POLL - Moving sensors

[Damien_Demolder]

I suppose it’s because smaller camera manufacturers have to work a little harder to gain attention that it seems to be them that comes up with all the new features these days. As Angela says in this week's test of the Canon EOS 500D and the Olympus E-30, it is companies such as Olympus that have brought us Live View, sensor cleaning and flip-out screens in DSLRs.

In this week's magazine we see more new developments, but from Pentax. We’ve seen in-camera levels and moving sensors before, but this time the two functions are combined to produce a camera that effectively levels itself. It seems revolutionary, and it will certainly be extremely useful, but it’s hardly the first self-levelling device invented.

It’s been six years since the invention of the shiftable camera sensor and the in-camera orientation sensor was patented in 1995, but only now have they come together. When Konica Minolta first showed me the Dimage A1 I suggested to they might use their moving sensor to create tilt and shift movements in normal lenses. Combining APS-C sensors with full frame lenses allows plenty of covering circle, and cuts out the excessive cost of specialist optics. It would be useful not only for architecture but as a creative tool for portraiture, landscape and still life.

To vote in this week's poll, 'Would you like to see more moving sensor features in DSLR cameras?' head to the homepage.

Thanks for taking part.

Damien

[GeoffR]

Absolutely not. I want the sensor firmly nailed to the body, that may mean I get camera shake and sloping horizons but the errors will be mine. If the sensor can be moved by the on board electronics you can bet that sooner or later it will move all by its self.

[huwevans]

Absolutely not. I want the sensor firmly nailed to the body,

WHS, with knobs on. Nailed, bolted down, tied to a humungous great block of granite, and the foundations sunk 200' down into the bedrock. :-)

Apologies in advance for this next bit, but it has to come out.

<u>WARNING: MAY CONTAIN RANTS</u>

The thing about shift is it's not really necessary in the digital age - the small amount of software correction needed for converging parallel lines is really neither here nor there from a quality point of view unless extreme corrections are necessary - in which case (a) the lens mounts aren't wide enough to incorporate the amounts of shift that would be needed to do it optically, (b) the large image circles needed mean that the photographer will be losing large amounts of optical quality anyway, off-setting gains made by foregoing the resampling done by software correction, and (c) reintroducing all those bad corner issues - vignetting, fringing due to extreme incidence angles on the microlenses and anti-aliasing filter, smearing due to photons being captured in wells adjacent to the ones they should be in, and so on.


And then there's tilt - absolutely essential for large formats, where depth of field is practically non-existent, but almost totally over the top for small formats. Even for 'full-frame', which now seems to have established itself as the largest format in common use, and to have replaced, say, medium format, as the one that enthusiasts commonly aspire to, it's really pushing it.

The buzz on internet forums these days seems to imply, sometimes, that no self-respecting manufacturer's lens inventory would be complete without a swathe of T&S lenses covering ultra-wide right through to moderate telephoto. The usual gleeful 'new toy' threads appear whenever someone gets hold of their new baby, and what do we get from them? Bl**dy mock 'toy-town' shots! That's all these small format T&S lenses have done for photography in the digital era. Tilts used to be all about making best use of the little DoF that was available on large format - now the formats are comparitively tiny, DoF is much greater by default, and some other justification needs to be found for an expensive and highly specialized piece of kit - enter the 'reverse tilt', where the goal is to get everything apart from the subject as out of focus as possible, producing the toy-town effect. And if I see one more of them I might just start to develop a fondness for That Bl**dy Mountain, and bluebell wood shots - I'm that p*ssed of with them!

That's it. Thank-you. All over now.

[sey]

Absolutely not. I want the sensor firmly nailed to the body,

WHS, with knobs on. Nailed, bolted down, tied to a humungous great block of granite, and the foundations sunk 200' down into the bedrock. :-)

Apologies in advance for this next bit, but it has to come out.

<u>WARNING: MAY CONTAIN RANTS</u>

The thing about shift is it's not really necessary in the digital age - the small amount of software correction needed for converging parallel lines is really neither here nor there from a quality point of view unless extreme corrections are necessary - in which case (a) the lens mounts aren't wide enough to incorporate the amounts of shift that would be needed to do it optically, (b) the large image circles needed mean that the photographer will be losing large amounts of optical quality anyway, off-setting gains made by foregoing the resampling done by software correction, and (c) reintroducing all those bad corner issues - vignetting, fringing due to extreme incidence angles on the microlenses and anti-aliasing filter, smearing due to photons being captured in wells adjacent to the ones they should be in, and so on.


