Add physical shutter to 5D2 to eliminate wobble, Dan Vance style?

Lee Wilson said:
But the light (the charge) that is being converted to a digital value will still have a temporal offset
Click to expand...

No. Think of it this way: first you enable all the pixels, in whatever order and fashion yo like, after this step all pixels are ready to be exposed, then you open the shutter for a period of time, then shut it again, all pixels have now been exposed for the same amount of time, and more importantly at the same time, then you read it out, again in whatever order you like. The result has no rolling shutter artifacts. This isn't a revolutionary idea by far, after all it's exactly what the camera does in still mode. I suggest you google it instead of ruining this otherwise interesting thread with, no offence, redundant questions.
 
Last edited:
mattsand said:
Haha, said by a man who himself didn't get that my post was dead serious. What i suggested is exactly how film cameras implement their optical vf's. Same thing. :-)
Click to expand...

I am serious, too. Implementing a mirror shutter and focusing-screen based optical viewfinder is the last thing I am interested in given the 5D2 amazing LCD screen and "live view" capabilities, complete with near 1:1 pixel magnification for focusing purposes.

Smiley faces often imply that what precedes them is written in jest.

I just got a bunch of **** from the hobby shop...
 
mattsand said:
No. Think of it this way: first you enable all the pixels, in whatever order and fashion yo like, after this step all pixels are ready to be exposed, then you open the shutter for a period of time, then shut it again, all pixels have now been exposed for the same amount of time, and more importantly at the same time, then you read it out, again in whatever order you like. The result has no rolling shutter artifacts. This isn't a revolutionary idea by far, after all it's exactly what the camera does in still mode. I suggest you google it instead of ruining this otherwise interesting thread with, no offence, redundant questions.
Click to expand...

Ladies and gentlemen! Time to go back to the technical issues of this interesting idear. Matt got it, I was quoted to have gotten it :)... fine.

But in video mode it would be complicated to sync the shutter with the sensor. How does the sensor "know" when he has to change to "exposure mode" and finally to "read out mode"? I suspect the shutter must send a signal, a timecode, or whatsoever, to the sensor.
 
It doesn't really matter as long as the frequency is the same. You just have to adjust the phase once and it should remain constant at least until you turn off the camera. Possible sync sources are crystals or the black burst from the composite out. One obvious but still problematic thing is that the electronic shutter has to be slower so you'd need to wave the camera around until the auto exposure agrees.
 
PaulM said:
Smiley faces often imply that what precedes them is written in jest.
Click to expand...

Which not only means that it can't be true, it also means it's a personal attack as well as that the poster didn't "get" it? Damn, I need to get my tongue out of my cheek quick. :-)
 
From a technical standpoint how would you sync it? It almost seems that you would need to hack the circuit, or maybe rewrite the firmware. You'd need to rebody a lens to fit the shutter, or maybe rebody the camera itself.

I can't help but think that by the time a viable camera was created, we will have the D400, D800, Scarlet and maybe something even better from Canon that's 24p.
 
you can sync it with a crystal controlled motor and a phasing knob, then you just turn the knob until you don't see the shutter in live view. it might also be possible to use the black burst from the composite out, but i guess it's 59.94 and not 60 hz. this is not a unique problem, it's the same as when syncing cameras, projectors, tv screens or whatever that will otherwise strobe or get scanlines, and the solution is the same as well.

i think you can probably put the shutter in front of the lens to keep the design simple.

as for the waving around you need to lock the electronic shutter in a slow enough speed. too much light and it will be faster than the mechanical one, thus you lose the effect.

/matt
 
Thank you Matt, "got" it.

