Why resolution matters

[eccles]

If you had the A700 with a longer reach and equally as high quality lens, do you reckon the quality of the image woudl be roughly the same or far superior on the longer lens?

This is the kind of test I'm talking about

I can probably do a reasonably meaningful test with A700 + 70-400 at 400mm compared to A77 + 70-400 at 285mm. I'll give it a try shortly.

 

[gman]

ha! I never even thought of using a zoom lens! Nice one, thanks and sorry for the hijack - although somewhat related I guess?

 

[eccles]

ha! I never even thought of using a zoom lens! Nice one, thanks and sorry for the hijack - although somewhat related I guess?

NP, and very related Id say. I'll try that test tomorrow if I have time. Empirically speaking I think the extra digital reach is very real but I'd be interested to know by how much myself.

 

[gman]

I as having a think about the argument regarding the lens quality and how it wouldn't be fair to use different lenses but is that a fair point to raise?

I'm not looking to find out the definitive scientific answer but simply the real world result with the currently available tools (aka lenses) and if there isn't a lens available that can resolve the detail of the D800 sensor well enough to match a longer focal length lens on a lower resolution camera, then that has to be a factor to be taken into consideration surely?

I'm of the opinion that a high resolution sensor will be able to crop and retain quality up to a certain level but perhaps at extreme cropping I reckon there would be a noticable difference as one would probably expect?

 

[HoppyUK]

I as having a think about the argument regarding the lens quality and how it wouldn't be fair to use different lenses but is that a fair point to raise?

I'm not looking to find out the definitive scientific answer but simply the real world result with the currently available tools (aka lenses) and if there isn't a lens available that can resolve the detail of the D800 sensor well enough to match a longer focal length lens on a lower resolution camera, then that has to be a factor to be taken into consideration surely?

I'm of the opinion that a high resolution sensor will be able to crop and retain quality up to a certain level but perhaps at extreme cropping I reckon there would be a noticable difference as one would probably expect?

Lenses are the main problem. As resolution goes up, so contrast goes down. And the smaller the sensor, for a given level of resolution, the lens has to work much harder. It's a bit like expecting a car that accellerates from 0-60 in six seconds, to go from 60-120 in the same time. It doesn't happen.

Within reason, sharpness is much more about lenses and the area of the sensor than it is about pixels.

We're looking at on screen images here and saying oh that looks great, but how much detail are we actually looking at? Probably 1m max. Work it out. The kind of detail you can get from even modest sensors is way above what we can even see in a normal size output. It's the lens doing the real work, maintaining high contrast which is what we preceive as 'sharpness' far more than high numbers of pixels.

 

[gman]

So is it the MP quantity or the pixel density that causes the problem?

 

[HoppyUK]

So is it the MP quantity or the pixel density that causes the problem?

All things being equal, ie number of pixels and similar lens, it is the physical area of the sensor that makes the difference.

A larger sensor collects more light, it has more of the basic raw material to play with, so ISO performance is better - wider dynamic range, lower noise.

Secondly, the bit that many people forget is the lens. It has to work harder with a smaller sensor to deliver the same resolution, so contrast is lower and the image has less punch/pop/whatever that you cannot replicate in post processing.

The differences are not huge in visual terms, but they are there, and these are the reasons why bigger sensors always win in all aspects of image quality.

Then there is the 'all things being equal' bit, which of course they rarely are. Technology plays a part and until recently is has been smaller formats that have been getting most of the attention. Crop format and 4/3rds has been closing the gap

What is particularly interesting about the D800 and 5D3 is that they are the first new full-frame sensors for a few years, so that should level the playing field a bit and FF should show a clearer advantage again.

 

[gman]

Ah I was wondering about medium format so pixel density does play an important role then. I'm still curious to see how noticeable the difference may be.

