[asw66]

I've started wondering about this after reading yet another FF vs APS-C rant in the DPR forums. It seems to me that basic physics and optics should be able to tell us that there is a range of sensor sizes that are "just right" for general use.

For instance:

• I think we can take for granted that sensors like those in cell phones will never be optimal, unless there is some dramatic technical breakthrough
• At the other extreme, I think it's safe to say that we would never want a camera with an 8x10 inch sensor: the physical size of it would be prohibitive

A more sophisticated argument I've heard is that FF cameras have a really shallow DOF at wide apertures, and that FF lenses are inevitably heavier than their APS equivalents. If this is so, then I would think that this would be even more problematic with medium format.

Putting this broader question another way, bigger is sometimes better but not always. So what are the limiting factors on the size of a camera sensor for general use?

[Cosmo]

8x10 inches.

[flyer]

The optimal sensor size is always dependant upon the intended use of the camera. It would be stupid to use a MF size sensor with 65mb. to print postage stamps. So, to answer your question, there is no optimum size.

[RioRico]

The best sensor, in the best camera, with the best lens, is the one you can't quite afford to buy. If you can afford it, it's not good enough. Ten years ago, that was a 1.1mpx chip. Today it's a 30mpx goodie. In a couple years, 100mpx will be just beyond reach. Et [expletive deleted] cetera.

There are exceptions and solutions. Affordable 70+mpx sensors were available years ago, and still are -- one need merely adapt a flatbed scanner, add the right lens. It can be done for a few bucks. Another solution, suggested to me when I first started inquiring into digital photography, is to buy a MF film cam and a good neg scanner. The cash saved on multimegapickle electronics (soon outmoded) can buy a lot of film and processing. Another solution is to stitch-up shots from low-megapickle cams. I can turn my 1.1mpx Sony P20 into a 24mpx monster just by taking a 6x4 array of shots and applying AutoStitch. Would a P645D or K7 or K30D be easier to use? Sure. Money buys convenience, not creativity.

---

Last edited by RioRico; 05-24-2009 at 06:49 PM.

[bymy141]

Yves gave the right answer.
It mainly depends upon the purpose of the camera.

If you need to send the camera to mars, the moon or if you need to print 30 x 90 feet prints, you need something different.
I travel a lot, if I didn't had any investments in Pentax glass and the Oly sensors had better DR, I'd probably switch to Olympus. Weigth is more important to me than resolution.
FF means less length with long lenses, much more weigth, more expensive glass.
I know a girl shooting product ad pictures with a Hasselblad camera.
They use 4x6" sensors I believe......
So, it is relative. There is no one truth here.

- Bert

[newarts]

It isn't a bad question because it forces us to think about what we value.

It isn't answerable without assigning a measure of value to each parameter so from a practical standpoint can't be answered exactly.

But just thinking about it offers insights. I'm pretty happy with the DOF range I can get with an APS-C sensor, but would like higher ISOs at the same resolution. Physics tells me this will require a bigger sensor & associated costs (unless some much smarter software is developed.)

For now, APS-C seems a pretty good balance for my purposes.

Iowa Dave

[elkarrde]

Money buys convenience, not creativity.

that is really well said! i couldn't agree more.

[simons-photography]

it all depends sensor size alters the perceived focal length and also the perspective of the lens, my old bridge camera had a 6.2mm lens to give the angle of view of 28 mm my pentax 18mm give the same agle of view, clearly the perspective of the two lenses is not the same, smaller lens: greater DOF the combinations are mind blowing if you think of it, the "standard" on medium format is 200mm i beleive but this will compress perspective by 4 times compatred to the "same" 50mm lend on a 35mm format.

[simons-photography]

The best sensor, in the best camera, with the best lens, is the one you can't quite afford to buy. If you can afford it, it's not good enough. Ten years ago, that was a 1.1mpx chip. Today it's a 30mpx goodie. In a couple years, 100mpx will be just beyond reach. Et [expletive deleted] cetera.

There are exceptions and solutions. Affordable 70+mpx sensors were available years ago, and still are -- one need merely adapt a flatbed scanner, add the right lens. It can be done for a few bucks. Another solution, suggested to me when I first started inquiring into digital photography, is to buy a MF film cam and a good neg scanner. The cash saved on multimegapickle electronics (soon outmoded) can buy a lot of film and processing. Another solution is to stitch-up shots from low-megapickle cams. I can turn my 1.1mpx Sony P20 into a 24mpx monster just by taking a 6x4 array of shots and applying AutoStitch. Would a P645D or K7 or K30D be easier to use? Sure. Money buys convenience, not creativity.

very very true, I've often captured 50+ Mpx images as a standard with my fuji bridge camera and K10D, if you are prepared to spend time and effort you can do things most people would not trouble to do. of course it won't work on moving subjects, but then what about 2+ cameras set up to be syncronised ? oops thing I'm going a bit far now...

