when are we going to get better dynamic range?

[mrklaw]

never mind noise (although thats welcome). Are there any advances in sensors coming that will give us more dynamic range?

How about being able to apply a digital ND grad to a scene, to bring the brightness of the sky down?

[Anke]

....How about being able to apply a digital ND grad to a scene, to bring the brightness of the sky down?

Why not use a regular filter ND? Why add more technology for the sake of it?

[qtfsniper]

We're getting new technology for sure. I dream of the day a digital camera sensor is able capture the dynamic range a human eye can see in one frame.. but for now sony has developed a back-illuminated cmos sensor. Here is the article, but it claims a HUGE boost. It might be a revolutionary sensor (for me). http://www.sony.net/SonyInfo/News/Press/200806/08-069E/index.html
and from another forum

"Sony says 8 decibel improvement. 6 in sensitivity (gain) + 2 due to noise reduction.

8 decibels = 6.3x.

Thats 2 1/2 stops."

[mrklaw]

Why not use a regular filter ND? Why add more technology for the sake of it?


because I want to take a simple shot with a bright sky without it blowing out. I'm not hugely into landscapes, so I don't have the investment of a set of filters etc. I understand if you are then its worth it, but I think some assisting tech would help in this situation. I don't think its tech for tech's sake.

[JCH77Yanks]

because I want to take a simple shot with a bright sky without it blowing out. I'm not hugely into landscapes, so I don't have the investment of a set of filters etc. I understand if you are then its worth it, but I think some assisting tech would help in this situation. I don't think its tech for tech's sake.


There's a "digital" ND grad in Lightroom 2.0.:D

[johnz]

Why not use a regular filter ND? Why add more technology for the sake of it?

Come oon.
Like why build IS lenses because you can always use a tripod?
Or why build better LCD's in cameras because you can view them better at your computer anyway?

[blackcap]

Why not use a regular filter ND? Why add more technology for the sake of it?

Exactly. They should have stopped at the wheel.

[cdifoto]

Why not use a regular filter ND?
Because filters suck.

[DC Fan]

If you want more dynamic range, switch to medium format - either film or a Phase One P45.

[poloman]

Soooo
How do you propose to implement this built in ND filter in an elegant way?
Sort of like hoping for underwear that does it all for you when you need to eliminate. (and is dry when you are done)
"Here's my new camera. It weighs 15 lbs but look at all the features!"

[johnz]

I think the OP gets too much critic for his queestion.

Come oon, it's not far fetched, it would be great to have, and it will happen oneday - quite soon i think. My guess is that it will be done by controlling the ISO of pixel individually, so that you can have iso 100 in some parts and iso 400 is others.

i would love it.

[Stocky]

A build in NDG filter should be easy. They just need the shutter to move faster at the top of the frame than at the bottom. Canon are you listening?

[mrklaw]

well I'd hope Canon have experts who do the design, but how about this?

lets say you have a 3 pixel sensor. You set a shutter speed of eg 1/100 of a second. The shutter opens, but instead of waiting for the full duration and then reading the sensor data, the sensor data is read in realtime (or at least in discrete, timed lumps - eg every 1/1000 of a second).

So you end up with not only the actual final exposed sensor data, but also a snapshot of that sensor data while it was being collected.

Pixel 1 is showing a value of 200 after the exposure, so that is left alone.
Pixel 2 is showing a value of 255 after the exposure. So the processor now starts looking back to see how long it took to fill up. Its data shows it was full halfway through the open shutter time (1/200 second), so it can extrapolate how bright it would have been after the full amount of time (512).
Pixel 3 also shows as full, and it filled up very quickly, after only 1/400 of a second, so it is the equivalent of 1000.

After doing this, the camera now has either a RAW image with expanded values that you can tweak, or it can go back and apply an ND filter to overexposed pixels, bringing them back below 255. Perhaps it'd need to only apply the ND filtering across a rectangular area, not sure.


No idea if thats possible at all. Might need a completely new type of sensor or method of reading it. Don't know, don't care :P

[poloman]

That's an interesting idea....not nearly enough data to give you all the gradations in a graduated filter.

[Doug Pardee]

How about being able to apply a digital ND grad to a scene, to bring the brightness of the sky down?

The problem isn't in the digital domain, or there would be Photoshop filters for it. The problem is the quantity of photons collected by the sensor, and by the noise generated from various sources. A "digital ND grad" won't help there.

