User avatar
HermannSW
Posts: 1576
Joined: Fri Jul 22, 2016 9:09 pm
Location: Eberbach, Germany
Contact: Website Twitter YouTube

Is there any photo/video of flying rifle bullet done with v2 camera?

Sun Feb 24, 2019 9:55 pm

This is screenshot from "Pistol Shot Recorded at 73,000 Frames Per Second" on youtube:
https://www.youtube.com/watch?v=7y9apnbI6GA
Image


Is similar photo or video of flying rifle bullet ever done with v2 camera?

Since two days I know that it is possible (with very very bright light, 10000lm) to shoot 1000fps videos with 19.5µs(!) shutter time. But because of v2 camera being rolling shutter camera and 640x480 mode 7 line_time_ns=19517ns=19.517µs most likely a v2 camera bullet image will never look that sharp as above (the phantom camera has 73 times the fps than the v2 camera 1007fps).

I do not own fire arms nor have used any in the last 35 years after general conscription, so it is the question how to try v2 camera bullet photo/video. Anyway from movie I knew the name Glock and looked up its muzzle velocity as 375m/s. That is 19 times the 20m/s I was able to deal with with v2 camera in this posting:
https://www.raspberrypi.org/forums/view ... 3&t=234406

This is a frame from 1007fps video of closing mouse trap where the closing bar has 20m/s speed:
Image


The frame horizontal 320 pixels correspond to 10cm, the vertical 75 pixels to more than 2cm.
With that view a 20m/s bullet would move 2cm every millisecond of a 1000fps video.
So the 10cm allow to capture movements of up to 100m/s(!).
Now doubling the current distance between camera and mouse trap (36cm) would allow for 200m/s.
After several shots and videos some bullets should show up by probability > 0.5 of this setup.
With 72cm distance vertical 75 pixels correspond to more than 4cm.
As example, a 9x19mm Parabellum firearm cartridge bullet has 9mm diameter. That would results in perhaps 15 pixels vertically when shot from left to right through camera view.

Now to the rolling shutter effect. The 15th row will be captured 14*19.5µs=275µs=0.275ms after the 1st row. In that time the bullet will have flown 375m/s*0.000275s=0.103m=10.3cm ... with 1cm bullet length only a very distorted bullet frame (11.3cm x 0.9cm instead 1cm x 0.9cm) could be captured with v2 camera ...


P.S:
I once did a v1 camera 665fps video of a Nerf Toy gun, but the dart moves with only 18.6m/s:
https://www.raspberrypi.org/forums/view ... 3#p1277323
Image
⇨https://stamm-wilbrandt.de/en/Raspberry_camera.html

https://github.com/Hermann-SW/Raspberry_v1_camera_global_external_shutter
https://gitlab.freedesktop.org/HermannSW/gst-template
https://github.com/Hermann-SW/fork-raspiraw
https://twitter.com/HermannSW

ethanol100
Posts: 586
Joined: Wed Oct 02, 2013 12:28 pm

Re: Is there any photo/video of flying rifle bullet done with v2 camera?

Mon Feb 25, 2019 10:48 am

If you look at the "COMMISSION DELEGATED REGULATION (EU) 2018/1922 for "Community regime for the control of
exports, transfer, brokering and transit of dual-use items"
, you will find the following paragraph 6A003.a.5
6A003 Cameras, systems or equipment, and components therefor, as follows:
a. Instrumentation cameras and specially designed components therefor, as follows:
5. Electronic cameras having all of the following:
a. An electronic shutter speed (gating capability) of less than 1 μs per full frame; and
b. A read out time allowing a framing rate of more than 125 full frames per second;
This gives a hint that you would need to have a exposure of less then 1μs to get a picture of a bullet without motion blur. This paragraph is there to counteract weapon development.

To make an example of your data, to get a image with acceptable motion blur, your object should not move more than 2% of its length during exposure. This means for your about 19mm long projectile, we are only allowing 0.38mm of motion during exposure, taking your 375m/s velocity, this equates to about 1μs.

The classical way to take an image of a bullet is not about using the shutter but using a very short duration flash. If you get your flash down to about 1μs you will be able to capture a sharp bullet regardless of your exposure time(Take a look at Harold E. Edgerton photographs). This leaves only the problem of triggering the capture at the right moment...

jamesh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 23874
Joined: Sat Jul 30, 2011 7:41 pm

Re: Is there any photo/video of flying rifle bullet done with v2 camera?

Mon Feb 25, 2019 2:31 pm

What an interesting read. If you have trouble sleeping.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Contrary to popular belief, humorous signatures are allowed. Here's an example...
“I think it’s wrong that only one company makes the game Monopoly.” – Steven Wright

ethanol100
Posts: 586
Joined: Wed Oct 02, 2013 12:28 pm

Re: Is there any photo/video of flying rifle bullet done with v2 camera?

Mon Feb 25, 2019 5:50 pm

Yes, its boring, but it is the reason why you can't easily capture bullets with consumer grade cameras. Most cameras start with a minimum exposure time of 1µs, even the normal high speed cameras you can get in Europe. The Phantom cameras are limited to 1µs if you want to import them.

