Sensors

JulesP

JulesP

Messages
2,748
Name
Julian
Edit My Images
Yes
Here comes another full frame V crop thread:)

I hope I'm missing something here as I've been doing research regarding FX V DX (full frame V crop).
I always understood it that the FX sensors were "better" at light gathering due to the size of the pixels on the sensor and that larger pixels meant less noise at higher ISO's - correct?

OK - here's the bit I don't get

D7000 pixel size - 4.78 µm
D7100 pixel size - 3.9 µm
D700 pixel size - 8.45 µm
D800 pixel size - 4.88 µm
D810 pixel size - 4.8 µm
D610 pixel size - 5.96 µm

Surely if the above theory is correct, The D700 should have the best sensor as its got the largest pixels. Then should come the D610 . The D7000, D800 and D810 should be pretty much equal and the D7100 should be downright terrible as it has the tiniest pixels.

I thought that the larger pixels that gathered more light, had less noise and had a better dynamic range than the smaller pixels. Somewhere along the line this theory seems to have a major flaw in it as it seems the smaller pixels are winning these days.
Yes I get it that smaller and more pixels mean sharper detail but I don't get the FX V DX thing when it comes to noise and dynamic range.
I also get the bit about sensor size and depth of field so don't need an explanation of that bit.
I was doing this research as I was looking at upgrading my D7000 to a D610 but am struggling now to justify it at all.

Anyone got any explanation for this?

Thanks

Julian
 
There is a similar thread a bit lower down entitled "when to go FF"
http://www.talkphotography.co.uk/index.php?threads/When-to-go-FF.556916/

However in summary I think with the latest sensor technology an PP software, there isn't the same differential between Crop and FF, unless you are going to be printing large images.

I love Zack Arias's take on this subject in the below video, it's a bit tongue in cheek, but if you read between the lines he has a very valid point.


Having upgraded myself I only really notice the difference once I get past ISO 1600. What I will say is that when you get to FF lenses cost a lot more, because instead of using a 400mm on a crop costing circa £1k you are looking at a 600mm at circa £7k to fill the same amount of paper the alternative is cropping the FF shot and enlarging, therefore losing 1/3 of the pixels and stretching the remaining ones.
 
Last edited:
Yeah I read that thread and watched the video.
Still doesn't really explain the noise / dynamic range thing though apart from the word 'negligible' which probably pretty much sums it up.
Most of my lenses are full frame anyway - bought them that way with the intention of following that upgrade path so I wouldn't need to replace them.
 
JulesP said:
Here comes another full frame V crop thread:)

Surely if the above theory is correct, The D700 should have the best sensor as its got the largest pixels. Then should come the D610 . The D7000, D800 and D810 should be pretty much equal and the D7100 should be downright terrible as it has the tiniest pixels.
Click to expand...
Improvements in sensor and amplifier technology means that you can't compare two bodies that are 4 1/2 years apart in their design and expect not to see the later ones showing improvements in noise and dynamic range. The only realistic comparison would be within a fairly tight timeframe.

Bob
 
Canon Bob said:
Improvements in sensor and amplifier technology means that you can't compare two bodies that are 4 1/2 years apart in their design and expect not to see the later ones showing improvements in noise and dynamic range. The only realistic comparison would be within a fairly tight timeframe.

Bob
Click to expand...
Yes that is true - which is why I added the sensors of the D7100, D800 and D610 (same sensor as D600) which are all about the same age. Surely in that respect the D610 should be better than the D810 when it comes to dynamic range and noise?
 
I'm not really Nikon literate but isn't the D810 classed as an "high-end" body whereas the D610 is only an "advanced" body. Simply looking at the sensor will never tell the whole story....it certainly doesn't with Canon's offerings. The amplifiers, unity gain and other in-camera processing all have an effect on the final raw output.

Bob
 
Full-frame has better ISO performance, less noise and greater dynamic range, because it's physically larger than APS-C. More light falls on FF, more photons - and that's unrelated to pixel size.

Final image quality then depends on numerous other factors, like the size of pixels and how well they collect light, with microlenses, gapless microlenses etc, and then there's the efficiency of the processing engine (amplifier). All these things are getting better.