But I have some doubt that it will work that easy with life view. Basically, the rolling shutter effect happens, because the lines of the sensor get exposed and red out at the same time, line after line. If there is not such a thing as a "exposure mode" and "read out mode" there still would be a rolling shutter or worse - you might lose some lines on the upper or lower end of the frame if not syncronized perfectly.
 
it doesn't matter how it's read out, as long as all the pixels are exposed at the same time and for the same amount of time. think about it. the exposure doesn't happen momentarily when you read that line, it starts when the electronic shutter "opens" (at a different time for ech line) and ends when you read the data (again at a different time for each line but at a constant time after that line was opened). if during that time a mechanical shutter has exposed the entire frame evenly, what could possibly roll? i don't quite understand what you're suggesting. and what does it have to do with live view?

yes, you "lose" some lines (the shutter will be visible) if the shutters aren't perfectly synced, hence the phasing knob. i'm not sure how many times i have to say it before you realize that i'm not putting things on the table for discussion, these are hard facts. still a long way from a working device but the theory is simple, straightforward and tried and tested in varous applications for centuries. for two things to run in sync they need to have the same frequency (or even multiple) and the same phase, that's all there is to it.

/matt
 
by the way i think you're worried for the same reason i'm trying to explain that the mechanical shutter has to be faster than the electronic one. the mechanical shutter can only be open for the electronic shutter time minus the time it takes to "roll" over all the lines, from when the last row is enabled until the first row is read. this is not a sync issue though, just a simple prerequisite, hence the possibly redundant explanation in the post above. if this prerequisite is met the problem is void and if it's not no sync device can ever fix it. it has to be met.

/matt
 
I was just pixel peeping some of my test footage. It looks to me as if it takes around 1/24th of a second to read the sensor. That is to say that the top of a skewed line is where the bottom of the previous frame's skewed line was in the previous frame; assuming that the line was plumb in the frame before the pan.

If this is the case (is it really that long!?!), wouldn't that present a problem with this concept?

ADDED- THIS IS ON MY D90 FOOTAGE
 
Last edited:
Hey Matt, sorry for being a dick earlier! You definitely get it!

I just talked to an engineer at Jameco Electronics who worked for 10 years at Stereographics, a company that made the highest quality LC shutter glasses around. He's now got me thinking that liquid crystal shuttering is the way to go.

There's something called a Polymer Stabilized Cholesteric Textured (PSCT) liquid crystal optical-shutter which seems perfect:

http://www.lctecdisplays.com/PSCT technology.asp

www.lctecdisplays.com/files/datasheets/FOS-PSCT.pdf
75% transmissive with 30v applied, 97% light scattering with no voltage applied.

That's right, scattering, not blocking.

But it scatters so thoroughly that almost "no spacial information" is relayed to the sensor.

Capable of very fast in switching between on and off. Like half a millisecond.

Switches as fast as 1000 Hz!

Imagine putting the 5D2 in bulb mode and switching the shutter on and off a few thousand times. Could be useful for certain applications.

At LcTec, the 25.4 x 30.5mm version is about $350 USD, the 307.3 x 406.4mm version (!) is about $450.

Not sure how to drive them though...

Both the engineer I spoke with and Dan Vance have assured me that syncing LC shutters to a camera's framerate isn't the hardest thing in the world.

Biggest issue is that a PSCT LC shutter wants AC power. It doesn't use much, though. One of those portable 100w 120v AC-outputting Black and Decker things oughta work...

There's also a company in Santa Fe, http://www.retrievertech.com/Shutter/Shutter Right.htm, which sells one 50mm in diameter for I don't know how much.
 
mattsand said:
by the way i think you're worried for the same reason i'm trying to explain that the mechanical shutter has to be faster than the electronic one. the mechanical shutter can only be open for the electronic shutter time minus the time it takes to "roll" over all the lines, from when the last row is enabled until the first row is read. this is not a sync issue though, just a simple prerequisite, hence the possibly redundant explanation in the post above. if this prerequisite is met the problem is void and if it's not no sync device can ever fix it. it has to be met.

/matt
Click to expand...

Now I've got the clear picture! That's what I've missed. Thank you, Matt!:)
 
PaulM said:
75% transmissive with 30v applied, 97% light scattering with no voltage applied.
Click to expand...

75% transmissive = 1/2 stop of light loss

97% light scattering - not sure if there could be a ghost image under certain circumstances.

Would be nice to place that device directly in front of the sensor to keep everything compact. (There should be enough space if you quit or disable the mirror of the 5D)
 
You must log in or register to reply here.
Back
Top