 

[eccles]

@Gman, I don't think there will be a problem finding lenses to outresolve the D800 sensor. The SLT-A77 has even greater pixel density and I have three primes that can match it. I think my two 'G' zooms give it a good run for its money as well. Some of my lesser lenses such as the 70-210 F/4 beercan and 28-105mm series 2 zoom are slipping behind though.
What is remarkable about the D800, if the DXOMark results are correct, is that its sensor has increased pixel density to that of a crop format camera, whilst still managing to significantly increase dynamic range and colour depth, maintaining good high iso performance as well. I recently downloaded a raw test file from Imaging Resource from the D800 at iso 6400 and ran it through ACR 6.7 candidate release and Topaz NR. The results were amazingly good and certainly bears out the DXO numbers.

 

[HoppyUK]

@Gman, I don't think there will be a problem finding lenses to outresolve the D800 sensor.

That isn't what Nikon says in their own literature, ie only the best lenses will do (16 out of 60 Nikkors recommended) and no higher than f/8.

The SLT-A77 has even greater pixel density and I have three primes that can match it. I think my two 'G' zooms give it a good run for its money as well. Some of my lesser lenses such as the 70-210 F/4 beercan and 28-105mm series 2 zoom are slipping behind though.

What is remarkable about the D800, if the DXOMark results are correct, is that its sensor has increased pixel density to that of a crop format camera, whilst still managing to significantly increase dynamic range and colour depth, maintaining good high iso performance as well. I recently downloaded a raw test file from Imaging Resource from the D800 at iso 6400 and ran it through ACR 6.7 candidate release and Topaz NR. The results were amazingly good and certainly bears out the DXO numbers.

That Sony A77 sensor is excellent, but in comparing formats it would be interesting to put the same lens on a Sony A900, where I suspect the A900 would win.

I have had a Sony NEX-7 on loan (same sensor) and, very good though it is, my Canon 5D2 betters it.

 

[SamHH]

I would love to see a side by side example of the same subject taken at the same distance but one with the subject filling the frame due to cropping and the other filling the frame without cropping but by using a longer lens. It would be interesting to see how cropping with the large mp sensors stack up against pure optics.

I think the examples one-third of the way down this page may help answer your question:

http://www.the-digital-picture.com/Reviews/Canon-Extender-EF-2x-III-Review.aspx

You can see that, even though teleconverters degrade the quality of the image, using a longer focal length is better than using a shorter focal length and cropping.

 

[Ed Sutton]

Secondly, the bit that many people forget is the lens. It has to work harder with a smaller sensor to deliver the same resolution, so contrast is lower and the image has less punch/pop/whatever that you cannot replicate in post processing.

Is that why the results on full frame from a 'cheap' lens look better than when the same lens is used on a crop sensor? Because that's what my eyes tell me.

 

[gman]

I think the examples one-third of the way down this page may help answer your question:

http://www.the-digital-picture.com/Reviews/Canon-Extender-EF-2x-III-Review.aspx

You can see that, even though teleconverters degrade the quality of the image, using a longer focal length is better than using a shorter focal length and cropping.

That's quite a substantial difference to my eyes anyway! Thanks for the link, it's affirming to me that there is definitely scope for looking into the extra reach by cropping concept deeper. The first images in this thread certainly show an excellent quality image by cropping but now I wonder would the image be even better to a significant level were the A700 used with a longer reach quality lens? So if one is intending to get the extra resolution simply for extra reach alone, it may be more cost effective to simply upgrade the lens rather than the camera perhaps?

So many questions, so many thoughts! lol

 

[simonblue]

Someone who doesn't shoot small birds very often, I'll wager.

I understand,but even with small birds,they will be an lens long enought to fill the frame.

And to me part of being a photographer is to solve the any problems at the camera end,not to just shoot away,and think it doesnt matter,because theirs alway pp.

 

[StewartR]

if your croping all the time,your using the wrong lens

Someone who doesn't shoot small birds very often, I'll wager.

I understand,but even with small birds,they will be an lens long enought to fill the frame.

I'm with Simon here. There's always a lens solution, though it may not always be practical. Have you seen my other thread?
Garden birding fun at f=2000mm

 

[eccles]

Guys, these are arguments for the sake of being pedantic.

Yes, full frame cameras will generally speaking give better images than APS-C but they will also lack digital reach. You cannot compare like for like because crop cameras invariably have higher pixel densities, and in any case the development of both formats is at different technological levels.