[DAZ]

To optimize is to make perfect, but to do this you need to state what the goal is, as others have stated. This implies making compromises (cost is compromise) and weighing them off against your goal. If you change the goal, the compromise or how you weigh them you will have a different optimize point. So what I think is good is not what someone else may think is a good point.

At the risk of getting called out for be to general here is the way I look at it. To go from APS-C to FF you get (with all other things the same) about 1 stop more or less. 1 stop more DR with FF. 1 stop less noise over APS-C. 1 stop less DOF with FF. This is the general part. What this means is you get about the same noise at lets say ISO 1600 on FF as you get with ISO 800 on APS-C but at some thing low like ISO 100 the noise for both is so low it is hard to tell them apart. For DR, if you had a DR of lets say 10 stops on APS-C you would get 11 stops on FF. For the DOF you liked on FF you would need to go 1 stop less for less DOF or 1 stop more for more DOF for APS-C. These are only approximant but you see the point. If you need or are willing to pay the about 2-3 times more fro FF over APS-C then by all means get FF. But remember cost is a factor and only you can determent if the cost to benefit works for you.

Now for me I don’t think it is worth 2-3 times for 1 stop but I can see the 4-5 times for the 2+ stops from APS-C to MF. Only because if you are going to spend that much more (and you are probably getting paid for it so why not just rent it) then get a big enough jump so that the change is in your face and not subtle.

DAZ

[WerTicus]

....5..........

[RioRico]

But remember cost is a factor and only you can determent if the cost to benefit works for you.

Ah yes, cost-benefit analysis. A pro needs to spend money to make money. (Or they get an employer or client to spend THEIR money.) An amateur takes their time, or blows their budget, depending on their desires and means. 'Tis ever so.

EDIT:

To optimize is to make perfect, but to do this you need to state what the goal is, as others have stated. This implies making compromises (cost is compromise) and weighing them off against your goal. If you change the goal, the compromise or how you weigh them you will have a different optimize point.

In Game Theory and Linear Programming, this is called a MiniMax Solution - trying to reach a maximum gain with a minimum cost - but the gains and costs are always defined differently with each distinct problem (goal), and they can be manipulated with time. So we try to get the maximum gain NOW while delaying the costs until LATER, like using a credit card. When I software-engineered financial systems, a corporate paradigm was: Never Time To Do It Right, Always Time To Do It Again. Which meant, get the software system written fast so it can be announced and sold, then spend the next couple years fixing major bugs. How many of us play similar games with our lives, hobbies, careers?

[Digitalis]

take a look at this...

[RioRico]

take a look at this...

Great comparison. I'm too lazy to generate quadrangles, but here's a film frame comparison list:

FORMAT
135-FF ... 24x36= 864 mm2
135-HF ... 18x24= 432 (50% of FF ~= APS-C)
135-SQ ... 24x24= 576 (67%)
110 ...... 13x17= 221 (26%)
126 ...... 26x26= 676 (78%)
127-40 .. 40x40= 1600 (185%)
127-30 .. 40x30= 1200 (139%)
127-60 .. 40x60= 2400 (278%)
120-620:
645 ... 56x41.5= 2324 (269%)
66 ..... 56x56 = 3136 (363%)
67 ..... 56x70 = 3920 (454%)
69 ..... 56x84 = 4704 (544%)
612 .... 56x118= 6608 (765%)

We can see that 110 film is about 5 times larger than the biggest PNS sensor, and about the same size as Four Thirds. Yup, the P110 is what Olympus should aspire to. Heh heh. And film frames in my Ikonta 6x6 (pocket Hassy), and Kodak and Voigtlander 6x9s (pocket Rolleicords), dwarf all the digital sensors, even the Kodak KAF.

[Lowell Goudge]

this discussion seems to be like all the old sayings

chairman Mao "one people one uniform"
henry ford "you can have any color you want as long as its black"
Unknown "if you design something such that one size fits all, it does not fit any"

There is no universal ideal "standard"

if you consider all the trade off's, field of view, depth of field, cost, final print size, print viewing distance, required resolution, grain/noise (or lack of ) until you define exactly what you want to do with a print, there is no "optimum". even then, optimum for one situation is not optimum for another.

Let's look at it another way. Until we get noiseless, 20 stop dynamic range and a function called "mindread.exe" so that we can print in infinite detail what we think we see we will never have optimum.

We are already far far ahead of what we could do with film. The real question is, what should we consider as good enough.