User avatar
HermannSW
Posts: 1576
Joined: Fri Jul 22, 2016 9:09 pm
Location: Eberbach, Germany
Contact: Website Twitter YouTube

Re: Is there any photo/video of flying rifle bullet done with v2 camera?

Mon Feb 25, 2019 6:18 pm

ethanol100 wrote:
Mon Feb 25, 2019 10:48 am
The classical way to take an image of a bullet is not about using the shutter but using a very short duration flash. If you get your flash down to about 1μs you will be able to capture a sharp bullet regardless of your exposure time(Take a look at Harold E. Edgerton photographs). This leaves only the problem of triggering the capture at the right moment...
Good information, but what I read the shutter duration is typically in the ms range, not µs.

And here it is mentioned that the flash duration lower bound is 30µs and not 1µs:
https://electronics.howstuffworks.com/h ... raphy2.htm
Sufficient flash durations can be as short as 30 microseconds, or 0.00003 seconds.
Do you know flashes that can produce shorter flash durations?

In 30µs the bullet travels a lot less, though:

Code: Select all

375m/s * 0.000030s = 0.01125m = 1.125cm

There is the next problem even with 30µs flash duration:
v2 camera takes 19.5µs to record a single line of frame in mode 7 (640xH).
30µs flash duration would allow to capture at most two lines of a single frame only ...


Perhaps its time to attack v2 camera registers 0x0162/0x0163 LINE_LENGTH_A which is constant for all 8 raspiraw modes in order to decrease line_time_ns for capturing frames that are less than full width:

Code: Select all

$ grep 016[23] imx219_modes.h 
      {0x0162, 0x0D},      // Line Length [15:8]
      {0x0163, 0x78},      // Line Length [7:0]
      {0x0162, 0x0d},
      {0x0163, 0x78},
      {0x0162, 0x0d},
      {0x0163, 0x78},
      {0x0162, 0x0d},
      {0x0163, 0x78},
      {0x0162, 0x0d},
      {0x0163, 0x78},
      {0x0162, 0x0d},
      {0x0163, 0x78},
      {0x0162, 0x0d},
      {0x0163, 0x78},
      {0x0162, 0x0d},
      {0x0163, 0xe7},
      {0x0162, 0x0d},
      {0x0163, 0xe7},
$ 
⇨https://stamm-wilbrandt.de/en/Raspberry_camera.html

https://github.com/Hermann-SW/Raspberry_v1_camera_global_external_shutter
https://gitlab.freedesktop.org/HermannSW/gst-template
https://github.com/Hermann-SW/fork-raspiraw
https://twitter.com/HermannSW


User avatar
HermannSW
Posts: 1576
Joined: Fri Jul 22, 2016 9:09 pm
Location: Eberbach, Germany
Contact: Website Twitter YouTube

Re: Is there any photo/video of flying rifle bullet done with v2 camera?

Mon Feb 25, 2019 10:18 pm

Thanks for the links, and pointing to "air-gap flash" for 1µs flash duration.
I found this article on building one:
http://www.glacialwanderer.com/hobbyrobotics/?p=490
The author essentially wants nobody to rebuild because it is really dangerous:
... The below information is for educational purposes only. Do not build one! If you go against my advice and do build one, I am not responsible for any injury, death, or any other problems it causes. ...

I now understand how the "Air Spark Flash" image in your previous posting was taken.
But this does only work with a camera taking the image altogether, like old cameras with film, or perhaps a global shutter camera.

v2 camera cannot do this, since it takes 19.5µs to capture a single line, the 1µs duration flash light will not even suffice to capture a single line of a single frame ...


P.S:
The air-gap flash from description uses a 35KV capacitor (that costs 90$ even for used).
Just remembered that I have a 1000KV cheap spark generator, not sure whether that can help, maybe it can:
viewtopic.php?t=207537#p1304035
Image
⇨https://stamm-wilbrandt.de/en/Raspberry_camera.html

https://github.com/Hermann-SW/Raspberry_v1_camera_global_external_shutter
https://gitlab.freedesktop.org/HermannSW/gst-template
https://github.com/Hermann-SW/fork-raspiraw
https://twitter.com/HermannSW

User avatar
Gavinmc42
Posts: 3886
Joined: Wed Aug 28, 2013 3:31 am

Re: Is there any photo/video of flying rifle bullet done with v2 camera?

Tue Feb 26, 2019 12:51 am

How fast can those flash LEDs on camera phones go?
I'm dancing on Rainbows.
Raspberries are not Apples or Oranges

User avatar
HermannSW
Posts: 1576
Joined: Fri Jul 22, 2016 9:09 pm
Location: Eberbach, Germany
Contact: Website Twitter YouTube

Re: Is there any photo/video of flying rifle bullet done with v2 camera?