And then there's tilt - absolutely essential for large formats, where depth of field is practically non-existent, but almost totally over the top for small formats. Even for 'full-frame', which now seems to have established itself as the largest format in common use, and to have replaced, say, medium format, as the one that enthusiasts commonly aspire to, it's really pushing it.

The buzz on internet forums these days seems to imply, sometimes, that no self-respecting manufacturer's lens inventory would be complete without a swathe of T&S lenses covering ultra-wide right through to moderate telephoto. The usual gleeful 'new toy' threads appear whenever someone gets hold of their new baby, and what do we get from them? Bl**dy mock 'toy-town' shots! That's all these small format T&S lenses have done for photography in the digital era. Tilts used to be all about making best use of the little DoF that was available on large format - now the formats are comparitively tiny, DoF is much greater by default, and some other justification needs to be found for an expensive and highly specialized piece of kit - enter the 'reverse tilt', where the goal is to get everything apart from the subject as out of focus as possible, producing the toy-town effect. And if I see one more of them I might just start to develop a fondness for That Bl**dy Mountain, and bluebell wood shots - I'm that p*ssed of with them!

That's it. Thank-you. All over now.

DITTO!!!

besides we want simpler cameras not more complicated with more things to go wrong and cockup a perfectly simple, straightforward and great shot.
I'm very tired of this "anything goes" so called progress to get one over on the competition and 'improve' sales.
Don't need photographers any longer, just need camera porters. Soon the cameras will be self-propelled and who will need humans at all!
I'm really starting to get p*ssed off with the mad attempt cater to and satiate/satisfy the no-talent gadget geeks at the expense of basic human skill and talent.
Just give us affordable, simple and straight forward high quality kit and let us get on with photography.

[nspur]

If the sensor can be moved by the on board electronics you can bet that sooner or later it will move all by its self.

They do. My late Ricoh R4, very early on and when in anti-shake mode always moved its sensor to the right so I not only got camera shake but also an incorrectly famed shot. Back it went to be fixed and a new one appeared that lasted for 2 years or so doing what it ought to until it did the same thing again.

[Learning]

IF VR goes wrong you change lenses if it matters. If an oscillating sensor goes wrong then you have to change to your backup camera. If you haven't got a back up then you are [****].

[RogerMac]

IF VR goes wrong you change lenses if it matters. If an oscillating sensor goes wrong then you have to change to your backup camera. If you haven't got a back up then you are [****].

Disagree, if in body IS goes wrong you can just switch it off. The worst case scenario is a very slightly missplaced sensor. If you are using live view then the screen will view will still show the actual final image

Roger

[Learning]

IF VR goes wrong you change lenses if it matters. If an oscillating sensor goes wrong then you have to change to your backup camera. If you haven't got a back up then you are [****].

Disagree, if in body IS goes wrong you can just switch it off. The worst case scenario is a very slightly missplaced sensor. If you are using live view then the screen will view will still show the actual final image.


Roger

I have to agree. And a modern camera is just another computer with a lens and sensor so I suppose switching it off and on again might help.

[GeoffR]

IF VR goes wrong you change lenses if it matters. If an oscillating sensor goes wrong then you have to change to your backup camera. If you haven't got a back up then you are [****].

Disagree, if in body IS goes wrong you can just switch it off. The worst case scenario is a very slightly missplaced sensor. If you are using live view then the screen will view will still show the actual final image

Roger

If only it were that simple!

I have lost track of the number of times I have come across electronic equipment that could only be switched off by physical removal of the power supply. Hardly a practical solution to a sensor translation system problem.

The more degrees of freedom the sensor has the more possibilities for it to adopt an unacceptable orientation.

Live view will indeed give you a view of what the sensor is seeing, as long as you don't have one of those clever cameras with a separate live view sensor. However that does rather presuppose that you are aware of the problem. If not you will carry on shooting, using the view finder, and the get home to card full of displaced images.