Since it sounds like you're interested in extra highlight headroom, and not too concerned about noise in the shadows, there are the Fuji DSLRs with their Super CCD SR sensors, and the newer Canon DSLRs with Highlight Tone Priority.

Fuji's sensor has two kinds of sensor element: normal and highlight. The data from the two groups of sensor elements are combined to produce the final photo. It's also possible to take photos with just the normal sensor elements, to reduce file size and increase camera responsiveness.

Canon's Highlight Tone Priority basically "uprates" the sensor by one stop, giving an extra stop of highlight headroom in the Raw data. The in-camera JPEG processing compensates for that, and provides a somewhat film-like roll-off of the extra highlights—remember, a JPEG can't express over 100% white. Canon's Raw software (Raw Image Task and Digital Photo Professional) can do that too.

[Zivnuska]

never mind noise (although thats welcome). Are there any advances in sensors coming that will give us more dynamic range?

How about being able to apply a digital ND grad to a scene, to bring the brightness of the sky down?

FWIW, I liked your question.

[waterj2]

My idea for a digital grad filter would be to find the horizon (if digital cameras can find a face, they can find a horizon), and decrease the time the shutter is open for the pixels above it (as I understand it, for things like Live View, an electronic "first curtain" is used, while the second curtain operates normally, so it should be feasible to open an electronic curtain some interval after the physical curtain passes, depending on the vertical position of the pixel). If you toss in a reasonably decent auto-horizon finder, and can get a smooth transition, it would offer at least some advantages over grad filters in that it allows the horizon to be at any height (as well as, presumably, some range of hard to soft grads), and finds it automatically. Of course, I suppose it would only work with a level horizon in landscape (you could shoot backlit corner walls in portrait, I guess)

I'm not sure it would be worth it, but if it's easily implemented, might be a feature that consumers would be interested in. I also wonder how much ISO sensitivity might be able to be mapped out pixel-wise (in the AE stage), and if that would have any chance of being recovered in RAW.

[CyberDyneSystems]

Fujii had an interesting concept, with it's differing sized and sensitivity photo sites, it added quite a bit of DR over a standard CCD of the time.

However it can be taken further. Jeremy just touched on it, and that it is now being talked about more often, I am convinced it will happen, not if but when.

Back in 2004 was the first time I mentioned this idea.
We often are blinded by what we think we know and ignore the obvious as being too new.
Again in 2004 the idea of adding iso adjustment to the two factors that make up our program AE (aperture and shutter priority) was scoffed at. In fact allowing the camera to auto adjust the ISO to maintain the program AE now looks perfectly acceptable with the noise capabilities of modern high end DSLRs.

So.. what does this have to do with dynamic range?

Again, lets think ahead, and outside of the box that confines us to rudimentary thinking.

Everyone of your CMOS photo sites is hooked up to your cameras processing computer already.
With a more powerful and faster set of CPUs, we could adjust that same ISO we adjust now on a "per pixel" level.

We would be able to adjust how much exposure leeway we would let the camera impose, and set it to back off the ISO pf any pixels approaching blow out, and/or boost the ISO pf pixels in severely dark areas. Taken too far it would be a mess, but with small incremental adjustments, you could double or triple the dynamic range with no changes to the CMOS itself.

[blackcap]

Soooo
How do you propose to implement this built in ND filter in an elegant way?
Sort of like hoping for underwear that does it all for you when you need to eliminate. (and is dry when you are done)
"Here's my new camera. It weighs 15 lbs but look at all the features!"

Why would it weigh heavier? The filter should be controlled electronically, via the LCD. Or are you thinking of a camera with a piece of plastic built onto the body? :lol:

However you wouldn't need a GND filter (electronic or otherwise) if the camera had high enough dynamic range to start with. I'm sure they'll get there one day, I just wish it would be sooner rather than later.

[poloman]

Why would it weigh heavier? The filter should be controlled electronically, via the LCD. Or are you thinking of a camera with a piece of plastic built onto the body? :lol:

However you wouldn't need a GND filter (electronic or otherwise) if the camera had high enough dynamic range to start with. I'm sure they'll get there one day, I just wish it would be sooner rather than later.

I had my tongue in my cheek. :)

CyberDyneSystems....you have the most sensible idea so far....IMO.