Tue Feb 26, 2019 5:57 am

3 orders or magnitude away from 1µs:
https://en.wikipedia.org/wiki/Flash_(photography)
A full-power flash from a modern built-in or hot shoe mounted electronic flash has a typical duration of about 1ms, or a little less, so the minimum possible exposure time for even exposure across the sensor with a full-power flash is about 2.4ms + 1.0 ms = 3.4ms, corresponding to a shutter speed of about 1/290 s.

In order to achieve the 1µs for capturing a frame ethanol originally mentioned, and with the setup I described where a 9mm diameter bullet flies through 15 lines of a frame only, 1µs would be sufficient for 15 lines. This would require 1µs/15=66ns for line_time_ns somehow (the current value is 19517ns).

So first step to see if that is possible at all, is trying to reduce line_time_ns for raspiraw with some tuned register settings (perhaps a 100x20 or 20x100 frame in top left position of 3280x2464 sensor array):
https://github.com/rellimmot/Sony-IMX21 ... et_XXX.PDF
⇨https://stamm-wilbrandt.de/en/Raspberry_camera.html

https://github.com/Hermann-SW/Raspberry_v1_camera_global_external_shutter
https://gitlab.freedesktop.org/HermannSW/gst-template
https://github.com/Hermann-SW/fork-raspiraw
https://twitter.com/HermannSW

ethanol100
Posts: 586
Joined: Wed Oct 02, 2013 12:28 pm

Re: Is there any photo/video of flying rifle bullet done with v2 camera?

Tue Feb 26, 2019 8:50 am

Gavinmc42 wrote:How fast can those flash LEDs on camera phones go?
White LEDs won't work because of the phosphor which will glow for a longer time. You can use R,G and B LEDs. But the required current and number of LEDs to get a good lighting is insane. It would be possible, though.
HermannSW wrote: I now understand how the "Air Spark Flash" image in your previous posting was taken.
But this does only work with a camera taking the image altogether, like old cameras with film, or perhaps a global shutter camera.

v2 camera cannot do this, since it takes 19.5µs to capture a single line, the 1µs duration flash light will not even suffice to capture a single line of a single frame ...
The v1 cam had a register to perform a "global reset" where after that all lines are stated to be exposed on the same time. The readout is then longer for the later lines, but as the flash behaves like a global shutter in a way, this does not matter.

For the v2 camera, I think a long enough exposure should be fine, too. If you start each line with your 19.5µs each, you could wait for about 1ms to "flash" and then start reading the frames out with i.e. 1.1ms exposure time. As previously this will then behave similar to a global shutter system.

User avatar
Gavinmc42
Posts: 3886
Joined: Wed Aug 28, 2013 3:31 am

Re: Is there any photo/video of flying rifle bullet done with v2 camera?

Tue Feb 26, 2019 10:47 am

High power 820nm IR LED 8ns rise time, 14ns fall time.
Yep those white phosphors are slow.
Does it have to be color?
I'm dancing on Rainbows.
Raspberries are not Apples or Oranges

User avatar
HermannSW
Posts: 1576
Joined: Fri Jul 22, 2016 9:09 pm
Location: Eberbach, Germany
Contact: Website Twitter YouTube

Re: Is there any photo/video of flying rifle bullet done with v2 camera?

Wed Feb 27, 2019 5:40 pm

There is a LED flash from Vela that has adjustable flash duration in range 0.5µs..5µs, unfortunately it does cost £790.00:
https://shop.vela.io/products/vela-one- ... mera-flash

One point to note is that the Vela One provides 1,000,000lm for such short flash durations.
So even when adding a parabolic reflector to the 1000KV Boost Step up High-voltage Generator
https://www.aliexpress.com/item/New-100 ... 75803.html
the effective focussed lumen might not be bright enough to get meaningful frames in 1µs flash duration. The photo from the spark in previous posting was not that bright, although unfocussed.
⇨https://stamm-wilbrandt.de/en/Raspberry_camera.html

https://github.com/Hermann-SW/Raspberry_v1_camera_global_external_shutter
https://gitlab.freedesktop.org/HermannSW/gst-template
https://github.com/Hermann-SW/fork-raspiraw
https://twitter.com/HermannSW

User avatar
HermannSW
Posts: 1576
Joined: Fri Jul 22, 2016 9:09 pm
Location: Eberbach, Germany
Contact: Website Twitter YouTube

Re: Is there any photo/video of flying rifle bullet done with v2 camera?

Wed Feb 27, 2019 9:14 pm

I looked at the parabolic reflectors of Vela One:
Image

and they looked like this 10$ 75mm reflector from aliexpress:
https://www.aliexpress.com/item/75mm-Di ... 29520.html
Image

There is a similar 92mm reflector on aliexpress as well, but Vela One product page says "Dimensions 26 x 26 x 10 cm", so it cannot be the 92mm reflector. I just ordered one 75mm reflector, if it is good enough for Vela One, it should be good enough for me. Delivery from China to Germany with free shipping in maximal 32 days ...
⇨https://stamm-wilbrandt.de/en/Raspberry_camera.html

https://github.com/Hermann-SW/Raspberry_v1_camera_global_external_shutter
https://gitlab.freedesktop.org/HermannSW/gst-template
https://github.com/Hermann-SW/fork-raspiraw
https://twitter.com/HermannSW

User avatar
Gavinmc42
Posts: 3886
Joined: Wed Aug 28, 2013 3:31 am

Re: Is there any photo/video of flying rifle bullet done with v2 camera?

