[jebjeb]

Just read this very interesting article from Luminous Landscape.

It explains lots about equivilant focal lengths, apertures etc.

What is most interesting is something I have always believed in but never understood the science behind; the fact that high pixel counts are pretty much wasted on many digital cameras, particularly small point and shoots.

The article explains about the relationship between aperture, depth of field, focal length, sensor size and sensor resolution. Through a series of calculations, one is able to calculate the optimum maximum f-stop before a camera can not deliver its full resolution.

Here are some of their calculations.

Canon 5D f/8.6
Canon 1Ds f/9.3
Canon 1Ds Mark II f/7.6
Nikon D2X f/5.8
Canon 20D f/6.76
Canon G7 f/2.06

The first thing I find interesting about this is that those of us with digital SLRs (which have larger sensors) have probably thought that the best way to get the sharpest picture is to reduce the aperture (use a higher f-stop). This then increases the DOF keeping more in focus. I know that I have always thought this but have also remembered not to go too low. This article doesn't dispute the DOF point. What it does say though is that by going beyond the calculated f-stop, one may gain higher DOF but the effective resolution drops significantly. Put it this way, if one had a Canon EOS 1Ds MkII which is 16 megapixels, if you took a picture with a "perfect" lense at f22, the camera would be the equivalent of 2 megapixels! However, if the same shot was taken at f8 then one would have the full 16mp.

Jeepers!

Now remember, all these calculations are based on using a perfect lense with no aberrations or other faults. This means that in the real world, things will be even worse.

There is a much bigger problem in point-and-shoots. I have always maintained that the lenses in these can not really support the huge pixel counts that are about these days, making them a bit of a waste and being inefficient with memory space. This is backed up by the calculations for a fairly top end P&S which most people would consider has a better than average lense; the Canon G7. The optimum maximum f-stop to achieve the full 10 mp is f/2.06. The lense only has a minimum aperture of f/2.8! This means that all available apertures are not delivering the full resolution. Even worse when one zooms in and the lense is slower. Imagine what it would be like with a smaller pocket size camera with a much slower lense.

So I guess this all backs up that manufacters have to stop it with this silly pixel race and start working on better lenses, higher sensitivity with less noise and other features.

Also, I have been a firm believer in full-frame is better. However, this article does suggest that this may not be the case for many things. Hmmm.

[Dark Helmet]

I stopped caring about MP's once they hit 5 as I even set 90% of my picts to be 3MP.
It is stupid seeing low end point and shoots with 7MP's as with it set at 7 at default people wonder why the camera holds so few pics, takes long to transfer, how they can't directly email a few of them and why it fills so much room on the camera.

They really should concentrate on better colour accuracy, low light levels and image stabilization over megapixels.

Naturally for pro photographers MP's still matter.

[Goldfinger]

The only thing that difraction does is make your image less sharp. There is a certain max point of sharpness. If you decrease your aperture beyond that point your DOF will be larger but your image will be less sharp. This is nothing new and the problem of difraction exists with film as well.

The claim that you turn your camera into a 2mp camera is bogus.

[Eug]

Issues of diffraction existed with film too. You don't see anyone from the film days saying film had too much resolution. Thus, I consider that argument very odd.

I think the MUCH bigger issue is that point-and-shoots give us resolution at the expense of noise. At the proper aperture, many can give us tons of detail if there is adequate light, but unfortunately, the sensors are extremely noisy in low light.

The number one reason I want from a point-and-shoot digital to a digital SLR was flexibility, but the number 2 reason was because of low light performance. The vast majority of point-and-shoots suck in low light.

[Dark Helmet]

Speed is also still an issue on most camera with the shutter delay.

[Eug]

Originally Posted by Dark Helmet 

Speed is also still an issue on most camera with the shutter delay.

Many newer higher end point-and-shoot cameras don't have a huge shutter delay (apparently).

[Dark Helmet]

Originally Posted by Eug 

Many newer higher end point-and-shoot cameras don't have a huge shutter delay (apparently).

Ya but many many of them still do. Even your camera phone (admittingly still a camera phone) takes forever to take a pic once you hit the button.

