16 Mpixels, Sony you've lost me !

The test is good for what it is. The problem is that people don't understand what the test is intended to show. This (mostly) isn't John's fault either. He did a pretty good job of explaining the point he was making. The problem is that people don't even try to understand the details of the experiment.
What is so difficult to comprehend that one sample is at 100% -- i.e., nothing resized in any way, and one sample is at 330% -- i.e., resized to over three times its actual size?

Have you ever resized anything to three times its original size? And then pixel-peeped at it? Rubbish, right?
There are different ways to compare images with different pixel counts and each has pros and cons. He could have kept one image smaller, or he could have downsampled the larger image.

No matter how you look at it though, it's hard to avoid the fact that the data support his hypothesis - at least in this case.

--
Ron Parr
Digital Photography FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/
 
I guess most people do understand the point in this test, they just find it completely silly and absolutely useless. It is merely comparing apples and oranges.
I'm not so sure. So far, I haven't seen anybody in this thread showing any symptoms of understanding.

The point of the experiment is to show that small pixels do not necessarily make less efficient use of sensor area than large ones. John used a very extreme example to make his point - he compared huge pixels from a DSLR with tiny ones from a compact camera.

Note that this result is in direct contradiction to the sophomoric claims that people around here so frequently make about pixel sizes. Every time a new camera is introduced with higher pixel density, the self-righteousness come out to lambaste the camera companies for sacrificing image quality for the sake of marketing higher MP numbers on the box.

The point is that smaller doesn't mean less efficient. Many people have tried to make this point in many ways over they years but people have a remarkable ability to ignore the facts here. One way to do it is to look at camera performance over time as pixel density has increased. There's a great example of this here:

http://pixinfo.com/en/articles/canon_powershot_g_evolution/

Another is to show an extreme case where cramming many more pixels into the same amount of sensor area leads to better overall performance, not worse. That's what John did.

--
Ron Parr
Digital Photography FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/
 
just dont buy it, i mean i only buy a camera that serves my needs and meet my standards closest as possible.

--
All my Post Processing is done with Capture NX2

http://www.flickr.com/photos/marti58/
Exactly! Unfortunately Sony currently doesn't offer any P&S camera with a "reasonable" sensor. Fortunately there is still Canon with its S90/95/G11/G12, or Panasonic with LX5, or Olympus, etc. What is however very sad, is the fact that those Canon cameras abe build around the Sony sensors.
 
It's a popular, sophomoric generalization. Sometimes it's true; sometimes it isn't. It all depends upon the implementation.
I would like to see one, single, real-life example of a compact camera beating an APS-C equipped camera. Show me two photographs of the same thing where a compact will produce a better image than a camera with a larger sensor.
Why would you ask for such a thing? Nothing I wrote or anybody has written in this thread would encourage such an expectation.
Except for maybe this, said by TechOutsider-
There is no advantage to bigger pixels emoticon - surprise ... just as John's test (below) shows that the compact's sensor with very small pixels (left) outclasses a slice of the DSLR's sensor (right).
I don't know where you got the notion that smaller pixels are worse, but that's wrong. Wrong in theory and wrong in real-life.
I was simply asking for a real-life example of small pixels outperforming larger pixels, within the realm of the photography we do. I contradict the above quote made by TechOutsider and suggest that there is enough proof that in a digital camera of the type we normally use, a larger sensor, with larger pixels, would do the best job.

And that the statement made that smaller pixels are not worse lacks factual basis, again backed up by the millions of digital files available for review taken with cameras with small pixels, and with larger ones.
 
so everyones bashin sony for increasin the pixel count on the hx series and others and yet canons usin the sony sensors in their cameras

laffin
 
that there is enough proof that in a digital camera of the type we normally use, a larger sensor, with larger pixels, would do the best job.
[snip]
millions of digital files available for review taken with cameras with small pixels, and with larger ones.
And here lies the problem - you are treating two separate variables - sensor size and pixel size - as interchangeable. When you say "small pixels" you mean "small sensors". A smaller sensor is at a disadvantage compared to a large sensor not because of the size of its pixels but of the size of its (smaller) total area.