Thu Feb 28, 2019 1:25 am

Osram has become my favourite IR led maker. 8ns rise 14 ns fall time.
https://www.osram.com/os/ecat/OSLUX®%20 ... e_2219950/

White leds have a typical 120 degree light output, that does not match the 60 degree or so Pi camera lens.
I have not strobed my IR LEDS yet, others have and got 60m+ spotlighting with the pi cameras.

Got some VCSEL laser diodes OPV332, 100ps rise and falls times but probably not a wide enough beam spread for this application?
I'm dancing on Rainbows.
Raspberries are not Apples or Oranges

User avatar
HermannSW
Posts: 1576
Joined: Fri Jul 22, 2016 9:09 pm
Location: Eberbach, Germany
Contact: Website Twitter YouTube

Re: Is there any photo/video of flying rifle bullet done with v2 camera?

Sun Mar 03, 2019 1:44 pm

Thanks, that are impressive rise/fall times.


In the other thread
https://www.raspberrypi.org/forums/view ... 7#p1436715
I am still trying to get the correct register settings for global shutter mode with line_time_ns shutter time for v1 camera.


The Vela One flash light has £38 Velo Pop flash sound trigger as addon:
https://shop.vela.io/products/vela-pop- ... to-trigger
Trigger speed: ~7µs (measured from point sound passes threshold)
A 375m/s bullet will only fly 7*0.375mm = 2.625mm during that time, quite impressive.


Since the v1 camera does not expose FREX and STROBE pins some mechanism for triggering air-gap spark flash (STROBE) as well as capturing a frame (FREX) need to be done otherwise.

If using two PIs with one v1 camera each a bullet can be captured with any mode, even with 5MP.

1st Pi does:
  • 640x128_s 665fps raspiraw recording
  • its camera is positioned to capture 1.2m in horizontal FoV
  • frames are not store but analyzed immediately
  • 640x128_s tool does really capture 320x64 raw Bayer frames
  • the left bottom green pixel of 2x2 rg/Gb raw Bayer data is brightest
  • take only left bottom green pixels from 2x2 as monochrome approximation
  • that results in 160x32 byte GRAY8 image
  • scan the 160x32=5120 bytes of a frame for significant change
  • if bullet detected then detect horizontal position in in frame as well
  • 665fps gives a frame each 1500µs or 1.5ms
  • bullet flies 37.5*1.5 = 56.25cm in that time
  • this will trigger two or three times in 1.2m FoV
  • use first trigger and horizontal trigger position to send signal to 2nd Pi
2nd Pi does:
  • its camera is positioned with FoV 0.6m behind the 1st Pi camera FoV
  • with M12 lens camera gets adjusted to get a near full FoV image of bullet
  • exposure is set to line_time_ns = 32503ns for mode 2
  • raspiraw is waiting for global reset single fame getting triggered by frex i2c when triggered
  • triggers air-gap spark flash after some configurable delay as well to fire sometime within the frame exposure

This should result in a single 5MP 2592x1944 image of flying bullet. Although exposure time is 32.503µs, since scene is "dark" and spark flash will light for 1µs only , the global reset of v1 camera becomes global shutter as well, and there will be no rolling shutter effect. Bullet will move 0.375mm in 1µs only and even big image of bullet will be sharp. This should be doble with 2 Pi Zeros, two v1 cameras and an air-gap spark flash or the Osram leds Gavinmc42 mentioned. Even though the frames of 1st camera will be more blurry than the inages with xenon flash shown before, the left/right positions within horizontal FoV should allow for needed precision triggering of 2nd Pi.


P.S:
The technique for 2nd Pi camera will allow to get a single (almost sharp) 5MP "global shutter" frame of closing mouse trap in flight if triggered in time with normal (constant) lighting, since closing bar moves only

Code: Select all

0.000032503s * 20m/s = 0.000650060m = 0.650060mm
during the 32.5µs of exposure, and bar thickness is 1mm.
Image
⇨https://stamm-wilbrandt.de/en/Raspberry_camera.html

https://github.com/Hermann-SW/Raspberry_v1_camera_global_external_shutter
https://gitlab.freedesktop.org/HermannSW/gst-template
https://github.com/Hermann-SW/fork-raspiraw
https://twitter.com/HermannSW

User avatar
HermannSW
Posts: 1576
Joined: Fri Jul 22, 2016 9:09 pm
Location: Eberbach, Germany
Contact: Website Twitter YouTube

Re: Is there any photo/video of flying rifle bullet done with v2 camera?

