HDR Cameras

These Cameras are going to be the future of Digital Photography, these Cameras

may not even need a exposure setting, because a single HDR exposure will

capture everything! and exposure will then be made on the computer after the

picture has been taken. May be a while before these Cameras are on sale,

cannot wait. Whats other PN members veiw?

How will a "single" HDR exposure be taken in-camera, when multiple-exposures are required to create the single HDR image?

You need multiple exposures now to create a HDR image, but it is predicted that HDR cameras the whole range of exposures will be captured, to be adjusted on the computer.

<p>Wasn't it Fuji who, a couple of years ago, came out with a camera whose sensor had one large and one small photozone per pixel, to try to solve the dynamic range problem? Did anything ever come of that?</p>

Are you talking about developing a sensor that has more dynamic range than current sensors?

 

If this were to be developed (and I hope it will) how will that eliminate exposure settings? Exposure is based on: ISO of the capture media, shutter speed, aperture.

 

If you eliminate my ability to select aperture, how will I achieve shallow depth of field? If you eliminate my ability to select shutter speed, how will I capture motion blur or freeze motion for that matter?

Sorry Steve, dont know about that, just been reading a bit on HDR, i think Photography will not be as much about the Camera, but more about PC Processing (Photoshop) good thing for new Photographers (like me) but i suppose it is talking the Skill away from Photography.

Sorry Rob not an expert on HDR, Look up Paul Debevec a leading expert in HDR Imaging. www.debevec.org.

Terry,

 

My questions were more rhetorical.

The first time I read on LL who referred to HDR as the Holy Grail I was really excited and when I tried the various techniques I knew then that for me at least HDR was the future.

 

I use Photomatix as I just can't duplicate what it does in Photoshop, even though there are those who swear by the Ctrl, Alt, Shift, tilda method. The contrast I get in clouds in low lighting conditions is really impressive with Photomatix if you bracket say -1.67, 0, +1.67.

 

At the moment though, there are some problems with Photomatix (for example, even in RAW the -1.67 capture has a lot of noise in it that effects the final tone mapped file), and the HDR plug-in in Photoshop CS2 is next to useless.

 

The biggest problem with bracketing is that if things move the HDR effect doesn't work no matter how you post process. Well, that's not quite true. I've played with deleting say a bike rider going through one of my bracketted shots, but this is all too much work.

 

So, Terry, I am in with you. If and when this camera comes, we may well have one hand on the Grail of photography.

What "camera" is this? Debevec is talking about using multiple images from a single camera, which restricts HDR to subjects that are either stationary or moving together so motion doesn't affect the relationship between the exposures. Fuji's brilliant approach, the auxilliary pixel sensor beside each primary pixel sensor, works pretty well by getting, essentially, two simultaneous images to work with, but everybody else is separating their exposures over time.

 

There are other approaches. One is to use multiple matched lenses and sensor arrays, possibly with the arrays each having different sensitivity. This could work in some cases, but eliminating paralax could be a killer if the subject couldn't be kept at least a few dozen yards away from the camera. Possibly useful for some things, but it's never going to fly for general purposes. (Debevac is talking about seven exposures for sky photography, can you imagine buying seven identical 600mm lenses and figuring out how to mount them really close together, and perfectly parallel, and then being able to actually pick the thing up to shoot?)

 

Another would be to split the image from one lens onto different sensors, like the old Technicolor system did. This would involve transfer lenses that took the primary image and created a zone of parallel rays, as is done in some microscopes, so that you could dip into the image path multiple times. This would certainly get you limited depth of field, as the extra optics and the division of the light would mean that anything slower than an f:/1.4 would probably be completely unuseable, although in full sun and really high-sensitivity sensors slower lenses could be possible.

 

The only process that seems even vaguely practical would be a camera with electronics so fast that it could take five (or seven) shots in series, with the sensitivity of the sensor doubled for each exposure, with the entire sequence taking place within the time we currently use for a single exposure. Say you're willing to live with a maximum shutter speed of 1/500, or 2 milliseconds. Call that 2,000 microseconds. Within that time, the camera would have to "expose" five times and read the sensor data four times (the fifth time could be after the 2 milliseconds are over without causing blur). A camera that can currently take ten frames a second is probably going to use over half of that interval (say 50+ milliseconds) to read the sensor for each image. Assuming that our magical HDR camera was going to use half of that 2 milliseconds for exposure and half for data transfer, that's 250 microseconds per image, or a 200:1 improvement in data transfer rates. Are there any cameras today that can hit 10 fps?

 

Even if one of these three paths can someday be implemented in a camera that a normal person could lift, you'd still need to set the exposure. Life doesn't fit in a four-stop range, and any one scene is going to either need 20 exposures to come up with a set that can render the image or you need to get that center exposure pretty close to correct to start with.

 

In short, it ain't going to happen this way. I would consider it far more likely that sensors with broader dynamic range will come along, which would be more useful. And even then, any photographer is going to want to set exposure rather than having some one-size-fits-all approach take over.

 

Van

<p>Hmm ... how about a sensor that takes Fuji's idea of higher- and lower-sensitivity photosites, but instead of implementing the sites side-by-side, stacks them on top of one another? Like Foveon's idea of stacking RGB on top of one another, but instead of stacking colours, it would stack two sensors for the same colour on top of one another, with a sensitivity difference of maybe two or three stops between the sensors. Dang, maybe I should have patented that instead of posting it here :-)</p>

Here's my idea. Let the user select the brightest value allowable in the histogram, and individual photosites then shut themselves off as they reach say 250,250,250 for example. Also may as well let the lower limit be user specified as well so that individual photosites will stay activated at least until they reach 12,12,12 (for example). The overall exposure will obviously be as long as it takes to reach a minimum level in the shadows, but the individual photosites will not all be "on" for the same length of time. In other words, the sensor will stay powered on, and the shutter will remain open as the individual photosites record then shut down as per user parameters until the entire image has been recorded.

 

I don't know if existing technology allows each individual photosite to be controlled this way or not, but it would seem to me to be the ideal solution.

Frank, if the speed of reading the sensor was high enough your suggestion might be a way to accomplish things by taking successive readings at different times in the exposure, but letting the sensor continue to accumulate photons. You would still need to read the sensor multiple times, but instead of a 20ms exposure followed by 10, 5, 2.5, and 1.25 for a total of 38.25 ms they could just read it at 1.25, 2.5, 5, 10, and 20 ms. If that's possible, it would cut the speed needed in half.

 

Van

Thankyou all for your informative response (a little out of my depth some of it) Thankyou all Terry :-)

How about a camera that let's you focus AFTER the picture has been taken? Check: http://graphics.stanford.edu/papers/lfcamera/ . Iam

Unbeleavable Iam Nemo, makes you wonder what will be avalable in a few years time (at a price) Great stuff thanks for the link. Terry.