[huwevans]

Unfortunately, the practical reality of electronic goods means that faults in one area are not always so helpful as to just disable that particular part, whilst allowing everything else to go on working normally. They can also disable the entire device, giving the user no clue as to where the fault lies or how it might be got around, if that's even possible.

[Terrywoodenpic]

I am sure such things will sell... however they add greater obsolescence and potential for failure.

Now they are invented they will become the next must have.

But I am in no rush.

[beejaybee]

Soon the cameras will be self-propelled and who will need humans at all!

Just so long as it has an "arguing with PCSO" mode....

[DaveS]

Just weld the dratted sensor to the body. No sensor should move EVER. aargh!
No, no moving sensor for me. How many times do I have to say it!
Dave

[Roger_Provins]

I think it would be a great facility to be able to angle the sensor at will, not so much to correct verticals - although that would be good, but rather to place the plane of focus where it's required - particularly in macro work. Many professionals would be extremely happy with this form of tilt and shift.

As far as moving sensors in general is concerned I think any fear of failure is misplaced. They are not likely to fail any more than other parts of the camera. Don't most dSLRs, even those without body image stabilisation, move the sensor at least for the cleaning cycle?

The last camera I had, which had in-body image stabilisation plus sensor cleaning every time it was switched off - thus a much moving sensor, took over 35,000 shots in the time I had it - that's 1000 rolls of film. It suffered no problems at all and is still going strong. How many film cameras, in amateur hands, ever got through that number of films in their lifetime?

[GeoffR]

I remain unconvinced that trying to shoe horn all the movements of a large format monorail camera into a DX sized DSLR is a good idea. Just because it can be done doesn't necessarily mean it has to be done.

[RogerMac]

Unfortunately, the practical reality of electronic goods means that faults in one area are not always so helpful as to just disable that particular part, whilst allowing everything else to go on working normally. They can also disable the entire device, giving the user no clue as to where the fault lies or how it might be got around, if that's even possible.

I have had a camera with in body IS for 2 years and the IS has been totally reliable (I experienced two faults - the AF went wonky after I dropped it and I lost my wireless broadband connection whilst updating firmware) I know a sample of one is nowhere near statistically significant but neither the AP forums nor the three four third forums (fora?) that I frequent are full of complaints about unreliability. Two years is sufficiently long for valid field evidence of reliability (or lack of it) to become statistically significant. Perhaps we should ask Damien to find out the true facts regarding IS reliability.

Roger

[GeoffR]

My I refer you to the esteemed Mr Murphy whose law reads:

"If some thing can go wrong, it will."


Thank you.

[RogerMac]

My I refer you to the esteemed Mr Murphy whose law reads:

"If some thing can go wrong, it will."


Thank you.

The preliminary evidence seems to be that this particular sub-system is totally reliable. I just think that somebody (perhaps AP) should collect the evidence to prove or disprove this

Edit: That's not to say that it is the best possible system regardless of cost. I happen to think that in lens IS is superior for long lenses.


Roger

[Fire_Monkey]

It's probably competely reliable, but isn't this another example of the camera doing the photographers work for him?
The more gadgets a camera has, the more chance of somthing going wrong! Let's keep them simple!

[Terrywoodenpic]

I think it would be a great facility to be able to angle the sensor at will, not so much to correct verticals - although that would be good, but rather to place the plane of focus where it's required - particularly in macro work. Many professionals would be extremely happy with this form of tilt and shift.

As far as moving sensors in general is concerned I think any fear of failure is misplaced. They are not likely to fail any more than other parts of the camera. Don't most dSLRs, even those without body image stabilisation, move the sensor at least for the cleaning cycle?

The last camera I had, which had in-body image stabilisation plus sensor cleaning every time it was switched off - thus a much moving sensor, took over 35,000 shots in the time I had it - that's 1000 rolls of film. It suffered no problems at all and is still going strong. How many film cameras, in amateur hands, ever got through that number of films in their lifetime?

Tilting the back (Sensor) could indeed shift the plane of focus and might have some advantages, However when you do this you reduce the depth of field at 90 degrees to it. So its usefulness can be limited.
It would be pretty useless for converging verticals as the amount of rise and fall would be very limited.