Tue Mar 12, 2019 12:57 am

This posting shows that capturing frames in FREX mode with global reset works fine with raspiraw:
viewtopic.php?f=43&t=218576&p=1440654#p1440654

It also describes that minimal exposure time is less than the 21.165µs (line_time_ns) in normal rolling shutter mode.
Minimal exposure time for FREX mode capturing with mode 7 640x480 46.5MHz pixel clock is 2.753µs(!).
It seems that no air-gap flash is needed to capture a 375m/s flying bullet because it will fly 2.753µs*0.375mm/µs = 1.032375mm only(!!) during frame exposure time. I have taken a photo of Parabellum 9x19 bullet only (top) and used gimp to see what 1mm of movement will look like. Not perfect, and less than 1µs exposure time would be better/sharper, but bottom construction shows what v1 camera global reset frame with 2.753µs exposure time will look like:
Image


After successfully capturing FREX mode frames today I ordered a 0.5 Joule spring load airsoft pistol.
Then I discovered muzzle speed calculator (German language):
https://www.softair-waffen-shop.eu/mete ... erechnung/

After having seen that maximizing muzzle speed is done by minimizing ammo weight for fix 0.5 Joule I ordered 1.2g ammo. The calculator says muzzle speed will be 28.86m/s. That is an order of magnitude less then the Glock 375m/s, but a good start for test at home. If all goes well I will have to contact local Schützenverein on whether they can help me with recording real flying bullets later.
⇨https://stamm-wilbrandt.de/en/Raspberry_camera.html

https://github.com/Hermann-SW/Raspberry_v1_camera_global_external_shutter
https://gitlab.freedesktop.org/HermannSW/gst-template
https://github.com/Hermann-SW/fork-raspiraw
https://twitter.com/HermannSW

User avatar
HermannSW
Posts: 1576
Joined: Fri Jul 22, 2016 9:09 pm
Location: Eberbach, Germany
Contact: Website Twitter YouTube

Re: Is there any photo/video of flying rifle bullet done with v2 camera?

Tue Mar 12, 2019 2:05 pm

These are my high speed home devices for testing high framerate and/or minimal exposure capturing with Raspberry cameras:

Years ago Motor test station, where outmost radial speed was 18.7m/s maximally:
https://forum.arduino.cc/index.php?topi ... msg2762009
Image


Recently closing mouse trap bar with nearly maximal radial speed measured as 20m/s:
viewtopic.php?f=43&t=234406&p=1433883#p1433883
Image


As described in previous posting the ordered airsoft pistol shoould give 28.86m/s linear velocity.


RC airplane propeller 20140rpm maximal radial speed measured as 49.5m/s:
viewtopic.php?t=190407#p1319617
Image
Radius of propeller is 47mm, this calculation gives speed in m/s:

Code: Select all

$ echo "pi=4*a(1); 0.047*pi*20140/60" | bc -ql
49.56281290058377472513
$
.


I revived outrunner motor 25.5cm diameter propeller experiment from 1.5 years ago, and it just worked:
https://twitter.com/HermannSW/status/892821584628641793

First test with 11.6V loaded 900mAh 25C 3S lipo showed 5222rpm maximally, which is 69.7m/s radial speed at the outside:

Code: Select all

$ echo "pi=4*a(1); 0.255*pi*5222/60" | bc -ql
69.72293655744507623712
$
Image


Reading the motor spec they talk of "Load current: 18.5A" and "Power (Watt): 240".
Seeing only 2.12A means I did something wrong (most likely the small breadboard cables cannot deliver higher amps).

Instead of 3S lipo I connected constant voiltage power supply with 12V and 5A maximal.
Even with that setup only 2.4A and 7.3V showed up measured, while power supply showed 2.540A with 12V.
Maximal rpm was a bit higher than before, 5517rpm means 73.6m/s or 265km/h:

Code: Select all

$ echo "pi=4*a(1); 0.255*pi*5517/60" | bc -ql
73.66170834688327950981
$

This is already 20% of 375m/s Glock muzzle speed, but at home and radially allowing for easy camera experiments.
⇨https://stamm-wilbrandt.de/en/Raspberry_camera.html

https://github.com/Hermann-SW/Raspberry_v1_camera_global_external_shutter
https://gitlab.freedesktop.org/HermannSW/gst-template
https://github.com/Hermann-SW/fork-raspiraw
https://twitter.com/HermannSW

User avatar
HermannSW
Posts: 1576
Joined: Fri Jul 22, 2016 9:09 pm
Location: Eberbach, Germany
Contact: Website Twitter YouTube

Re: Is there any photo/video of flying rifle bullet done with v2 camera?

Wed Mar 13, 2019 8:08 am

Two small corrections to the speed numbers.

