Loading Profile...
Twitter,
Facebook, or email.
Sensor Size Matters, Not Just Megapixels
Cool tool! I'd like to see the physical size of the sensor factored into the specs listed as well - sensors size is more important to me than megapixel count. I like that you guys tell the user that CMOS is generally better than CCD as far as sensors go. Now let's see the sensor size listed, with a corresponding numerical value for the size vs. megapixel value (typically measured in millions of pixels per square CM).
Official
Response
-
EMPLOYEE
I’m
excited
Hi Jason et Al,
We've just launched a significant update to the site, you can read about it here:
http://blog.snapsort.com/2010/06/snap...
The new comparisons only mention differences in our calculated true-resolution, not in megapixels, that way if two cameras have say 14MP and 12MP but both have tiny chips, then we won't call it out as an advantage, more details here:
http://snapsort.com/learn/sensor/true...
Also, the site shows and lets you search by pixel sizes.
Promoted
Response
-
3Showing the sensor size is not enough. We need to see the physical area per pixel to make a valuable comparison. So, we need to see:
Sensor Size in Square Microns / Megapixels
Then, take that information and use the camera min and max f-stop, multiplied by 1.35 to show the Airy disk diameter in microns for that camera. Now, for the best part:
By comparing the Airy disk size by the sensor size in square microns, one can get a feeling for how over-subscribed the sensor is. Many pocket cameras on the market today fit 4 whole pixels inside the minimum Airy disk diameter for their lens and physical sensor size! That means that the minimum resolvable dot of light the lens can provide (this is physics, not up for debate) falls onto 4 physical pixels! Why have the pixels be that small???
Anything over 6-7 megapixels in most pocket sized cameras is a waste. The physical size of the pixels is smaller than the minimum size of the disk of light focused onto the sensor by the lens.
-
EMPLOYEE
I’m
thankful for great users
Hey Jason, glad you like it, and thx for the great write up.
You're absolutely right, sensor size is important. We'll work on getting that in.
Currently it is included behind the scenes in our "Low Light Capability" rating. Also, we have a rather hidden feature, if you hover over the rating it will show you some of the details it used to make that rating.
We're working on making this more transparent!Comment -
-
3Showing the sensor size is not enough. We need to see the physical area per pixel to make a valuable comparison. So, we need to see:
Sensor Size in Square Microns / Megapixels
Then, take that information and use the camera min and max f-stop, multiplied by 1.35 to show the Airy disk diameter in microns for that camera. Now, for the best part:
By comparing the Airy disk size by the sensor size in square microns, one can get a feeling for how over-subscribed the sensor is. Many pocket cameras on the market today fit 4 whole pixels inside the minimum Airy disk diameter for their lens and physical sensor size! That means that the minimum resolvable dot of light the lens can provide (this is physics, not up for debate) falls onto 4 physical pixels! Why have the pixels be that small???
Anything over 6-7 megapixels in most pocket sized cameras is a waste. The physical size of the pixels is smaller than the minimum size of the disk of light focused onto the sensor by the lens. -
-
EMPLOYEE
I’m
thankful
@Shad: dude, you're hardcore :)
I read up a bit on Airy disk diameter just now on Wikipedia: http://en.wikipedia.org/wiki/Airy_disk
That looks really interesting. So basically you're saying that from the aperture size one can determine what the min physical area per pixel a camera can resolve is, and that many point and shoots are oversubscribed.
This the type of thing we'd love to include in our comparisons, as it sounds like it might scientifically debunk the megapixels arms race.
Thanks! -
-
EMPLOYEE
1I just saw this article on the same topic today:
http://petavoxel.wordpress.com/2010/0...
Anyone else have thoughts on this? I'm thinking we'll find a way to incorporate this into our comparisons.-
That would be great! It would help to package the raw data (yet still make it available) into a comprehensible concept. A label like "Physical Pixel Density", "Optical Efficiency", or similar.
Maybe a single value (scaled to between 1-5 stars) that is the pixel size in sq. microns divided by the min. Airy disk size plus pixel size in sq. microns divided by the max. Airy disk size. Larger values mean the sensor is less over-subscribed.
This is mostly relevant to fixed-lens cameras since DSLRs do not tend to suffer from sensor over-subscription, and depending upon the lens used one can easily go to a high enough f-stop to get non-optimal use of the comparatively large sensor. -
-
Another idea I had (because I'm a geek and cannot leave well enough alone) is something like "Effective Pixel Count" which is the megapixels divided by the ratio of the minimum Airy disk diameter to the physical pixel size. It sort of imagines what if the sensor only had one physical pixel per minimum Airy disk diameter? I think this is a good one because smaller pixels than the minimum Airy disk diameter are really just oversampling except that unlike audio oversampling, optical oversampling actually degrades quality.
-
-
-
-
-
EMPLOYEE
1I was thinking something similar. You'd probably want min(realMegapixels, effectiveMegapixels) I imagine? That way cameras that are oversubscribed get reduced down to their theoretical limit.
What f/stop would you use for the calculation? I gather you get higher effective pixels at higher f-stops, so maybe this calculation could be done at say f/5.6 assuming thats an average f-stop? Or maybe it should be done at the camera's widest f/stop for a worst case scenario, e.g. f/2.8?- view 1 more comment
-
-
I think I see what you're getting at with using a single, static f/stop value. If there is some f/stop that all P&S cameras seem to implement, that might be more easily understood than dynamically using each individual camera's minimum f/stop. In that case f/4 or f/5.6 would both be good options.
-
-
yes, I know what you mean about this concept not being well known. We'd have to find a clear and easy to understand way to present it, and provide details (or a link) for those who want to understand it in detail. This combined with something that estimates # of photos per disk might make a lower megapixel camera actually look better than a higher megapixel camera!
