How much better will digital get?

[BJL]

To keep the load light I want to use a medium format view camera (6x7 or 6x9) ... the current scanning back is smaller than 4x5 and with some design and production adjustments can work in a small format view camera.

Since I am fairly sure that digital sensors will overall be smaller than their film counterparts and that single shot sensors will not go beyond 645 format, and larger format photography often has little in the way of speed constraints so that scanning backs can give the best price/performance, I expect that there is a niche for scanning digital view cameras in formats like 6x7 and 6x8. For example, the Fuji GX680 might have an interesting digital future.

[BJL]

To take two pictures 1,000th of a second apart is just not physically possible with current sensor technologies ...

Yes, but from the little I have read about the Stanford research, there are possible new sensor technologies that could overcome this.

Anyway, the Fuji SR idea is clearly the one to watch for now. For one thing, it works with existing focal plane shutters while the other ideas do not. (I think of SR as simultaneously recording two images, one through an ND filter.)


P. S. I just found this reference mentioned in another thread, on a related dynamic range enhancement method being offered commercially for surveillance video cameras. One quote suggests that this is related to what I mention above: "Building upon technology developed at Stanford University in the 1990s, Pixim has created an image capture and processing system that provides high-quality pictures with enhanced dynamic range."
http://www.pixim.com/technology/technology.phtml

[Kenneth Sky]

To make smaller less noisy sensor sites and higher resolving lenses we will have to await a breakthrough in nano technology. That could be tomorrow or in 50 years depending who you read.

[Jonathan Wienke]

Of couse we save in film, but then didn't we charge the film costs to the client anyway? we sure arn't charging the computer time...

Speak for yourself. If you aren't, you should be...

[John Camp]

For those interested in narrow-angle mosaic photos, and haven't seen it, look at this:

http://www.tpd.tno.nl/smartsite966.html

(This is the Delft photo; I suspect many have already seen it. I was amazed to click on the tower in the far background and being able to see the clock face and tell what time the photo was shot. Also, there are numberous artifacts in the streets, like half-buses. But this technique hold the possibility of printing enormous photographs that will resolve down to loupe level.)

[jani]

From a satellite 100Km up, that resolution limit at 50% MTF would be about one line pair per 14 meters. With the far lower MTF needed to barely resolve text, the graphs suggest a limit of about 20,000 lp/radian, still 5 meters per line pair.

Maybe on a very clear day they can see grapefruit, on the top of a mountain above all the dust and heat haze!

From what I recall, such satellites often have a distinct orbit, bringing them as close as 30-40 km from the ground in extreme cases. But that's still nowhere near good enough to look at license plates lying flat on the ground.

[Stef_T]

John, I don't think that I am going to help solve your dilema, I have the same problem, deciding whether to wait, or to get something now. However, on the topic of what you want, and how good something is, let me tell you a short story.

This is a true story, that happened, or rather I saw, durring the summer, a year ago, about 2 minutes walk from my house in Toronto. I was crossing the street, when I car pulls up on the red light. it was a brand new corvette, driven by a young teenager, I'm guessing that he was 19, no more then 21. He thinks that he is 'all that'. Listeing to lound rap music, rocking his head, arm up on the driver's seat. We lock eyes for a second, and I get the impression that he is telling me 'haha, you wish you were me!' Now a corvette is a very nice car, and I wish I have one even though I can't drive yet, but here's the good part of the story, and remember this actually happened: Beside the corvette drives up a red Ferrari, driven by an older man, probably in his 40s. I don't exactly remember what Ferrari it was, It might even have been an F50, tho my memory is a bit hazy on that fact. The teenager in the Corvette looks over, and it was like his heart skipped a beat. His face was as pall as snow. He litterally, turned off his music, and sunk so low in his chair, that I could no longer see him. He couldn't blieve that the car beside him, was so much better then his. AS the light turned green, the Corvette was a little slower going off then the Ferrari, the kid was obviously still shaken after being showed up in such a way.

Now, I know that had nothing to do directly with photography, but the moral remains the same in both cases. No matter what you get, how much you spend, there will always be something better, that your equipement will look like crap compared to. in that moment you would do anything to have that better equipment, but you can't. All you can have is the knowledge, that one day, even that equipment will be outdone, will get old and will look bad compared to something newer and better.

The moral of all this: don't worry so much about the future, and if someone else's equipment will be better then yours or he got a better deal on it. Life's too short to wait, get the best that you can now, and enjoy it. If you simply save up your money waiting forever, you will die a very rich, lonely and sad man.

Carpa Diem, Stefan Tarnawsky

[didger]

I realize the 24x36 sensor has it's limits in megapixels.  Maybe once 35mm reaches 22mp, things will slow down a bit, and Medium Format will begin to develop more rapidly.

The problem is not so much that 24x36 sensors have pixel density limits (who knows what those limits might prove to be?), but that even now, with 16.7MP, the bottleneck is lens quality.  Even for 11MP, very few lenses keep up.  For 16.7, perhaps NO lenses really keep up.  For 22MP, what's the point?  I can't believe there will ever be a 22MP 24x36 sensor unless there's a very dramatic breakthrough in lens technology.  I see no hint of this happening.  Modern lenses are on the whole not even as good as those old Leica and Zeiss designs, so where are these miracle lenses for ultra high density sensor cameras supposed to come from.  