For the RC airplane the 20140rpm were determined from 1007fps video only roughly. Later I did add a small reflective mark to one of the two propeller blades and used digital laser photo tachometer to measure real 18233rpm:
viewtopic.php?f=43&t=190407&start=25#p1396192

Slightly reduced speed is 44.8m/s:

Code: Select all

$ echo "pi=4*a(1); 0.047*pi*18233/60" | bc -ql
44.86984943477378175587
$ 

For measuring speed of big propeller on outrunner motor I did not add reflective material, the blades were reflective enough. And that is a problem, both blades triggered tachometer measurements, so the real rpm value for the propeller is only half the displayed rpm, which means 36.8m/s:

Code: Select all

$ echo "pi=4*a(1); 0.255*pi*(5517/2)/60" | bc -ql
36.83085417344163975490
$ 

This table shows the corrected speed numbers, speed type and whether the experiment is one-time (closing mouse-trap or airsoft pistol shot) or not (easier for camera since frames can be taken anytime). I let the outrunner motor propeller at bottom because it will definitely be faster than RC airplane propeller when powered with >10A instead of <3A.
Image
⇨https://stamm-wilbrandt.de/en/Raspberry_camera.html

https://github.com/Hermann-SW/Raspberry_v1_camera_global_external_shutter
https://gitlab.freedesktop.org/HermannSW/gst-template
https://github.com/Hermann-SW/fork-raspiraw
https://twitter.com/HermannSW

User avatar
HermannSW
Posts: 1576
Joined: Fri Jul 22, 2016 9:09 pm
Location: Eberbach, Germany
Contact: Website Twitter YouTube

Re: Is there any photo/video of flying rifle bullet done with v2 camera?

Wed Mar 13, 2019 12:52 pm

One more speed to update (above table is new version already) -- I just received the airsoft pistol bullets, and I misread their weight. The product description states 0.12g(!) and not 1.2g. In fact they are a bit lighter, I did weigh 50 and the precision digital scale showed 5.8g, so one bullet does weigh 0.116g on average. Now I did run muzzle speed calculator again with 0.5Joule and 0.000116kg and got 92.8m/s(!!) calculated. I have to say that I completely underestimated what a 0.5Joule airsoft pistol can do muzzle speed wise. I did calculate that a 3D printed 6mm diameter ball from PETG of a colleague would weigh more (0.156g, or "only" 80.0m/s if the ball could handle the pressure on shot).

I will verify the (theoretical) 92.8m/s speed with a [email protected] v2 camera video. The bullet will move "92.8m/s × 0.001s = 9.28cm" between two successive frames, I just need to make FoV of v2 camera capture more than 20cm horizontally and get at least three frames with bullet. Because of rolling shutter effect lines will be skewed. With 19.5µs shutter time each line will be nearly sharp (bullet moves "92.8m/s × 0.0000195 = 1.8mm" during exposure time, 30% of bullet diameter). 6by9 said that a single 640xH mode line takes 31µs to read out, so a next line will start capturing bullet "92.8m/s × 0.000031 = 2.87mm" further away from the airsoft pistol.

Btw, 92.8m/s (or 334km/h) is already a quarter of target muzzle speed of Glock with 9x19 Parabellum bullet ...

Will have to wait until Saturday to capture the video, then airsoft pistol is planned to arrive.

Kinetic energy of the flying bullet is 0.005Joule only (W_kin = 0.5*m*v, 1Nm=1Joule), a hundedth of airsoft pistol 0.5Joule value:

Code: Select all

$ echo "0.5*0.000116*92.8" | bc -ql
.00538240
$ 
Image
⇨https://stamm-wilbrandt.de/en/Raspberry_camera.html

https://github.com/Hermann-SW/Raspberry_v1_camera_global_external_shutter
https://gitlab.freedesktop.org/HermannSW/gst-template
https://github.com/Hermann-SW/fork-raspiraw
https://twitter.com/HermannSW

User avatar
HermannSW
Posts: 1576
Joined: Fri Jul 22, 2016 9:09 pm
Location: Eberbach, Germany
Contact: Website Twitter YouTube

Re: Is there any photo/video of flying rifle bullet done with v2 camera?

Thu Mar 14, 2019 12:03 am

HermannSW wrote:
Wed Feb 27, 2019 9:14 pm
and they looked like this 10$ 75mm reflector from aliexpress:
https://www.aliexpress.com/item/75mm-Di ... 29520.html
Image
The reflector arrived today, took only 15 days from China to Germany, and it seems to fit the 5000lm (and 10000lm I recieved earlier this week) leds well (photo through welding shield safety glass, I was surprised that I could convince the camera to focus on the 5000lm led and not on the safety glass):
Image


The reflector just sits vertically on 5000lm led, the light on 2m above ceiling looks much brighter subjectively than 5000lm led alone without reflector.


I found a way to see how the light goes out of the reflector, just holding an A4 piece of paper 10cm above the reflector, and taking photo of the top:
Image
⇨https://stamm-wilbrandt.de/en/Raspberry_camera.html

https://github.com/Hermann-SW/Raspberry_v1_camera_global_external_shutter
https://gitlab.freedesktop.org/HermannSW/gst-template
https://github.com/Hermann-SW/fork-raspiraw
https://twitter.com/HermannSW

User avatar
HermannSW
Posts: 1576
Joined: Fri Jul 22, 2016 9:09 pm
Location: Eberbach, Germany
Contact: Website Twitter YouTube

Re: Is there any photo/video of flying rifle bullet done with v2 camera?

