Coke_Lens.htm

[Pete Svarrior]

First of all, extreme barreleye?

You're the only one to have called it extreme. I called it obvious. And if is obvious - all you need to do is inspect the box, as we already did.

Before it enters the cloud layer, draw lines along straight objects like the truck bed, table, roads, etc. They all seem fairly straight.

If we're operating with such nebulous and imprecise terms as "fairly straight", then we might as well say that the horizon "seems fairly straight" in the screenshot presented. We already have a perfectly good reference point to gauge the level of distortion, one that's always in frame. Why should we discard it and keep looking for new ones? Just to match your hypothesis?

However, nobody has demonstrated distortion across the middle of the camera view, where the horizon is shown in the screen capture displayed in the OP.

The horizon is not in the middle of the view in that picture, and its perceivable curvature is more subtle than that of the tray seam line. This is consistent with barrel distortion.

The best way forward here, if you're really unsure, would be to establish what lens was used and adjust the photo for distortion in something like Lightroom.

[Longtitube]

I beg your pardon, the screen capture was in TomInAustin’s post:-



Which part of the image was the “middle” again?

[Pete Svarrior]

Which part of the image was the “middle” again?

Bright green lines added for your convenience - though I'm not sure why that needed my input. Surely you could have measured it yourself?

[Longtitube]

Thank you so much.

You earlier said the visible seam on the garlic bread tray demonstrated barrel distortion was present. It is, although very mild, so here's what a uniform, straight-lined grid looks like viewed through a lens with strong barrel distortion, with green lines in the middle of the picture:–



These green lines are straight, showing there is no horizontal distortion along the vertical green line or vertical distortion along the horizontal green line, but the further away from the middle you look, the stronger the distortion becomes. At the edges the distortion is unmistakeable.

Now here is the screen capture first posted by TomInAustin, with the amount of distortion showing on the tray seam marked in blue and the amount of curve in the horizon marked in green.



Near the edge of the image, the distortion measures 3 vertical pixels and we would correctly expect 3 or 4 pixels in the opposite direction near the top of the picture, while also expecting less distortion nearer the middle of the capture.

That is not what the capture shows: the curved horizon is only a short way above the middle of the picture, yet is curved by at least 23 pixels. That's not what mere barrel distortion would show. The horizon is plainly curved.

[Pete Svarrior]

I'm sorry, but I'm not really interested in your whataboutism.

Your measurements are extremely fuzzy: the horizon is not at all well-defined within the picture, but you're pretending it is; you're also comparing the curvature of two arcs of different spans by simply measuring the difference in their sagittae - a mathematical non-starter. The red line you're measuring from doesn't even touch the horizon on either edge of the frame.

You also describe your expectations as "correct" despite knowing absolutely nothing about the setup in use - you call them "correct" because they match your preconception.

In fact, we can commit similar crimes on geometry with your own grid:



I already provided you with simple steps to correct the photo for barrel distortion. If you really deny its significance, you can verify your own claim.

It's strange that you're so desperate on this particular point, too. It's not like it changes how useful the photo is - it's just a drop in an ocean of issues.

[Longtitube]

I'm sorry, but I'm not really interested in your whataboutism
....
It's strange that you're so desperate on this particular point, too.

There's no need to get personal, mate, even if you were right about desperation (which you aren't.)

you're also comparing the curvature of two arcs of different spans by simply measuring the difference in their chords

You're quite right, so I measured the length of the straight blue line by the tray's seam, originally highlighted by yourself. It's 715 pixels.




The tray seam, shown as an arc in the screen capture, dips 3 pixels below the long straight blue line. Were it the same width as the image, 1117 pixels, it would dip (1117/715) x 3 = 4.6 pixels. This is still sustantially less than the estimated 23 pixels the horizon rises above the straight red line, in fact about a fifth of the rise.

I repeat the general point about barrel distortion being important near the edges of an image, whereas it's very slight near the middle of the same image. I've seen many images displaying barrel distortion over many years and many showing much more distortion than the screenshot. Distortion affecting the middle of an image more than the edges is not barrel distortion, and is not evident in the screenshot.

I deliberately do not intend to process the image for distortion in Lightroom or any other program, because we have both seen many claims over the years that this, that or another image can't be trusted as it's been "Photoshopped" and we both know the pointlessness of such arguments. Indeed, the wiki makes this very point in the FAQ.