As far as sharpness is concerned, that hasn't got much to do with pixels either (within reason). FF is sharper than APS-C again because it is larger. So the image doesn't have to be enlarged so much, the resolution demands are lower, and therefore lens image contrast goes up (basic lens MTF theory).
 
But how can it be unrelated to pixel size as it's those pixels that record the light falling on them?
I was always lead to believe that FF was 'better' as the larger pixels could record more light information / dynamic range.
 
HoppyUK said:
Full-frame has better ISO performance, less noise and greater dynamic range, because it's physically larger than APS-C. More light falls on FF, more photons - and that's unrelated to pixel size.

Final image quality then depends on numerous other factors, like the size of pixels and how well they collect light, with microlenses, gapless microlenses etc, and then there's the efficiency of the processing engine (amplifier). All these things are getting better.
Click to expand...

As I've said before I've never read an explanation of this that makes sense to me. Here's my reasoning... and I'm assuming a perfect lens and the same settings on FF and APS-C cameras bolted to a tripod and shooting the same scene from the same point.

-More photons will fall onto the FF sensor than onto the APS-C sensor because the FF sensor is physically larger than the APS-C sensor.
-The same number of photons will fall onto an APS-C sized centre section of the FF sensor as will fall on the APS-C sensor.
-The extra photons which the FF chip collects over those of the APS-C sensor fall outside of the APS-C sized centre section of the FF chip.
-The APS-C sensor should therefore have the same number of photons and the same image properties as the APS-C sized section of the FF sensor.
-Any difference in properties should therefore be due to the degree of image enlargement applied.

Perhaps noise and DR are affected by enlargement? From my days in electronics and step wedge tests I'd say that looking at something larger or looking closer at something smaller enables you to see more detail and perhaps you'll see another step you couldn't quite see before and you'll certainly see more faults but actual DR shouldn't be affected by enlargement, within reason, surely? All bets are off if you go to stupid differences like comparing a postage stamp sized image to a barn door by eye.

There will be problems comparing FF and APS-C sensors in different cameras as even if you find two cameras that you think have the same technology inside them you can't be sure that the technology is actually identical. Maybe the only way to test the theory that FF sensors are better because they gather more light in total because they are larger is to guarantee that the technology is actually identical. You could do this by using a FF camera in APS-C crop mode. This will guarantee the same pixels, the same micro lenses, the same electronics and the same in camera processing. Or, some cameras have the ability to digitally zoom. However, maybe it's easier to just crop a FF image post capture?

If larger sensors gave less noise and more dynamic range than smaller ones of exactly the same design and technology perhaps we should see an increase in noise and a drop in dynamic range if we use a FF sensor in APS-C crop mode or indeed if we digitally zoom or simply crop a FF image. Does this happen? Perhaps that's mad reasoning but comparing a 7D to a 5D seems pointless to me as there are too many differences.

And just to be clear, I was an electronic engineer and I worked in computer manufacture and repair and this explains my bafflement at optical systems. If I had any APS-C lenses that would fit my A7 I'd try a few shots in crop mode but I haven't and all I can say for now is that cropping an A7 image doesn't seem to lower the DR to a degree that is noticeable or indeed show any effect that can't be explained by the different levels of enlargement.

As I worked in electronics it's easier for me to see reasons for one sensor to perform better than another in differences in the technology rather than simply and purely the physical size of the chip, other than the differences that can be explained by different levels of enlargement. Maybe this is why we use the word "crop"?
 
JulesP said:
But how can it be unrelated to pixel size as it's those pixels that record the light falling on them?
I was always lead to believe that FF was 'better' as the larger pixels could record more light information / dynamic range.
Click to expand...

I believe that a lot (maybe all) can be explained by the advances in technology. When sensors first came out things weren't done as well as they could have been and now designers and production engineers have learned and things can and are done... better. On chip analogue to digital conversion and what/how it's done being one example, micro lenses another. Most people wouldn't believe the effect that even the length and shape or a track either on a circuit board or inside a chip can have on a signal and then there's the route taken, noise, grounding and everything else that goes into the mix.
 
JulesP said:
But how can it be unrelated to pixel size as it's those pixels that record the light falling on them?
I was always lead to believe that FF was 'better' as the larger pixels could record more light information / dynamic range.
Click to expand...

Please re-read my post.

