Propeller blades drop in flight!

mikea said:
The camera on an iPhone does a circular(?) scan of the scene so you get effects like this on videos.
Click to expand...

Unspoilering it as it is horrendously wrong.

CMOS imagers generally don't do full frame integration (snapshot) modes. They do what's called "electronic rolling shutter" where they progressively expose one scan line at a time. If an object is moving fast enough, you get motion effects.

The Effect illustrated:

http://www.youtube.com/watch?v=T055cp-JFUA

--Carlos V.
 
Yeah, you get all kinds of strange artifacts using a CMOS/CCD sensor when taking a picutre of a fast moving object. As each line on the chip is scanned, the image on the next line has moved. Combine this with a prop moving that fast, and you set up a strobe effect, too.
 
The only digital recordings I've seen of propellers that look correct have been the TV cameras. iPhones have the effect noted there (along with all kinds of other interestingly-shaped props), one of my cameras makes it look like the props just jump rapidly into different positions where they sit before jumping to the next one. My old camera made it look like the props were slowly spinning backwards.

Nevertheless, the blades falling off are fun. That looks like a Dash-8, no? :)
 
Ted DuPuis said:
The only digital recordings I've seen of propellers that look correct have been the TV cameras. iPhones have the effect noted there (along with all kinds of other interestingly-shaped props)
Click to expand...

Cell phones pretty much never have full field integration capable sensors.

However, dedicated camcorders are more likely to have them.

I find it interesting how much pitch the blades have at cruise speed.

http://www.youtube.com/watch?v=e8o7M4-Kc0E

--Carlos V.
 
Unbeliever said:
I find it interesting how much pitch the blades have at cruise speed.

http://www.youtube.com/watch?v=e8o7M4-Kc0E
Click to expand...

The prop has to increase its pitch to keep the blade's angle of attack positive, usually between 2 to 4 degrees. With the inflowing air coming at such a speed the blades get to some pretty awesome angles. A friend took some high-speed photos of an old Lockheed Electra in cruise flight (it was converted to water bomber and he was flying the bird-dog airplane) and the prop tips were at more than 45°. The blade roots were even steeper, probably 60 or 70°.

Dan
 
is there video software that you can change the FPS after the fact and remove crazy prop dances? My newer camera's have the prop all over the place, my older Sony mini doesn't show the prop at all (or any different that we see it realtime).
 
OtisAir said:
is there video software that you can change the FPS after the fact and remove crazy prop dances?
Click to expand...


I doubt it, but I don't know for sure. If there is something, it might be expensive. I use a lot of industrial robotics-type cameras and vision processing s/w. Once the image is recorded, there's not much you can do about those kinds of things.

Possibly a slower shutter speed or frame rate would help, then it might just become a blur. If you had a strobe that was synchronized to the prop, though...
 
Out of curiosity, I just checked -the best I could find is that an iPhone records video at roughly 30fps. Apparently, that's in daylight and it will throttle down to about 24fps in low light conditions. At 30fps, that's 1800 frames per minute. At a setting of 1800 rpm on the prop you might be able to fix the prop blades in space, but they'll probably still show a curve to them because they'll be moving during the frame scan.
 
Matthew said:
Out of curiosity, I just checked -the best I could find is that an iPhone records video at roughly 30fps. Apparently, that's in daylight and it will throttle down to about 24fps in low light conditions. At 30fps, that's 1800 frames per minute. At a setting of 1800 rpm on the prop you might be able to fix the prop blades in space, but they'll probably still show a curve to them because they'll be moving during the frame scan.
Click to expand...

Yep. To get a prop disk, the image has to be exposed for 1/n revolutions of the prop where n is the number of blades. At 2400 RPM the prop is making 40 revs per second so it would be theoretically possible to get a disk with anything less than 80 FPS but unfortunately the exposure time is typically a tiny fraction of the frame time because that's how camcorders handle excessive light levels. One can often add an attenuating neutral density filter though and get the exposure time up to a large fraction of the frame time.

To stop the prop and have it look like a prop you need a camera with a frame transfer imager. Most read the image line by line, some camcorder imagers can only read half the image (odd or even rows) at one time and that produces a "sawtooth" prop with jagged edges. Some of the HD stuff and a few of the older analog camcorders have a "progressive" mode that will yield an image of a prop that looks correct.

BTW, the "bent prop" effect isn't unique to digital cameras, many high end SLR film cameras use a shutter that only exposes a small part of the film at the same time when the shutter speed is high with a small shutter "window" that slides across the film.
 
Last edited:
Matthew said:
Out of curiosity, I just checked -the best I could find is that an iPhone records video at roughly 30fps. Apparently, that's in daylight and it will throttle down to about 24fps in low light conditions. At 30fps, that's 1800 frames per minute. At a setting of 1800 rpm on the prop you might be able to fix the prop blades in space, but they'll probably still show a curve to them because they'll be moving during the frame scan.
Click to expand...