Yours must be a thankless task, moderating forums populated mostly by people who doubt or outright reject the flat earth hypothesis. Another week and another sceptic with experience and real-world knowledge asks difficult questions casting doubt on another part, or from another angle on the same part as last month, which the believers here struggle with. I'm sure you must groan at the next hazy, wobbly video shot over miles of water to make or disprove the case for the horizon or ships disappearing over it. It's not for the money anyway!

However, should anyone throw the barrel distortion problem back into the mix, you have an example to produce in answer – I'll leave the original test chart where it is. The lens was an 18-135mm zoom, the chart image was lifted directly from a review of that lens. What does surprise is the FAQ suggests a curved horizon will be seen from high up: I'd have thought you'd welcome the video evidence of such.

[Pete Svarrior]

The tray seam, shown as an arc in the screen capture, dips 3 pixels below the long straight blue line. Were it the same width as the image, 1117 pixels, it would dip (1117/715) x 3 = 4.6 pixels. This is still sustantially less than the estimated 23 pixels the horizon rises above the straight red line, in fact about a fifth of the rise.

This continues not to be how geometry works. I'm sorry, but there's nothing more I can do to help you with this. The ratio between the "line" and "dip" have nothing to do with the eccentricity of these two arcs. I can keep pointing out that you're wrong and explaining why you're wrong, but in the end of the day you won't accept it, because you're not interested in being correct - you just want to confirm your preconceptions.

But hey, let's keep on keeping on. Let's illustrate the issue with my previous example:



The span of the white frame is 590 pixels, and the "dip" (sagitta) of the first arc is 14 pixels.

The second arc has a span of 195 pixels and a sagitta of 1.5 pixels.

You assert that I can make these comparable through a simple ratio. Let's do that.

1.5*590/195 = 4.53

As you can see, by your logic, I should expect that the arc towards the bottom of the image is much less curved than the one closer to the middle. However, even a cursory visual comparison will reveal that not to be the case. In reality, when measured correctly, the sagitta of the second arc is 23 pixels. Your logic fails.

The physical reason behind your error is that (as you astutely pointed out) the effect is more pronounced the further away from the centre of the frame you are. By sampling the curvature from just the (horizontal) middle of the frame, you fail to account for the significant increase in effect towards the edges.

You also entirely ignored the many issues with your "23 pixels" estimate - the line you're using as your reference point does not touch the horizon on either edge of the frame (or, indeed, at all). When adjusted appropriately, the correct sagitta is more akin to 10 pixels.

You already know how to correctly verify your claim. If you choose to "deliberately refrain" from proving your position, then I don't think we have much left to discuss.

I repeat the general point about barrel distortion being important near the edges of an image, whereas it's very slight near the middle of the same image.

Terms like "very slight" and "important" continue to be meaningless. The supposed curvature of the horizon in the image you're focusing on is "very slight" and "not important" and yet here you are fixating on it.

The lens was an 18-135mm zoom

Do you realise how extremely wide this range is, and how useless that statement is as a result? An 18mm focal would be bordering on a fisheye lens, which this obviously isn't. A 135mm focal wouldn't capture anything remotely close to this wide an area. Before you can perform your experiment, you need to know the actual focal of the lens at the time of filming, not what the particular device is capable of.

I deliberately do not intend to process the image for distortion in Lightroom or any other program, because we have both seen many claims over the years that this, that or another image can't be trusted as it's been "Photoshopped" and we both know the pointlessness of such arguments.

Yes, as I pointed out many times, there are more severe issues at play here than you trying to circumvent simple optics and geometry. Nonetheless, my focus for now is on pointing out these two failures, and altering the photograph to more accurately represent what you should see from an altitude of 27km would be beneficial to you.

Yours must be a thankless task, moderating forums populated mostly by people who doubt or outright reject the flat earth hypothesis.

Oi now, governor. There is absolutely no need to get personal, innit.

Another week and another sceptic with experience and real-world knowledge asks difficult questions

Yes, with insight and expertise like referring to a lens as an "18mm-135mm zoom" when trying to determine its barrel distortion, claiming that the Earth is round but 5 times smaller than in RET, or demonstrating your excellent knowledge of geometry as you did above, you're guaranteed to blow me away.

EDIT: I note that I've been referring to these curves as "arcs" which may be a bit hasty - they could be arcs, but they might not be depending on the specific situation. I'll leave the phrasing as-is since it doesn't particularly affect any of the underlying reasoning, but it's only fair that I highlight that inaccuracy.