( Short Exposure ) CCD Camera

hatter

What I'm trying to build: a B/W camera system to take a snapshot of a very fast (250mph maximum) projectile.·The camera will be approximately 6-12 inches away from the projectile.

www.ovt.com mentions the following (which is great):

www.ovt.com said...

To better suit the intended application, we normally allow the user control over how fast the exposure and gain algorithms work in the chip.

Some datasheets are here:http://www.ovt.com/p_cameraChips.html·but I don't quite understand how to read them . I'm mostly concerned with exposure time, since I need as little blur as possible.

Can someone point me in the direction of what to look for in the datasheets, as well as other cameras to look at?

Any and all help/comments appreciated.

-noobie

steve_b

I'm not entirely sure if I've got my terms right....
but I thought exposure time was the amount of time spent reading an image....so, the old film cameras with their 600exposure times etc...
I think you want a combination of frame rates and a certain exposure.· The faster the frame rate the lower the exposure, and vice versa.

Someone can correct me on this!

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
·

Steve
http://members.rogers.com/steve.brady
"Inside each and every one of us is our one, true authentic swing. Something we was born with. Something that's ours and ours alone. Something that can't be learned... something that's got to be remembered."

hatter

In a traditional camera, shutter speed and the film speed both affect the net exposure, correct? Since a CCD is a fixed speed "film" in this case, wouldn't I need to control how long light·affects the array, to affect the exposure?

So, another question... is CCD sensitivity basically the "film speed"?

Edit: Thanks for the quick reply!

David B

What about leaving your camera fully open in the dark, then flash the target with a strobe light, like Doc Edgarton did with his famous bullet images? Would that approach suit your project?

kelvin james

I'm not a pro photo guy, but i can help out a bit. What you are looking for is a " fast shutter speed" to freeze a fast moving object. The amount of time the shutter opens will determine how much blur is captured when an object is moving. And with a faster shutter, a larger appature or light exposure is required to offset the lack of light hitting the ccd.That is why when capturing a very fast object a flash is required for the extra light, and the flash will appear to freeze the object. Since you did not go into detail about what you know about cameras, i will give a little detail about what you are attempting, to make your own. First, the ccd you are looking at is only the ccd chip, you would still have to buy the interface to control it, a lense adapter and lense, some sort of electronic shutter, and since most of these only output a raw format, you would need special photo software to edit and view the image. The video ccd won't help you out much either, as they top out at 30 fps, meaning the shutter opens/closes 30 times a second, which would not be fast enough for what you are trying to do. So, from a video chip you would be getting a shutter of 1/30th of a sec, and for what you are trying to do, you would be probably looking at 1/5000th /sec, just a guesstimation. You would be better off with an old 35mm SLR, with a good lense and flash, with the appropriate film. These cameras are pretty cheap these days with the digitals now owning the market. They have great manual control for shutter, exposure, etc. Not trying to deter you from a "project", but just wanted to let you know what you are getting yourself into. kelvin

Nate

Why not use the flash idea with an electronic shutter - fire the flash and then close the shutter ~1/5000 sec later.· You would have to experiment with the intensity of the flash and the actual time lapse between the flash firing and the shutter closing to get the right exposure and sharpness.· Of course you would still have to·deal with·the lens, ccd control, and decoding of the image data.· Hope you're going to make a lot of money with this project, because I'm guessing if you actually go forward with it by the end of it you will have learned a lot, but spent a lot of time with it.· [noparse]:)[/noparse]

nate

steve_b

If you want a 'ready to go' cheap (and hackable) digital camera....go grab one of them 'keychain' cameras.

I've seen them in Radioshack for $30Can.

I picked one up and it's from DIGIGR8.· They're only 100Kpixels....but this will get you going with some cheap experimenting.

If you can find a copy of dec2004 Nutz&voltz, there's a guy that hacked one of these and put a small 555 pulse trigger connect to the shutter button on this camera.· He mounted it in his R/C airplane and took some aerial pics.

Issue there was that he had to set the 555 timer wide enough in pulsewidths so that the camera didn't go through all it's pictures before he got up!· (this one only has 26pics but runs on one AAA battery).

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
·

Steve
http://members.rogers.com/steve.brady
"Inside each and every one of us is our one, true authentic swing. Something we was born with. Something that's ours and ours alone. Something that can't be learned... something that's got to be remembered."

hatter

Thanks a steve_b, I'll definitely grab one of those to hack, even if just for fun! Also, thanks again to everyone for your input on this. I’ll give you guys a rundown of what I think so far…

First of all, the link I posted, is for CMOS cameras, which I’m afraid are too slow to suit my purpose (too noisy as well). Second, I’m trying to create a digital system from head to toe, with no hand-tooling necessary. So, an old school camera won’t work (thanks for the suggestion though).

From what I’ve read so far (data sheets, product briefs, and explanations from across the web), most CCDs are fairly sensitive to infrared. This should help solve part of my problem, lighting, since I can create an IR strobe very easily. However, I can’t leave the CCD open because there will be ambient light in the environment(s) I’ll be shooting in. Like several of you have suggested, precisely controlling the strobe and shutter, is key to the crisp image capture I’m looking for.