1dsMKII is a sort of a stopgap measure.  The real step up will have to be medium format.  As to whether Mamiya is qualified; I'd bet yes, but until the camera is out and thoroughly tested, we're all just speculating. I'm surprised that some of you couch your speculations in such confident phrasing.  Have you had a secret sneak preview of ZD, and found it lacking?

[Dan Sroka]

That's why I mentioned the idea of a used 1Ds...

Yes, one great thing about the digital market is its robust used market. Photography has enough early adopters that for those with a little patience, many deals can be had. (Although for me $4K is still way too much for a body.)

[Bobtrips]

Some little upstart like Foveon could make a breakthrough but they  will be bought out or forced to keep the prices high.  

Foveon is manufactured under contact by National Semiconductor.  Hardly a little upstart, that National Semiconductor.

[Bobtrips]

Maybe theres a different way to approach the "I want more pixels" issue.

Think back to Michael's comparison of the 1D MkII and the Pro1.  Both 8 megs and if shot at lower ISO settings the quality was quite close.

The Mk II has a 24 x 36 mm sensor - a total area of 862 mm sq.
The Pro I has a 6.6 x 8.8 mm sensor - a total area of 58 mm sq.
The Pro I's sensor is only 6.7% of that of the Mk II.  

Them pixels is squeezed in tight!  

Seems to me there's some powerful resolving going on in the Pro I's glass.  Is there something about smaller lenses that makes them better?  (Or am I lost in space?)

How would a small sensor 16/24/32 meg camera perform?  (The sensor might have to be larger than the Pro I's 2/3".)

(Forget high ISOs, dynamic range, and shallow DOF for the moment.  I think there are solutions to these problems.)

[didger]

If someone should happen to point this message out to Michael he might just search his memory who he might have met that would call him a jerk on a public forum.  Is this the way to create a kinder gentler LL?  Michael Fatali is just another struggling human being doing the best he can with the hand life is dealing him.  We're all "jerks" one way or another now and then.  I found him fun to hang out with at his house a bit and in the field a bit and at his galleries and I find his photography and his enthusiasm for it beautiful and inspiring.  I'll certainly look him up again when I get back into another big Colorado Plateau shooting phase.

[BobMcCarthy]

So, increasng pixel and sensor size gives you choices like
(a) less visible noise, but less DOF and/or longer exposure time ("same highlight range, better shadows, more OOF or motion blur"), and
( same DOF and shutter speed, same visible noise levels, more highlight headroom ("same shadows, more highlight range").

Can software improvments (ie post processing in the camera or raw converter), overcome flaws and errors,and subsequently remold the data, to provide a higher quality output. I know lens aberations can be mitigated, can not noise (errors) at the pixel level be smoothed or reassembled.

bob

[Bobtrips]

OK, I'm trying to put all this together in my possibly-slower head.

Small photo sites would be fine if they didn't fill as quickly.  (Putting the DOF issue to one side for the moment.)

Smaller photo sites would mean smaller sensors; smaller, lighter, cheaper, and easier to build lenses.

There are two possible solutions for the 'filling too quickly' problem.

First, 'back of sensor' processing.  Build a CMOS chip that can either measure the amount of time that it takes the site well to fill and/or count the number of times it fills.  

Dynamic range would be limited only by the minimum shutter speed that was necessary to freeze motion.

Second, take multiple readings of the sensor.  For example, when the shutter was set to 1/250th the camera might actually take four 1/1,000 exposures and blend the four to achieve a very high dynamic range.

(Smart algorithms could probably detect the point at which motion blur appears and set the maximum shutter speed after the capture.)

OK, how befuddled am I?    

[Bobtrips]

And the DOF issue.

First, very deep DOF works for most landscape shots.  That means that a smaller sensor would be preferable.

But for those times when one wants to isolate the subject with a shallow DOF seems like you could do so with multiple, very fast exposures.  (My K-M A1 has a max. electronic shutter speed of 1/16,000th second.)

Let's say the camera takes 3-6 quick exposures to get the 'blend' shots.  

The it takes one or more shots on each side of 'focus' to determine which parts of the frame are closer and which are farther than the chosen focus plane.

(Sure we would need some serious buffer space and processing speed, but Moore's law is still working for non-sensor stuff....)

Now all we need is 'brokeh algorithms' and we can dial in the DOF back at home.

(Hey, I'm not a sci-fi fan so let me know if I'm drifting too far in that direction.)

[didger]

Peter, I like all your ideas and maybe most of them are even doable, but not for free.  However, what are "legacy lenses" that would allow you to use 25 Mpixels effectively in 35mm format?  I don't know of any lenses shorter than 35mm that can even keep up with 16 Mpixels or more than barely 11.  What lenses are you talking about?  If Zeiss distagon lenses are "legacy", I'm afraid you'll be disappointed in the ultrawide realm.  They're OK for 11 Mpixel, but for anything over that I'll move to medium format.  