[centerchannel68]

Neat article. But yeah, I stopped caring long ago, once I saw the pictures that slick looking sony with the HUGE lcd and carl ziess lense.... I loved that camera... thought it was perfect...5MP..huge screen..then I saw the pictures. HORRIBLE color. Everything looked 'dead' and flat. My 4 mp canon takes much, MUCH better pictures.

[cszar2001]

The only thing that matters when it comes to photography is the person behind the camera.
And that is not going to change - ever.

[pathogen]

Apes take awesome pictures, so it's not the person. It's the viewer: Beauty = Eye of the beholder.

[Dark Helmet]

Originally Posted by cszar2001 

The only thing that matters when it comes to photography is the person behind the camera.
And that is not going to change - ever.

Even if they are using this?

[OreoCookie]

This is actually quite old news. I remember a comparative test of the D2X (12 MP) and Canon's full-frame counterpart (16 MP). Both of them were tested with the 105 mm fixed-focal length macro lenses and it was shown that the D2X's resolution was actually higher at low apertures (I think it was even at f5.6).

Similarly, it was easily shown that 10 MP point-and-shoot cameras usually don't have a higher resolution than their predecessors, especially at high ISO the noise filter washes out any detail the picture might have had.

[Hawkeye_a]

H3D

Im not a photographer by any means, im more of a technically oriented person, so i could be wrong....

the thing that matters when talking MPs, is actually the ratio of MPs to censor size. (ie the number of pixels you plan on cramming onto the censor). the smaller the ratio, the lesser the noise, the better the picture.

If you have a really tiny censor, it might be why your 3MP pictures look better than your 5MP pictures. had your camera had a bigger censor, 5MP would definately look better (more detail/higher resolution).

I have a 10MP cmera with a 1/1.65" CCD (compared to a G7's 10MP with a 1/1.8" CCD (which is a smaller censor))

post processing and noise reduction (or any sort of processing) doesnt always guarantee a better picture, and any non-full-frame camera will produce some level of noise in low light conditions.

If you are looking for a good camera, balance is the key...
-you want a nice piece of glass in the front
-a reasonably sized censor
-a decent number of MPs for the resolution you are looking for
-decent processing (Canons are the best imo).
-portability/controls/etc...depending on what you want.

Im not a pro photographer, so i could be way off here..... feedback would be appreciated.

Cheers

[GSixZero]

Originally Posted by Dark Helmet 

Even if they are using this?

I was actually trying to set up a Quicktake 100 in OS X this week, but was having issues. Has anyone been able to do this?

Thanks.

[Dark Helmet]

Originally Posted by GSixZero 

I was actually trying to set up a Quicktake 100 in OS X this week, but was having issues. Has anyone been able to do this?

Thanks.

Why in the world...

[GSixZero]

Originally Posted by Dark Helmet 

Why in the world...

Why not? It still works!

[jebjeb]

Originally Posted by Hawkeye_a 

If you are looking for a good camera, balance is the key...
-you want a nice piece of glass in the front
-a reasonably sized censor
-a decent number of MPs for the resolution you are looking for
-decent processing (Canons are the best imo).
-portability/controls/etc...depending on what you want.

Im not a pro photographer, so i could be way off here..... feedback would be appreciated.

Cheers

I agree with most of this, Hawkeye. I have held off buying a new body (currently have an EOS 10D) as I wanted to build up decent glass. Have a bunch of L f/4 stuff and others so have that covered now.
Big fan of big sensors and would love to go full frame but don't mind the crop frames either.
I am not too worried about processing as I shoot everything in RAW. However, if and when I jump up to 12 mp + then RAW will certainly be sucking up a whole lot more memory

[Dark Helmet]

Originally Posted by GSixZero 

Why not? It still works!

The worst camera phone in the world must take better pics. And work on a Mac.

[Jim_MDP]

Originally Posted by Dark Helmet 

The worst camera phone in the world must take better pics. And work on a Mac.

Yeah, but you still don't belong in this discussion.

[Mastrap]

Originally Posted by Eug 

The number one reason I want from a point-and-shoot digital to a digital SLR was flexibility, but the number 2 reason was because of low light performance. The vast majority of point-and-shoots suck in low light.