Sun Mar 17, 2019 2:03 pm

In order to have a high speed radial experiment for first "global reset with 2.65µs exposure = global shutter" photos/videos with Arducam ov5647 camera I did replace the thin cables with thick ones. And I repleced the 3S lipo with my constant-voltage power supply. That can only deliver 5A (instead of 18A the outrunner motor can take, but it is a start). So the power supply is used with 12V/5A. In order to avoid two blade measurements per round I did add reflective material to the end of one blade, and black tape to the end of the other (you can see Raspberry v1 cameras mounted on left and front side of motor test station, they are not used now):
Image


As can be seen when run at maximal measured rpm value (4082) the voltmeter/amperemeter shows 11.9V and 4.09A. The reflective material on one blade paints a nice circle, I did use Android phone camera with flash to take the pictures. You can see the outrunner ESC speed tester being outside of motor test station for safety:
Image


As said the digital laser tachometer showed 4082rpm maximal, that is 54.5m/s or 196km/h at the outer radius:

Code: Select all

$ echo "pi=4*a(1); 4082/60*pi*0.255" | bc -ql
54.50192015080252799707
$ echo "pi=4*a(1); 4082/60*pi*0.255*3.6" | bc -ql
196.20691254288910078945
$ 

So how far will the blade (only) move with the (hopefully) 2.65µs exposure time? Only 0.14mm ... to be verified now:
viewtopic.php?f=43&t=235523&p=1443478#p1442871

Code: Select all

$ echo "pi=4*a(1); 4082/60*pi*0.255*0.00000265" | bc -ql
.00014443008839962669
$ 


So this is updated list with new speed for outrunner propeller, up 48% for now:
Image


P.S:
The airsoft pistol arrived yesterday as well, but it will have to wait. The test shots seem to confirm high speed of the 1.2g bullets. If I shoot into a 3m distant cardboard box. the yellow 6mm diameter bullet gets reflected to me so fast that I cannot see it at any time, neither before hitting the box, nor after.
⇨https://stamm-wilbrandt.de/en/Raspberry_camera.html

https://github.com/Hermann-SW/Raspberry_v1_camera_global_external_shutter
https://gitlab.freedesktop.org/HermannSW/gst-template
https://github.com/Hermann-SW/fork-raspiraw
https://twitter.com/HermannSW

User avatar
HermannSW
Posts: 1576
Joined: Fri Jul 22, 2016 9:09 pm
Location: Eberbach, Germany
Contact: Website Twitter YouTube

Re: Is there any photo/video of flying rifle bullet done with v2 camera?

Sun Mar 17, 2019 9:33 pm

Just trying 1007fps 640x150_s capturing of airsoft pistol bullet with Raspberry v2 camera.
This was my initial setup, at the far end a little cardboard box with cardboard in front that has air behind allowing to swing back a bit and reduce the energy of the airsoft pistol bullet when being hit. A measuring tape for being able to determine bullet position in order to determine its speed by looking at several frames with bullet on. A small shooting slit at bottom of front motor test station wall, and a 5000lm led lying on top of plexiglas safety cover of motor test station:
Image


The first videos I took with that setup looked very washed, and I had to increase exposure to 200µs in order to see something.
Then I remembered what I forgot, the 5000lm light needs do be focused on the scene!
I used the reflector I received recently and just put 5000lm led on top.
This is Android smartphone photo without flash, it nicely shows the focusing of the 5000lm led, will see whether videos become better with it (as expected):
Image
⇨https://stamm-wilbrandt.de/en/Raspberry_camera.html

https://github.com/Hermann-SW/Raspberry_v1_camera_global_external_shutter
https://gitlab.freedesktop.org/HermannSW/gst-template
https://github.com/Hermann-SW/fork-raspiraw
https://twitter.com/HermannSW

User avatar
HermannSW
Posts: 1576
Joined: Fri Jul 22, 2016 9:09 pm
Location: Eberbach, Germany
Contact: Website Twitter YouTube

Re: Is there any photo/video of flying rifle bullet done with v2 camera?

Mon Mar 18, 2019 8:29 pm

I really wanted to start with sub-3µs exposure global reset videos of Arducam ov5647 today.
My fist target was the outrunner propeller.
I found a 150W/12V mains adapter I bought for my "unlimited inhouse drone flight" project, that can deliver 12V at 12.5A:
http://forum.arduino.cc/index.php?topic ... msg3786115
I connected it and did some experiments with the propeller.
I lost 3(!) ESCs today (the red Electronic Speed Contol elements in previous photos).
During my months long circlular robot tests with two outrunner motors driving the wheels directly I never lost any ESC.

So while I searched for a 4th ESC (I am not sure whether I have) I found an old mini drone that stopped flying years ago.
I tested powering one of the 4 motors directly with 3.7V and it worked.
So I took a 2xAA battery holder, soldered its two cables directly to mini drone motor and used my constant voltage digital power supply's big clamps to connect to the battery holder. Finally I glued the mini drone into motor test station (near the place where the red ESC was) and now have a mini drone motor I can control from 0.01V in 0.01V steps up to 3.7V and above! And most importantly, totally reliable/repeatable:
Image


I did several rpm measurements, the rpm reported from digital tachometer have to be devided by two because both blades get counted for one rotation of propeller. I did measure 19K/39K/53K/65K rpm for 0.7/1.7/2.7/3.7 V. Increasing voltage above 3.7V does not have an effect on the measured rpm, keeps 65K. I went down to 0.00V and increased in 0.01V steps. When reaching 0.10V the motor starts to spin. Then it is possible to reduce to 0.06V while the motor keeps spinning, not less. At 0.06V the constant voltage power supply shows 0.016A delivered, and digital tachometer measured 348rpm.