So, the current prototype idea contains the following:

··· a) ·NTSC to VGA CCD w/lens
··· b) ·High speed CCD decoder/digitizer
··· c) ·LCD plane shutter
··· d) ·A fairly fast MCU (possibly a dedicated ccd controller)
··· e) ·Supporting ics, buffers, etc.
··· f) ·A ton of time

Here’s a hint… the projectile I’m trying to capture typically doesn’t travel more than 1000feet, and it spins a considerable amount (for some people, a lot more than others ).

If any of the assumptions I've made are wrong, please correct me!!!
·

Paul Baker

I by no means mean to burst your bubble, and I am an amateur photographer at best. But I don't know if you will be able to get no blur without using Edgartons method of a strobe in a dark environment. I've worked with LCD shutters before and thier capacitance is pretty bad (they are high) this means to get a fast, full cycle (opaque, clear, opaque) you'll need to pump a fairly large current pulse, but LCD's are sensitive things, if you overdo it they will permanently stick (typically remain forever dark) when the Eo exceeds a value (electromagnetic field between the plates).

Doing some quick trig calculations, assuming your equivalent format is 35mm (all but expensive digital SLR·CCDs are less than this, a smaller format=worse), the object is 12 inches away, and your depth of field is 1 inch (this is the range that is in focus, hence you object must be between 11.5" and 12.5") yeilds an angle of 90 degrees, meaning the path of your projectile within the field of view is 17". Traveling at 250mph and assuming the projectile is 0 dimensional (it has no width or height) it will be in the cameras sight less than 1/4000th of a second and this is the time it takes to travel the entire distance of the frame. An exposure of 1/40,000th of a second will make the projectile appear to be 1.7" long and this is for a 0" long object (a 1/2" object will appear as an object over 2" long). The LCD shutter I worked with was 35mm format and the fastest I could operate it was 1/100th of a second, a smaller shutter would be faster but it would also reduce the travel length of 17", and remember this is assuming your using a professional grade CCD (most consumer 2MPixel CCDs are 25mm or less). The cheapy pen cams will have a CCD around 10mm-15mm.

Im not saying its impossible, but you may need to approach the problem differently, a properly designed high speed synchro-flash can have illumination durations of 1/100,000th of a second, but to buy such an item is mucho-dinaro. You may be able to design an arc discharge lamp for a fraction of the price.

Paul

You maybe able to get some better figures by using an LCD that is mounted next to the lens (and hence is the size of the smaller lens not the CCD), the one I used was mounted to the format (the 35mm slide). I was able to get better than 1/100th of a second exposurse by combining the mechanical iris with the LCD where the LCD was clear, the iris opened, then the LCD went black but I still couldn't get 1/1000th of a second with·ISO-2000 and the picture have anything but dim shadows. Of course this was with physical film,·but CCDs have an equivalent ISO rating they are optimized for and·my·Minolta A2 only goes to ISO-800 and·I have problems with graininess·in the photos,·of course in such situations as standard·photography I am going to have·a higher standard of whats acceptable.

One final note, depending on your application it may be possible to skirt some of the issues, say if your just trying to measure the speed of the object, measuring·the projectile's streak duration (the time when you first register its presence till the time you last register its presence) may ease some of your timing constraints.

Post Edited (Paul Baker) : 1/20/2005 7:00:44 AM GMT

hatter

By no means did you burst my bubble! Your reply is what I was hoping for -someone with serious insight into the problems I will face.

By my calculation, the optimal distance the projectile would travel in the area I have to work with was ~1/20th of an inch, which equates to something around a·.00001s·time interval. Obviously, extremely difficult, but more likely impossible·to achieve with a single-stage shutter of any type. Even with a dual stage shutter, I think the tolerances will be too high to achieve the timing accuracy I want.

Even with an arc-discharge setup, I still could·face a problem with ambient light... Maybe a pulsed IR diode setup would be sufficient... EDIT2: Actually, an IR strobe might be perfect for my application. I know for a fact they are reliable up to 50khz, maybe I can pulse an LED "Lamp" quickly enough to get a clear picture...

There's probably something I can do to deal with the [noparse][[/noparse]remaining] blur in·software. Also,·adding another camera in a different axis might help too. I don't need beatiful images, just ones I can analyze efficiently.

Ah the gears are turning!

Thanks again for your reply.·Keep em coming!

EDIT: The main reason I need·a clear image, is to track rotation. Haha, quite the "project", isn't it?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- Hatter -

Post Edited (hatter) : 1/20/2005 7:36:42 AM GMT

steve_b

Ever see the Matrix?!
See that freeze 3D motion thing they do?
They use a matrix of cameras setup and they take a picture all at the same time and splice each image in to the film so it looks like it's real-life freezing.· cool stuff.

Well, are you stuck using one camera?· Why not set up 4 or more!
This way you could do some quick trigger toggling!
Fire them like your spark plugs fire!
1, 2, 1, 3, 2, 4, 3, 4· or some such order to overlap enough!

