Developer-friendly / experimental branch (dataghost) - page 10 - CHDK Releases - CHDK Forum
supplierdeeply

Developer-friendly / experimental branch (dataghost)

  • 143 Replies
  • 75981 Views
*

Offline LjL

  • ****
  • 266
  • A720IS
Re: Developer-friendly / experimental branch (dataghost)
« Reply #90 on: 06 / July / 2008, 19:01:31 »
Advertisements
Right, but I was under the impression that the "orientation sensors" were kind of separate from the IS mechanisms (although IS might use their outputs, too, somehow), and mostly just for telling portrait from landscape.

The article you linked to about the SD870 would appear to confirm that, no?

Anyway, I've made another set of measurements, this time without the integration thing, I merely "integrated by eye" (with all the associated drawbacks). I don't know what to make of them, except remarking that the first 360 degrees do look kind of similar to the ones I posted before...

(Probably a completely stupid suggestion: could this be measuring the Corioli effect?)
« Last Edit: 06 / July / 2008, 19:10:44 by LjL »

*

Offline DataGhost

  • ****
  • 314
  • EOS 40D, S5IS
    • DataGhost.com
Re: Developer-friendly / experimental branch (dataghost)
« Reply #91 on: 06 / July / 2008, 19:26:47 »
Did you do something like 10 measurements and averaged them? One measurement (set, as in A/B?) isn't enough to get accurate results.

Anyway, these readings aren't necessarily from the IS mechanism. I've seen them referenced with different names though it's not easily apparent. I don't exactly recall whether or not I saw them referenced with non-IS names but I've seen them referenced with different 'purposes'.

*

Offline Bg~

  • *
  • 27
Re: Developer-friendly / experimental branch (dataghost)
« Reply #92 on: 06 / July / 2008, 19:51:51 »
Coriolis effect: coriolis acceleration from the earth's rotation is much too small to be measured these accelerometers.

LjL, is this what you mean by the integrated measurements?
filter_tau = 10;
EViAInt = (EViAInt*(100-filter_tau)+_GetISLensEViAaxis()*filter_tau)/100;

*

Offline Bg~

  • *
  • 27
Re: Developer-friendly / experimental branch (dataghost)
« Reply #93 on: 06 / July / 2008, 19:57:47 »
BTW DataGhost, where do these functions come from: _GetISLensEViAaxis() (and all the related ones)?  Are these just the function names as they exist in the Canon firmware?


*

Offline LjL

  • ****
  • 266
  • A720IS
Re: Developer-friendly / experimental branch (dataghost)
« Reply #94 on: 06 / July / 2008, 20:05:37 »
Yes Bg~, that's exactly what I meant, except I didn't use a "filter_tau" variable ;)

DataGhost, what Bg~ described is how I "averaged" the results - in the graphs where I did so. In that last one I posted, I just averaged by eye, which is of course a terrible method, but I just wanted to double-check somewhat (mostly fearing that my method was either not very sound, or suffered from rounding problems).

Anyway, the first graph at least (the one with just the plain pseudo sine and cosine when rotating the camera around the lens axis, accompanied by a lot of numbers) should be relatively accurate, since for most angles I averaged quite a few of those (already "average", integrated, or filtered, or what you want to call it) readings.
For angles or 30, 60, 120, 150, 210, 240, 300 and 330 I did just one reading, as those angles required hand-holding the camera... still, the resulting graph looks very much like a sine and a cosine without many bumps.
« Last Edit: 06 / July / 2008, 20:12:58 by LjL »

*

Offline LjL

  • ****
  • 266
  • A720IS
Re: Developer-friendly / experimental branch (dataghost)
« Reply #95 on: 07 / July / 2008, 13:33:00 »
Other graphs for your sinusoid enjoyment...

I've placed the A720 on a turntable (as in, gramophone), distance from the center about the radius of a 45rpm disc.

The graphs are what happens during 20 seconds to the A and B axis, turning at 33 and 45 rpm.
Note that there definitely is some tilting of the (flimsy) turntable with the relatively heavy camera on it - at 45 rpm especially, I could hear clicks and see one side of the turntable slightly sink (1mm or so, I'd say).

The X is the count of calls to dg_draw_orientation_demo(), which is what did the logging. I logged time(NULL), too, to make sure the graphs contained exactly 20 seconds (from the last "n"th frame to the last "n+19"th frame).
« Last Edit: 07 / July / 2008, 13:35:37 by LjL »

*

Offline Bg~

  • *
  • 27
Re: Developer-friendly / experimental branch (dataghost)
« Reply #96 on: 07 / July / 2008, 14:09:29 »
I'd be interested in seeing what the response looks like when you have the turntable stationary and then turn it on.

*

Offline LjL

  • ****
  • 266
  • A720IS
Re: Developer-friendly / experimental branch (dataghost)
« Reply #97 on: 07 / July / 2008, 15:16:18 »
Sure, I'll give you the whole graph and the data.

The times when I start, stop and switch to 45 rpm (I think that time i stopped, switched and then restarted) should be easily guessable.


*

Offline LjL

  • ****
  • 266
  • A720IS
Re: Developer-friendly / experimental branch (dataghost)
« Reply #98 on: 07 / July / 2008, 15:33:50 »
And here is also a graph done over a long period of time (2926 seconds) with the camera steady on a hopefully flat surface.

You can see that this time the A axis value is much closer to 4000 than to the 3500 of other graphs... it definitely varies at least when you take out the batteries, and here clearly it also increases with power-on time (temperature?).

The "rogue" values at the beginning are, of course, me hitting the SET key to start logging.

*

Offline LjL

  • ****
  • 266
  • A720IS
Re: Developer-friendly / experimental branch (dataghost)
« Reply #99 on: 08 / July / 2008, 10:08:53 »
Unrelated to the sensors - I've posted some patches for this build in another thread. I suppose they aren't really anything specific to this build, it's just that it's the one I have on my hard drive.

 

Related Topics