Yes, the rotor diameter of 34mm is minimal compared to the outrunner motor 25.5cm propeller diameter -- but the rpms are much higher (even when devided by two). The circumference is more than 10cm, and with 65Krpm this mini motor rotor does 57.8m/s(!) or 208km/h:

Code: Select all

$ echo "pi=4*a(1); pi*0.034" | bc -ql
.10681415022205297010
$ 
$ echo "pi=4*a(1); 65000/2/60*pi*0.034" | bc -ql
57.85766470361202547460
$ echo "pi=4*a(1); 65000/2/60*pi*0.034*3.6" | bc -ql
208.28759293300329170856
$ 

Since for the sub-3µs exposure video recordings only speed is important to me, this mini drone motor is all that I need. The rotor moves 0.14mm in the planned 2.4225µs exposure time, or 0.21 angular minute:

Code: Select all

$ echo "pi=4*a(1); 65000/2/60*pi*0.034*0.0000024225" | bc -ql
.00014016019274450013
$
⇨https://stamm-wilbrandt.de/en/Raspberry_camera.html

https://github.com/Hermann-SW/Raspberry_v1_camera_global_external_shutter
https://gitlab.freedesktop.org/HermannSW/gst-template
https://github.com/Hermann-SW/fork-raspiraw
https://twitter.com/HermannSW

User avatar
HermannSW
Posts: 1576
Joined: Fri Jul 22, 2016 9:09 pm
Location: Eberbach, Germany
Contact: Website Twitter YouTube

Re: Is there any photo/video of flying rifle bullet done with v2 camera?

Mon Mar 18, 2019 9:27 pm

The plan was to capture 5MP images with Arducam ov5647 camera.
But running raspivid with mode 2 or 3 (5MP) leads to all kinds of issues (failed to enable port, too many macroblocks with Segmentation fault, ...). I gave up and will do 2MP images in mode 1 (1920x1080). This is a totem video player screenshot from "raspivid -md 1 -fps 30 -o tst.h264" video:
Image


Nearly centered rotating propeller, with camera lens to propeller distance 8cm. Recording from the side is needed since I plan to light the scene with 5000lm led lying on motor test station top safety plexiglas shining vertically onto the scene. Although motor is running at minimal speed (0.06V with 348rpm), no details. What changes with using mode 1 over modes 2/3 is pixel clock, mode 2/3 have 80MHz pixel clock, mode 1 68MHz. The minimal strobe pulse changes slightly from 1/80µs to 1/68µs, not of importance. Exposure time unit is 128/68µs or 1.88µs, so setting exposure to "2" will keep it above real exposure of 2.5µs.


P.S:
This is how same propeller looks with Raspberry v1 camera global (external) shutter capturing:
https://github.com/Hermann-SW/Raspberry ... ter-videos
Image
Last edited by HermannSW on Wed Sep 18, 2019 3:19 pm, edited 1 time in total.
⇨https://stamm-wilbrandt.de/en/Raspberry_camera.html

https://github.com/Hermann-SW/Raspberry_v1_camera_global_external_shutter
https://gitlab.freedesktop.org/HermannSW/gst-template
https://github.com/Hermann-SW/fork-raspiraw
https://twitter.com/HermannSW

User avatar
HermannSW
Posts: 1576
Joined: Fri Jul 22, 2016 9:09 pm
Location: Eberbach, Germany
Contact: Website Twitter YouTube

Re: Is there any photo/video of flying rifle bullet done with v2 camera?

Tue May 14, 2019 7:20 pm

There will never be a v2 camera photo of a flying rifle bullet because of missing global reset function that v1 camera has.
The shortest I was able to capture sofar with Arducam ov5647 camera was with 6µs strobe pulse length.
I did get sharp frames from propeller moving with 124m/s (1/3rd of bullet target speed).
But others at MIT did it first, they not only captured bullet hitting apple with 0.5µs strobe, but also from 16(!) Raspberry cameras at the same time from different angles👍:
https://www.raspberrypi.org/forums/view ... 3&t=240442
Image


P.S:
Nice, they use "i2cwrite.c" I wrote for my "global reset + low µs strobe" experiments with (Arducam) v1 camera in their github repo.
⇨https://stamm-wilbrandt.de/en/Raspberry_camera.html

https://github.com/Hermann-SW/Raspberry_v1_camera_global_external_shutter
https://gitlab.freedesktop.org/HermannSW/gst-template
https://github.com/Hermann-SW/fork-raspiraw
https://twitter.com/HermannSW

Return to “Camera board”