In terms of noise, ISO and dynamic range etc, the key driver is the total amount of light falling on the sensor, ie total area, and FF has more than double the area of APS-C. The second phase is how much of that light is actually collected (pixel size, microlens efficiency etc), and the third phase is how good the camera's processing engine is in converting that into a recorded image.

Pixel size comes into phase two, and larger pixels tend to have a higher signal-to-noise ratio (though technology is constantly moving the goalposts there) but if you have smaller pixels, there are more of them collecting light. Bottom line is it's better to have lots of smaller pixels over a larger area (FF) than it is to have fewer larger pixels over a smaller area (APS-C). This is fact.

woof woof said:
As I've said before I've never read an explanation of this that makes sense to me. Here's my reasoning... and I'm assuming a perfect lens and the same settings on FF and APS-C cameras bolted to a tripod and shooting the same scene from the same point.

-More photons will fall onto the FF sensor than onto the APS-C sensor because the FF sensor is physically larger than the APS-C sensor.
-The same number of photons will fall onto an APS-C sized centre section of the FF sensor as will fall on the APS-C sensor.
-The extra photons which the FF chip collects over those of the APS-C sensor fall outside of the APS-C sized centre section of the FF chip.
-The APS-C sensor should therefore have the same number of photons and the same image properties as the APS-C sized section of the FF sensor.
-Any difference in properties should therefore be due to the degree of image enlargement applied.

Perhaps noise and DR are affected by enlargement? From my days in electronics and step wedge tests I'd say that looking at something larger or looking closer at something smaller enables you to see more detail and perhaps you'll see another step you couldn't quite see before and you'll certainly see more faults but actual DR shouldn't be affected by enlargement, within reason, surely? All bets are off if you go to stupid differences like comparing a postage stamp sized image to a barn door by eye.

There will be problems comparing FF and APS-C sensors in different cameras as even if you find two cameras that you think have the same technology inside them you can't be sure that the technology is actually identical. Maybe the only way to test the theory that FF sensors are better because they gather more light in total because they are larger is to guarantee that the technology is actually identical. You could do this by using a FF camera in APS-C crop mode. This will guarantee the same pixels, the same micro lenses, the same electronics and the same in camera processing. Or, some cameras have the ability to digitally zoom. However, maybe it's easier to just crop a FF image post capture?

If larger sensors gave less noise and more dynamic range than smaller ones of exactly the same design and technology perhaps we should see an increase in noise and a drop in dynamic range if we use a FF sensor in APS-C crop mode or indeed if we digitally zoom or simply crop a FF image. Does this happen? Perhaps that's mad reasoning but comparing a 7D to a 5D seems pointless to me as there are too many differences.

And just to be clear, I was an electronic engineer and I worked in computer manufacture and repair and this explains my bafflement at optical systems. If I had any APS-C lenses that would fit my A7 I'd try a few shots in crop mode but I haven't and all I can say for now is that cropping an A7 image doesn't seem to lower the DR to a degree that is noticeable or indeed show any effect that can't be explained by the different levels of enlargement.

As I worked in electronics it's easier for me to see reasons for one sensor to perform better than another in differences in the technology rather than simply and purely the physical size of the chip, other than the differences that can be explained by different levels of enlargement. Maybe this is why we use the word "crop"?
Click to expand...

Maybe you're making this more complicated than it is. In your scenario, you are comparing the APS-C sized area in the middle of a FF sensor - in which case, no difference; you've just created an APS-C format camera. But the whole point of full-frame is the larger image area, that's where all the advantage lies.
 
Forget numbers. Just shoot two identical pictures with the cameras you are considering buying... then look at the pictures. The one that looks best to you wins. The end.


This is just the same conversation with the same people, saying the same things. Just go search the forums... save us all typing this shizzle again. I've posted MANY things in here with real, practical examples of why looking numbers is just a load of [PLEASE DON'T TRY TO BYPASS THE SWEAR FILTER] that's only really of any interest to camera nerds.
 
Am I missing something here?

Isn't the point that if you can't justify a move to FF then you don't need to? If you needed FF then you'd already know why and move to it! You are researching the sensor rather than the results.
 
It was a lot simpler with film...
 
You must log in or register to reply here.
Back
Top