As for cropping bird shots, there may be a lens solution on an unlimited budget and a truck to carry the gear but if you carry your camera around and photograph on the go it's always going to be a compromise. I for one wouldn't risk my back carrying a 500mm F/4 lens weighing nearly 4kg around a wildlife park.
In such cases the higher pixel density wins.

And before anyone mentions iso, yes I know a 24mp crop sensor loses out at high iso when viewed at 100% and compared to one at 16mp also at 100%.

Anyway, I took two indoor shots using flash and a tripod, one with my 12mp A700 with the 70-400mm zoom at 100mm, the other with my 24mp A77 and the same lens at 70mm. Both images are the same scale when viewed at 100% and for all intents and purposes show the same amount of detail.
Here are 100% centre crops. Exif is intact. See if you can work out which is which before checking the exif data.

Mike.

 

[gman]

Thanks for doing the test and although the A700 does have the edge I wouldn't say it's substantial, in fact it's probably only really noticeable when you compare side to side which nobody would be doing in practice anyway! lol

I'd agree definitely that for portability reasons the high MP has major advantages

 

[flossie]

This is getting confusing. It's a really simple experiment, providing you have both a D800 and a D7000. Not a D700, that's not the comparison being discussed.

Very sharp lens, f/5.6 to optimise sharpness below the diffraction ceiling, then just swap cameras and crop the D800 image to the same framing as the D7000.

They will be practically indistinguishable.

Complete and utter rubbish. The DoF will be different. And that's just physics, before you get into the more subjective areas of contrast, sensitivity, colour rendition, etc etc... if photography was just about pixel densities we'd all be shooting with phones..

 

[squishy]

Complete and utter rubbish. The DoF will be different. And that's just physics, before you get into the more subjective areas of contrast, sensitivity, colour rendition, etc etc... if photography was just about pixel densities we'd all be shooting with phones..

How could the DoF be different? With the same lens at the same aperture at the same distance, a 1.5x crop from a D800 image will be identical to a D7000 image. It can't not be, the D7000's sensor technology is almost identical to the D800's, it's just 1/2.25 times the area.

Nobody is saying photography is just about pixel density, what's being discussed here are specific circumstances where having a higher pixel density can be advantageous in capturing more details.

 

[arad85]

How could the DoF be different? With the same lens at the same aperture at the same distance, a 1.5x crop from a D800 image will be identical to a D7000 image. It can't not be, the D7000's sensor technology is almost identical to the D800's, it's just 1/2.25 times the area.

I'll step away from the keyboard in a minute, but sensor size is inherent to the calculation of DoF. It's rolled in to the calcs as CoC. The bigger the sensor, the bigger the CoC.

 

[weybourne]

I'll step away from the keyboard in a minute, but sensor size is inherent to the calculation of DoF. It's rolled in to the calcs as CoC. The bigger the sensor, the bigger the CoC.

Eh? I don't know much about this stuff but surely in Squishy's explanation, as I understand it, the sensor is the same distance from the lens in both cases (even though one of the sensors is smaller) - the image cast by the lens is the same size in both instances, and the usable bit (i.e. after the crop of the FF image) that we're talking about is from equally sized portions of the sensors in each instance - so the CoC is the same.

 

[HoppyUK]

Complete and utter rubbish. The DoF will be different. And that's just physics, before you get into the more subjective areas of contrast, sensitivity, colour rendition, etc etc... if photography was just about pixel densities we'd all be shooting with phones..

Please read my post again properly.

 

[arad85]

Eh? I don't know much about this stuff but surely in Squishy's explanation, as I understand it, the sensor is the same distance from the lens in both cases (even though one of the sensors is smaller) - the image cast by the lens is the same size in both instances, and the usable bit (i.e. after the crop of the FF image) that we're talking about is from equally sized portions of the sensors in each instance - so the CoC is the same.

Oops... I missed the crop and just saw the 2.25 as big (I blame the wine ). If you crop the image, you effectively change the sensor size, so if you crop the image to the same field of view as the D7000, you effectively have a D7000 sized sensor - so yes, they will be "identical".