(and no, your plugs don't fire in that order! haha)

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
·

Steve
http://members.rogers.com/steve.brady
"Inside each and every one of us is our one, true authentic swing. Something we was born with. Something that's ours and ours alone. Something that can't be learned... something that's got to be remembered."

Paul Baker

You may be able to get away with the IR idea, I know most digital cameras put special coating on thier optics to reduce the CCD's sensitivity to IR. If you use a lens which doesn't have this coating and use an IR filter for a range of the IR spectrum that isn't prevalent in your ambient scene, it may work as well as being in a dark room. Does your projectile have markings which will enable you to track rate of rotation? You maybe able to find some IR paint that would work. Also by increasing the distance from the camera you'll increase the amount of time the projectile is in the camera's view, but depending on the size of your projectile, it may reduce your ability to detect rotation. A staggering of the IR lamp flash and the shutter can potentially yeild an exposure that is better that either can do alone. I strongly suspect this will require alot of trial and error at practically every phase of design, so Im glad to hear you have lots of time to devote to it, you'll likely need it.

Paul

Ive seen the staggered photo effect, but know little about it. So I have no idea if it would help, this approach would also require some hefty post processing time to stitch the photos together, especially if you have a real-time constraint. I do know that its difficult to compensate photo blur in post-processing, I've spent many hours in photoshop trying to recover a blurry photo with limited success, of course I only had one perspective to work with.

Post Edited (Paul Baker) : 1/20/2005 5:07:35 PM GMT

Nate

Golf gizmos sell because·desperate people·believe that they will make them better Golfers, but the gizmos seldom work·and are·almost never based on sound science.· I say fake the whole photo, give the customer what they want to hear, and make a ton of money.· [noparse]:)[/noparse]

nate

Paul Baker

Just got an idea, If you know that the rotation will be within a specific range. You could multiple expose your frame and flash in a sweep of frequencies then post process looking for your markings to line up with the previous exposures markings, by knowing the period between those two exposures you can calculate·the rate of rotation. If you mark your projectile with a line around the circumference·and lines down the length. The circumference line will denote the leading edge of the projectile for the exposure, the longitudinal lines will mark the rate of rotation, by calculating the distance between the exposures where the longitudinal lines line up you can get the flash period.

Paul

Though if Nate's assertion that this is for golf is true, doing this for a spherical projectile would complicate this method beause the projectile will be rotating on two seperate axis, the hook/slice rotation and the run/roll-back rotation.

Post Edited (Paul Baker) : 1/20/2005 5:30:18 PM GMT

steve_b

You can't multi-expose a digital image though....not in the film way!!

In some form of digital tv transmission (HDtv or something) they only transmit the pixels that have changed!· So, you transmit your background less often than the moving foreground (it's a 'swimming' kind of affect...not nice).

You could overexpose your image and leave the shutter open longer (depending on camera type)....hard to say if this would help or hinder.



as far as the camera 'array' goes....you'd have a bunch of stills that could just be viewed one after the other.· Do you need to have them in a movie sequence?· All movie cameras are just cascaded stills that give the perception of movement....what do they call that....visual perception or something?·

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
·

Steve
http://members.rogers.com/steve.brady
"Inside each and every one of us is our one, true authentic swing. Something we was born with. Something that's ours and ours alone. Something that can't be learned... something that's got to be remembered."

kelvin james

Maybe you would like one of these:www.visiblesolutions.com/phantomv5.html

Paul Baker

What I mean by multiple exposure is more than one flash strobe per photo, it would behave exactly as a traditional multiexposure would, sorry for the confusion, reading the CCD is destructive (for that matter so is traditional multi-exposure once you develop the film you can't reexpose the film). Multiexposure would also reduce bottleneck of reading the CCD.

Post Edited (Paul Baker) : 1/20/2005 10:51:29 PM GMT

hatter

Nate: you win! But I have to disagree with you. Typically, aids don’t work for a golfer because he/she has chosen poorly considering their specific problem area(s). It doesn’t help that people are extremely impatient and want instant results. Practice makes perfect, right?··
·
Paul: That’s exactly what I intend to do, either take two pictures, or double expose one. Double exposure –or more precisely, single exposure, double flash- does have a good side effect of keeping image output half the size. During CCD processing, I might be able to compress the data stream as well. The complexity added to the image processing step by using a single image is negligible.
·
As far as rotation tracking, I’m hoping a thick (1-2 mm) black line around the circumference of the ball will be adequate for rotation analysis. I’m working right now with some test images with a single line. The first section of code is in place to cull the scene, but the ideas for rotation aren’t implemented yet. As long as the image quality is clear enough, finding the rotation won’t be very difficult with a single plane dissecting the ball. So, obviously the only concerns are source image(s) and speed.
·
For right now, I’m going to get a prototype camera trigger in place using a visible-light flash and see what I can get for image quality.
·
Once again, thanks for “bench racing” with me

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- Hatter -

Paul Baker

A simple IR Emitter and transistor, where the ball stands in its path should work, There will be some lag time caused by the club head though.