A bigger sensor will - equal technology provided - always perform better than a smaller because it receives more light (because of its larger surface). But that has nothing to do with bigger pixels.
 
I don't know where you got the notion that smaller pixels are worse, but that's wrong. Wrong in theory and wrong in real-life.
I was simply asking for a real-life example of small pixels outperforming larger pixels, within the realm of the photography we do.
There were over 1000 posts on this very subject in the Canon and Fuji forums about a month ago. Briefly, one of the samples used was the comparison between the Canon G10 with 14mp downsized and the G11/12 with 10mp's.

Nothing personal, but I have no intention of getting involved in yet another highly charged debate on this complex subject, particularly as I am no authority on this.

However, you'd be wrong to assume that there are no examples supporting the view that smaller high density pixels can beat larger one's covering the same sensor area . Just check back the John Sheeny threads on the Canon forum and Great bustard on Fuji.

Nick
 
so everyones bashin sony for increasin the pixel count on the hx series and others and yet canons usin the sony sensors in their cameras

laffin
Do not be mistaken. That Sony can make excellent sensor they proved by latest S and G Canon cameras. Unfortunately Sony is not able to use those exellent sensors in its own cameras.
 
Ok. This is an argument that isn't going to go away. But I posed this position some time back in another thread and no one responded.

I have this straight question...

If there is no definitive proof that increasing pixel density on a sensor makes for better images; Why bother increasing pixel density on a sensor in the first place?

Its just a simple question. And it doesn't demand a complex answer. Why do it if it doesn't make the image better?

Dave
 
Ok. This is an argument that isn't going to go away. But I posed this position some time back in another thread and no one responded.

I have this straight question...

If there is no definitive proof that increasing pixel density on a sensor makes for better images; Why bother increasing pixel density on a sensor in the first place?

Its just a simple question. And it doesn't demand a complex answer. Why do it if it doesn't make the image better?

Dave
Only Sony (and other sensor manufacturers) could give a reasonable answer to this question, if such answer even exists. From my working experience as a software designer, oversampling (and this is exactly what’s going on with such dense sensors) gives you more “freedom” in digital signal processing. Unfortunately, having more freedom in digital photography translates into adding more gizmos or some other useless features. But fine, if you do not like some features, you do not need to use them. Unfortunately, oversampling in digital photography causes deterioration of the signal quality due to lower signal/noise ratio of smaller pixels. For sure, those modern 16 Mp sensors are technologically superior to, for example a sensor in my H5. But in spite of the technological advance, the S/N factor of the 16 Mp sensor must be worse than that of 5 years old 7 Mp sensor. So, then the question is, what customers care more about. To have a simple camera delivering crisp pictures, or to have a camera with a lot of features but delivering not so crisp pictures. I am convinced that everyone participating in this (or other, similar) forum prefers great image quality and doesn’t care about features at all. But we are just a tiny fraction of digital cameras consumers. We are simply not important for the camera manufacturers. Fortunately, some manufacturers listen to us as well and from time to time they come with what the others would call “niche” cameras. I mean cameras like S90/G12, XL5 and few others. For us however only those niche cameras are real cameras. And currently Sony is not giving us anything what we would like.
 
Ok. This is an argument that isn't going to go away. But I posed this position some time back in another thread and no one responded.

I have this straight question...

If there is no definitive proof that increasing pixel density on a sensor makes for better images; Why bother increasing pixel density on a sensor in the first place?
If we strictly follow this logic we should stick to one pixel sensors since increasing the pixel count doesn't increase image quality.

But your assumption is wrong. Increasing the pixel count actually does increase image quality (by increasing the detail captured - the fact that manufacturer's obviously try their best to offset this gain by applying heavy noise reduction in their jpegs is another issue).