Also, are you so confident that noise free 25 MPixel 35mm format can be done, or for that matter, that it can be done period?

And, yeah, shooting conditions and forecast for a few more days are not so great here.

[danborge]

This discussion has been focused on the optical, mechanical and electronic technicalities of achieving maximum quality images for a given format size. Very interesting and entertaining, but what if we  turned the question around? What are the MINIMUM digital camera/lens specs needed to produce an exhibition quality 16x20 print?  Assume that "exhibition quality" is defined as good enough that discerning print buyers would not perceive any technical flaws in the print, at normal viewing distance, that would detract from the aesthetics of the image or their willingness to buy it.  No lupes or microscopes allowed.  Assume that we are talking about realistic landscape photography and not art shots that employ deliberate distortions. Also assume that the print buyers do not know what camera was used.

If this is the goal (for at least some photograhers), do they need any more than a Digital  Rebel and the latest image processing software for reducing lens defects (DxO or equivalent), reducing noise (Neat Image or equivalent), upscaling (Genuine Fractals or equivalent) and sharpening (PhotoKit or equivalent)? Given the proper processing workflow, would discerning print buyers, at normal viewing distance, perceive any significant difference between a 16x20 print of a Digital Rebel image and 16x20 print of a high-end digital camera image? If so, where would the difference lie?

At what point would this photographer rationally decide that further investments in technically better gear would reduce, rather than increase, his net profit from selling 16x20 prints?

[Bobtrips]

This discussion has been focused on the optical, mechanical and electronic technicalities of achieving maximum quality images for a given format size. Very interesting and entertaining, but what if we  turned the question around? What are the MINIMUM digital camera/lens specs needed to produce an exhibition quality 16x20 print?

I  think there's something very important in this question that should be considered.

If this question were to be carefully answered then that data would provide a lot of help in deciding the 'how many pixels' question.

I'd love to see a real piece of carefully done research.

Something along the lines of the same shot/same conditions with various levels of cameras (8 meg fixed-lens, 8/12/16 meg dSLR, 16/22 meg MF).  Print the files out at standard sizes from 8x10 to huge.

Then hang the prints in a public place and have people judge them/have advanced photographers judge them.

Heck, throw in 35mm and a few sizes of film and lets get these arguments behind us.


(And, yes, it would take some work to make it a good study.  All research is that way.)

[DiaAzul]

I just had a recollection.  About 40 years ago I saw a photo of two men plying golf.  I sould easily see the golf ball on the green.  It helped to tell it was a golf ball to see the green and two golfers.  I could see the dimples or the name on the ball though.  The most amazing thing to me was the photo was taken with a MF camera on an airplane 65,000 feet above the golf course and traveling at 1200 mph.  That was 40 years ago and film.  How good can digital get?

That would be the SR-71 Blackbird with the Boeing Technical Objective Camera. The aircraft could carry two cameras either film or a 25Mpixel digital back (not tried operationally though - NB this is in the 1970s).

I would dispute that the cameras could resolve the dimples on a golf ball (or even the glof ball itself) from 80,000ft at 1,200mph. The cameras in the Blackbird were lateral (in that they took pictures sideways out of the aircraft) so atmospheric affects are going to be increased and motion blur at 1,200mph is going to require an unrealistic shutter speed to prevent motion blur of tiny objects.

From a commercial perspective I have dealt with aerial reconnaisance photography (usually 4-5 years behind in terms of quality compared with military applications, though ultimately the technology trickles down), and the fiction quite often exceeds reality by quite a considerable margin.

[Ray]

Reminds me a bit of the very first photo I sold about 6 years ago, taken with an old Minolta 35mm camera and cheap Tamron 70-300 zoom. With camera vertical on tripod, I took 13 shots of the city below from a tourist lookout, using Ektachrome 200 and lens at maximum zoom.

I scanned the slides on my Nikon LS-2000 but had to reduce the resolution to around 1500 dpi so Panavue's stitching software could handle the job which resulted in a final image of a mere 250MB.

I printed the image on my new Epson 1200 using roll paper and Corel's PhotoPaint which has a tiling facility. (Photoshop couldn't do the job). The print was 8ft long by 12.5". Sold it to the local council.

What excited me about the photo was the fact it had been taken using cheap consumer equipment yet was higher resolution than any single shot from any expensive professional camera that money could buy.

I intend going back to that spot on the right day, when conditions are perfect, and repeat the exercise using my 20D with 100-400 zoom plus 1.4x extender giving me an effective 35mm focal length close to 900mm. Instead of 13 shots in a row, it'll be something like 120 shots in 3 or 4 rows and I'll be able to read the logos on the aircraft landing in the distance.

Okay! Not as big as the Delft photo, but I think a more interesting scene   .

I'm waiting for computer technology to develop a bit so this job will be easier. I want stitching software to improve; 1GB sticks (or even 2GB sticks) of RAM to be faster, cheaper and more readily available; 64 bit operating systems to be standard and an upgrade to Photoshop that allows access to at least 8GB of memory.