I bought a Fuji F30 specifically for it's low light capabilities. It easily matches, and sometimes betters, the performance of a lower level SLR. Fuji has decided to concentrate on low light performance rather than upping the pixels in this camera and it shows.

To give an example, this images was taken in very low light at Mother's Dumplings. It's using ISO 1600 - the highest setting is 3200 - and while there's some softness in it it is still a perfectly usable snapshot.



Try and search on flickr for this little gem of a camera and you'll see some wonderful night photography.

[Eug]

^^^ It's nearly impossible to assess low light performance in a 500 pixel wide JPEG.

However, I do not doubt your claims of it having good low light performance. That's why I didn't say all point-and-shoots have this problem.

Also, that camera didn't come out until 2006, long after I had switched to a dSLR. Since then dSLRs have improved further, and still give a noticeable advantage in low light performance over most point-and-shoots. Anyways, your F30 does sound like a great little camera, but obviously it ain't perfect either.

And yeah, Mother's Dumplings is great. I feel sorry for the restaurant next door though. Their place isn't as nice, and they weren't written up in Toronto Life, so they have next to no business. However, the food there is also excellent, and I much prefer one of their hot sauces, which is a mixture of garlic and chili. It's also a little bit cheaper than Mother's Dumplings. Both restaurants are much superior to the Dumpling House on Spadina. (Dumpling House used to be good, but they changed owners and everything went downhill.)

[Mastrap]

Originally Posted by Eug 

Mother's Dumplings is great.

Glad that you made it, and liked it. I haven't tried next door, but I just might now.


Yeah, the F30 isn't perfect, but what it does it does well. I love my, by now ancient, Olympus SLR but the F30 is a camera I can take with me at all times.

[jebjeb]

I think what Mastrap has alluded to there is right; if one is happy with a product just becuse we learn info such as this doesn't mean that said product is now no good.

Info like this is very interesting and will probably influence how I use my gear but it does not make me instantly dislike what I have been happy with.

[Eug]

Originally Posted by jebjeb 

I think what Mastrap has alluded to there is right; if one is happy with a product just becuse we learn info such as this doesn't mean that said product is now no good.

Info like this is very interesting and will probably influence how I use my gear but it does not make me instantly dislike what I have been happy with.

The problem though is that the info isn't really new, and it's kind of misleading.

BTW, one thing that everyone has forgotten to mention is that our pixels aren't even complete pixels. In the old days, 1 Megapixel might have meant 1 million 3-colour pixels. Unfortunately, the pixels with current cameras are one colour only, so the other colours are interpolated.

[Hawkeye_a]

Agreed (jebjeb), different companies make different compromises and so no one product is prerfect. (that applies to ANY industry).

My camera doesnt perform all that well in low light conditions (high ISO)... i knew that before i got it (it is a compact after all), but to me it made up for in the other features it offered that mattered to me, and were not available on other compacts.

If i was primarily going to be doing low-light photography i would have probably gone with the Fuji, cause the results are really good. but since im not, i didnt weight that feature too heavily when making my choice.

result... im happy

[jebjeb]

Originally Posted by Eug 

The problem though is that the info isn't really new, and it's kind of misleading.

BTW, one thing that everyone has forgotten to mention is that our pixels aren't even complete pixels. In the old days, 1 Megapixel might have meant 1 million 3-colour pixels. Unfortunately, the pixels with current cameras are one colour only, so the other colours are interpolated.

Actually, you will find that they have addressed the issue of single colour sensitive pixels in the article. Nathan Myhrvold's perspective at the bottom of the article is what I found to be a bit clearer and more informative.

[MaxPower2k3]

Another important thing to remember is that numbers are easy to sell. From a sales point of view, it's much easier to say "This camera is 6MP. This one is $50 more and 8MP. That's 33% more for $50!" than "This camera is $50 more than that one and has better optics." So, from that point of view, it's much more worthwhile to invest in research to cram bigger numbers into their sales materials than better optics into their cameras. Especially since 90% of consumers who buy point-and-shoots probably couldn't tell the difference between a good camera phone and an SLR, much less a 1/3" or 1/4" sensor.