This has been linked elsewhere in this thread but it's worth mentioning again and again - the development of the Canon G-series' image quality over the year:
http://pixinfo.com/en/articles/canon_powershot_g_evolution/
 
And here lies the problem - you are treating two separate variables - sensor size and pixel size - as interchangeable. When you say "small pixels" you mean "small sensors". A smaller sensor is at a disadvantage compared to a large sensor not because of the size of its pixels but of the size of its (smaller) total area.
Sorry, but no. A large photodiode works better than a small one because the P/N junction noise remains a constant, but because larger photodiode captures more photons, the S/N ratio of the device works in favor of the user.
A bigger sensor will - equal technology provided - always perform better than a smaller because it receives more light (because of its larger surface). But that has nothing to do with bigger pixels.
Bigger sensors have bigger pixels. 8 million pixels on an APS-C chip will outperform 8 million pixels on a chip a fraction of the size. Both sensors have the same pixel count, but the larger chip has larger pixels, which work better.

Cheers
 
Ok. This is an argument that isn't going to go away. But I posed this position some time back in another thread and no one responded.

I have this straight question...

If there is no definitive proof that increasing pixel density on a sensor makes for better images; Why bother increasing pixel density on a sensor in the first place?
If we strictly follow this logic we should stick to one pixel sensors since increasing the pixel count doesn't increase image quality.

But your assumption is wrong. Increasing the pixel count actually does increase image quality (by increasing the detail captured - the fact that manufacturer's obviously try their best to offset this gain by applying heavy noise reduction in their jpegs is another issue).

This has been linked elsewhere in this thread but it's worth mentioning again and again - the development of the Canon G-series' image quality over the year:
http://pixinfo.com/en/articles/canon_powershot_g_evolution/
This is not exactly true. To squeeze out the maximum (resolution) from the lens you need to have only a certain sensor density. For sure that density is well below 16 Mp. Increasing further sensor density will not contribute to the picture quality/clarity at all.
 
Bigger sensors have bigger pixels. 8 million pixels on an APS-C chip will outperform 8 million pixels on a chip a fraction of the size. Both sensors have the same pixel count, but the larger chip has larger pixels, which work better.
No, the larger sensor is larger, which works better. For an identical image and exposure a sensor eight times bigger receives eight times more light (signal), of course it will outperform the smaller sensor.

Again, if small pixels are so detrimental to image quality how do you explain the outcome of John Sheehy's experiment (which controls for the effect of sensor size)?

http://forums.dpreview.com/forums/read.asp?forum=1018&message=37472402
 
This is not exactly true. To squeeze out the maximum (resolution) from the lens

you need to have only a certain sensor density. For sure that density is well below > 16 Mp. Increasing further sensor density will not contribute to the picture
quality/clarity at all.
Which lens are we talking about here? After all different lenses have different resolving powers. And oversampling a weaker lens will yield better results than sampling at a lower rate.
 
so everyones bashin sony for increasin the pixel count on the hx series and others and yet canons usin the sony sensors in their cameras

laffin
its more then just a sensor that comes to play .....in camera processing comes to mind.

combining more MP's.. more zoom range.. more features ect .....is leading to much more compromises, i simply dont believe that all negatives of such construtions can be compensated for.

sure we allway's have endless debates about topics like these, thousands of theories and such, some make sense some dont.

the question was raised why MGF keep making more MP camera's well i see it as just a sales argument.

if i compare a 12.1 APS c vs a 12.1 fullframe i clearly know what wins in my book.

i never have seen prove that a smaller sensor is equal or better then a bigger sensor.

people keep claiming sensors are getting better and i doubt that, i think the surrounding processing is getting better.

same you can see in how powerfull PP tools have become compared to a few years back, no matter if i process jpeg's or raws ... i can do more and refine better then before.

the electric darkroom is amazing.

--
All my Post Processing is done with Capture NX2

http://www.flickr.com/photos/marti58/
 
I swear, some people just like to argue.
You can set the number of MP recorded in camera - e.g. my camera has the option of 12 MP, 8 MP, 5 MP, 3 MP, and 0.3 MP ;).
No you can't. The sensor always record 16MP. All you pick how much downsampling the camera software does.
Do you know what the word recorded means? A sensor cannot "record," any more than a microphone can "record." You need a recording device.

Record: "to set down or register in some permanent form"

Besides, the ops complaint was that it took up to much memory. Changing the setting addresses his concerns completely and fully. Does anyone really care is you come up with some pedantic quibble, even if it was right, which it wasn't?
 

Keyboard shortcuts

Back
Top