-
-
-
-
I’m
excited
Some idea of how 'good' the pixels are is definitely worthwhile and I think would make a big difference (certainly it should), but the question about f-stop suggests that the lens is an important factor. For DSLRs which lens would you use? The 'kit' zoom? At what focal length? Or the standard 50mm - then is that 50mm at full frame equivalent or 50mm outright, i.e. equivalent to say 75mm on a 1.5x factor APS-C sensor. This is all a bit of a nightmare! I wonder if any of these variables cancel out nicely but I'm not enough of a mathematician to work it out.-
Shad pointed out that sensor sizes on DSLRs are usually large enough that they are not running into this limit with any f-stop. So perhaps you could do this calculation for all cameras at some nominal f-stop, like f/5.6, and what you'd find out is that for most DSLRs they can achieve their rated megapixels, but most point and shoots can't and are limited by their sensor size.
-
-
For example, a 10 megapixel DX sized sensor would start to be oversubscribed around f/27. One can certainly dial in that f/stop or higher on a DSLR, but it usually requires switching to aperture-priority mode, i.e. the kind of person who does it probably knows what they're doing.
-
-
-
-
-
Since I'm the one that started this off, I thought I'd chime in again: I've read the airy disc article that's blazing it's way across the online photogosphere (yup, made that word up!), and while it's indeed quite fascinating, it's above the heads of most people (myself included to a degree). Trying to encapsulate this equation would be tricky, especially with the f/stop issue.
I think the basics of sensor size vs. megapixels for a simple pixel density value gives an easier to understand number that would contribute real value to what Snapsort is trying to offer. dpreview.com does this and I find it helpful. This value doesn't vary based on f/stop, which I think is important to consider. -
-
EMPLOYEE
1Hey Jason, great to hear from you again.
Maybe we do both? I do agree with you that the airy disc idea is a complex one, but I'm hoping we can find a way to convey it simply. I think that is an important area in general that Snapsort adds value in: not just providing you with the information (it *is* all there on the internet), but instead giving you the important information and making it useful/understandable.
Maybe something like this:
Camera ABC
Resolution
Effective: 7.3MP (based on sensor size, learn more)
Advertised: 12MP
Storage
Estimated photos per 1gb disk: 175
Camera XYZ
Resolution
Effective: 8.5MP (based on sensor size, learn more)
Advertised: 9MP
Storage
Estimated photos per 1gb disk: 200
I just made all that up, and presumably "learn more" would be a link to details on how that effective resolution was calculated. -
-
I think the problem with the Airy disk approach is that it depends on the f/stop value - and that changes depending on the lens on the camera. I bought a Panasonic GF-1 in December, and it has a pixel density of 5 million pixels per square CM. My Panasonic DMC-ZS3 has a pixel density of 45 million pixels per square CM. Both of those values are fixed, and not based on the lens on the GF-1 (which can be changed) or the fixed lens on the ZS3. And those two numbers tell me a lot...
Anyway, I think it's always better to present more info to consumers, so regardless of which approach you take, it will be an improvement. :-) -
-
Yeah I do see what you're saying.
We'll look at both! -
-
EMPLOYEE
I’m
excited
Hi Jason et Al,
We've just launched a significant update to the site, you can read about it here:
http://blog.snapsort.com/2010/06/snap...
The new comparisons only mention differences in our calculated true-resolution, not in megapixels, that way if two cameras have say 14MP and 12MP but both have tiny chips, then we won't call it out as an advantage, more details here:
http://snapsort.com/learn/sensor/true...
Also, the site shows and lets you search by pixel sizes. -
-
I like to know when a product is made under the assumption that I'm gullible. That tells me volumes about how good the product really is. I'm not stupid, but I'm not a sophisticated camera user. If the "oversubscribed" pixel information were presented to me in a spoonfed form, then I can instantly tell if the company is trying to sell me snake-oil. For example, the fact that Canon has REDUCED the megapixel counts in some of their cameras, like the G11, really impresses me. They're not trying to grab my money, they're trying to sell me a quality product with fair published specs. I suppose you could call it a "spec exaggeration metric" or something like that. Then, I could get rid of most of the Sony products in my searches :)
-
-
EMPLOYEE
I’m
confident
Hi qwasty,
"I like to know when a product is made under the assumption that I'm gullible. That tells me volumes about how good the product really is."
You lost me here I'm afraid :)
"If the "oversubscribed" pixel information were presented to me in a spoonfed form, then I can instantly tell if the company is trying to sell me snake-oil."
Great point. We don't currently indicate this in our comparisons, we'll consider adding this in the future.
You might be interested in:
- Point and shoot cameras with large pixels
- Snapsort's article on true resolution
- Snapsort's article on pixel size
- Point and shoot cameras with large pixels
-
-
Of course it depends on what is done with those extra pixels. If you have a 20MP Sensor and throw out most of the information to record a 1920x1080P video then that is wasteful.
If like Sony (3CMOS, EXMOR and now supersampling) or Fujifilm (EXR), you make use of the extra pixels to calculate a better image then, more pixels can be good and smaller ones (if they are black, hot or noisy) show up as smaller faults, thus improving image quality.
The other way is to have bigger pixels (to collect more light per pixel) but they must ALL be perfect pixels or you will got a black, white or noisy spot in the image.
.
See also:
Some more Math to digest - with pictures ;)
http://image-sensors-world.blogspot.c...
Better than the Human Eye (and 35mm Film):
http://image-sensors-world.blogspot.c...
Both Sony and this Article http://www.cambridgeincolour.com/tuto... claim that small pixel size is usually preferable as is a larger Sensor size. This does not suggest that we all need 50MP Sensors anymore than it suggests that 5MP Sensors will be very versatile (croppable, usable for video, better electronic image stabilization, etc.). -
