Printable View
I have several sensor questions.
Why are sensors not round like the image from the lens.
Let us capture all the data and make our own decision on how to crop.
Do larger pixels:
make for better sharpness? Larger the better?
Why do we not put life expectancy on lenses, does the aperture mechanism not wear out?
Quote:
Originally Posted by iND
Cost, probably. Sensors are cut from larger wafers (~200 APS-C sensors or ~20 FF sensors from a single wafer). Imagine cutting round cookies from a disc of cookie dough - there's a lot of waste with the in-between spaces, but unlike dough, you can't just ball up the rest of the silicon wafer and roll it out again...
Quote:
Originally Posted by iND
Probably, but so can the IS system, the AF motor, etc. Lots of moving parts, each moving part has a different failure rate, which one will wear out first? The shutte mechanism (which includes the mirror assembly, etc.), on the other hand, is the primary moving part within the camera body - one number that's easy to count.
Quote:
Originally Posted by iND
Just as Neuro said, silcon wafers for sensors are very expensive and cutting round sensors would be very wastefull. Good idea though.
Quote:
Originally Posted by iND
Yes and no, larger pixels (assuming same sesor size) means you can't zoom in as much and don't see as many defects. Noise and lens softness, ect. But you are replacing it with something worse, lack of resolution. The defectswere always there, you just couldn'tsee them or were not as obvious before.
Noise isn't affected by pixels size, it's affected by sensor size. A larger sensor gathers more light, pixels just cut up the light into smaller pieces. With VERY slight losses between pixels with todays gapless micro lens sensor technology.
John.
Quote:
Originally Posted by iND
First, the image from current lenses is *not* always round. The more expensive the lens, the more likely it is to have rectangular baffles in and around lens elements. Many use petal or rectangular lens hoods. Those restrict the image from circular into a rectangular shape. The benefit is that it reduces veiling flare, ghosting, and increases contrast.
Second, even if the image was round, making the sensor round would cause stratospheric costs increases. It would make an $800 camera into an $8000 camera. Almost *double* the sensor area (1.7X) is required, not to mention the extra wasted space from unused wafer. I'm told it's a hugely difficult, non-standard semiconductor process as well; so much so that it would actually be cheaper to just make a large square sensor and only use the center circular portion.
So what is the optimal sensor and lens shape? It is the one that you use for most of your prints. If most of your prints are round, then a round lens with round sensor would give you the best result. Using a 3:2 sensor and 3:2 lens to make a round print is suboptimal. Similarly, if 8x10 is your most common print ratio, then the optimal lens and sensor ratio would be 8:10 (i.e. 4:5).
Which would be better for making 8x10: a round sensor and round lens? Or a 3:2 sensor and 3:2 lens? With a 3:2 lens, there is 17% extra unnecessary light causing veiling flare and reducing contrast, and you have to crop out 17% of the unused sensor portion. With a round sensor, you have to crop out 56% of the sensor, the lens has 56% more flare and reduced contrast. So 3:2 is much better than circular for 8x10. In fact, for any common print ratio, 3:2 has less flare, more contrast, and much lower sensor cost.
Quote:
Originally Posted by iND
No. Larger *sensors* tend to make for better sharpness, as a general rule. Given two sensors of equal size, larger pixels tend to make for *worse* sharpness.
Quote:
Originally Posted by iND
I'd say "smaller the better". Check out this thread for more information about pixel size:
community.the-digital-picture.com/.../1055.aspx
And here is another one I wrote (on a different forum) that is related:
photography-on-the.net/.../showthread.php
Quote:
Originally Posted by iND
By "we", do you mean lens manufacturers? Or do you mean "why don't we photographers maintain a chart of the long-term failure rates of lenses"? Either way, I don't know the answer, sorry. The closest thing I know of is the excellent statistical information posted by lensrentals.com every year:
www.lensrentals.com/.../lens-repair-data-4-0
It's awesome. I just wish more places (and the manufacturers themselves) would do that.
Quote:
Originally Posted by iND
In my experience, most lenses will be lost to bumps, bruises, mold, and other types of damage before they have any mechanical failures. But if you take good care of a lens, I'd want it to last at least 10 years before the I.S., autofocus motor, or the aperture mechanism starts to fail. Anything less than that and I would consider it poor quality manufacture. I don't know if today's advanced lens coatings lose effectiveness over the long term or not, but I wouldn't think so. If you have a completely manual lens (including the aperture), it can easily last over a century. I've seen lenses from the 50's that are as good as new.
Good point, some of my lenses are 20+ years old and work perfectly. Figure this, how long will it take to have lubricated pieces of metal that have very little force against each other rub each other such as in a aperture. Almost indefinetly.
John.
Quote:
Originally Posted by Daniel Browning
yeah, ditto that. i've got Takumar lenses my mum bought in the '60s when she went to Japan for a Uni trip. the 50/1.8 is sharper than my niftyfifty, focus ring is smooth, aperture works good. It all depends on how well you take care of them. One thing with the Takumars is that their coating had some sort of UV-reactive coating that makes the lens look yellow after a few years in the dark, apparently a few days in sunlight is enough to fix them.
.
But then, they don't make em like they used to. Apertures, IS, AF motors will wear out eventually. But how many would 'wear out' and how many would 'wear out of spec' (but still otherwise work)?
And think about the upgradeability. if i'd had a 50/1.8 v1 since EF was launched 30ish years ago, and it broke (like the motor fused over and focussing stuck), i wouldn't bother repairing it, i'd probably spend less on a new niftyfifty v2. The cost of repairing all but the most valuable lenses would make it cheaper to buy a new one, once you take labour into account (i've spent a few years in the electronics repair industry).
.
As for the round sensor, think about the days of film when it all started. how do you make a round negative in a strip without wasting any film? maybe alternating triangles could fit, hexagons could fit but still waste a bit. And most frames are squared (ok, we could make round frames more easily in bulk nowadays), go back hundreds of years to paintings and most were square/rectangle because round corners were harder for carpenters to make.
.
But then, nowadays it wouldn't be so hard, just expensive. I seem to recall reading about a camera with a sensor, which was basically a 3:2 overlayed over a 2:3 portrait and 1:1 (so it kinda looked like a fat + sign. but then, that would have to be cut down from a square sensor, so why not just put in the square sensor and disable pixels on the edge outside the image circle...
ok, fourthirds-user.com/.../the_biggest_four_thirds_sensor_yet.php is something along the lines i was thinking. simply extend the idea to 1:1 and portrait. but still, you'll have to make a lot of pixels outside the circle you're not going to use --> too expensive...
.
Also,Quote:
Originally Posted by Neuroanatomist
Yeah, you kind of can. but they just throw the rest of the wafer back into the crystal-growing vat, no sense in wasting all that purified silicon...