Oops. That one was linked above.

 
mikea said:
Oops. That one was linked above.

Click to expand...


The RPM match and the shutter speed match stop the props. The top image the prop looks normal, because the rotation of the prop is perpendicilar to the framescan.

In the lower image, there's a more oblique angle, so the blades are moving in a horizontal direction and being 'bent', but they are also 'stopped'.

Pretty cool -- of course, the conspiracy theorists will tell you that the engine has actually stopped and the blades were physically deformed.

edit: I just looked more closely at the 2nd video. It looks like there are markings on the blades that flicker. Sounds like the props are actually turning at some multiple of the frame scan rate, and different blades are being captured at the same location.
 
Last edited:
Ted DuPuis said:
Nevertheless, the blades falling off are fun. That looks like a Dash-8, no? :)
Click to expand...

Yup, no doubt.

Matthew said:
Out of curiosity, I just checked -the best I could find is that an iPhone records video at roughly 30fps. Apparently, that's in daylight and it will throttle down to about 24fps in low light conditions. At 30fps, that's 1800 frames per minute. At a setting of 1800 rpm on the prop you might be able to fix the prop blades in space, but they'll probably still show a curve to them because they'll be moving during the frame scan.
Click to expand...

gismo said:
Yep. To get a prop disk, the image has to be exposed for 1/n revolutions of the prop where n is the number of blades. At 2400 RPM the prop is making 40 revs per second so it would be theoretically possible to get a disk with anything less than 80 FPS but unfortunately the exposure time is typically a tiny fraction of the frame time because that's how camcorders handle excessive light levels. One can often add an attenuating neutral density filter though and get the exposure time up to a large fraction of the frame time.

To stop the prop and have it look like a prop you need a camera with a frame transfer imager. Most read the image line by line, some camcorder imagers can only read half the image (odd or even rows) at one time and that produces a "sawtooth" prop with jagged edges. Some of the HD stuff and a few of the older analog camcorders have a "progressive" mode that will yield an image of a prop that looks correct.

BTW, the "bent prop" effect isn't unique to digital cameras, many high end SLR film cameras use a shutter that only exposes a small part of the film at the same time when the shutter speed is high with a small shutter "window" that slides across the film.
Click to expand...

Chances are, in the OP at least, those blades are spinning at less than 1000 RPM, if that tells you anything. :dunno:
 
Matthew said:
The RPM match and the shutter speed match stop the props. The top image the prop looks normal, because the rotation of the prop is perpendicilar to the framescan.
Click to expand...

Even perpendicular to the frame scan, the blades are still moving fast enough for smearing, and I don't see any. That's what makes me think the first vid was recorded with a proper camcorder with full frame integration and not a cell phone CMOS imager.

--Carlos "does machine vision stuff as a day job" V.
 
Matthew said:
I do "machine vision stuff" for food, too! I don't have to deal with moving objects, though.
Click to expand...

I do. Surface robotic mobility and planetary entry/descent & landing.

We hate electronic rolling shutter. It doesn't take much movement to change object geometry. We can deal with blur better than shape changes, especially in stereo vision applications.

--Carlos V.
 
Unbeliever said:
I do. Surface robotic mobility and planetary entry/descent & landing.

We hate electronic rolling shutter. It doesn't take much movement to change object geometry. We can deal with blur better than shape changes, especially in stereo vision applications.

--Carlos V.
Click to expand...

Finally - when people ask me what I do I usually say "not much" since it's a whole lot easier to say that than explain it. I use high magnification machine vision pattern recognition for object measurements and locations. Then use that info for various other machine motions.
 
I think this one demonstrates the effect most, as the camera operator changes the camera orientation during the filming:

 
TangoWhiskey said:
I think this one demonstrates the effect most, as the camera operator changes the camera orientation during the filming:
Click to expand...

I don't get how the camera ever sees the object where is never is - in this case the prop blade descending under the wing. I could grok it if the image of the blades went behind due to persistence as the plane moves forward...or something. I guess it is a property of the image capture effect and relative motion. I'll have to look it up.
 
mikea said:
I don't get how the camera ever sees the object where is never is - in this case the prop blade descending under the wing. I could grok it if the image of the blades went behind due to persistence as the plane moves forward...or something. I guess it is a property of the image capture effect and relative motion. I'll have to look it up.
Click to expand...

I have absolutely no clue what a grok is. It's not even in my dictionary.

There's nothing weird going on. All the blade images are within the prop disk.

It's just a digital version of the classic air hockey table with a puck bouncing around under a strobe light with an open shutter camera mounted above. The exposure time, frame rate, image scan method and propeller rpm create the weird blade shapes and movement.

I don't know how many people missed it however while the blades are falling off the bottom, they're attaching to the top too. It makes a nice goofy cartoonish version of conservation of energy.
 
Last edited:
You must log in or register to reply here.
Back
Top