Why is there an ISO difference between Crop and Full Frame images I've taken?

rob-nikon

rob-nikon

Messages
6,951
Name
Rob
Edit My Images
Yes
Last year I tried to capture two identical images using a crop sensor and a full frame using the 'equivalent field of views' of the two (ie the same framing). For a while I've been think of writing it up into a blog post about crop and full frame sensors for wildlife photography. I tried to use the same settings on both the crop and full frame cameras, the only difference would have been the focal lengths for the same framing (ie crop at 200mm, FF at 300mm). What I've noticed is the full frame camera is one stop lower ISO (I was using auto ISO on both). As the crop camera had a 70-200 f2.8 attached and the full frame camera a 300 f2.8 would the size of the aperture opening on the lens account for the one stop difference in ISO or is caused by the difference in sensor?

I think I need to repeat the test using the same lens on both camera to ensure all is equal. The images where taken quite close together, really only seconds apart in woods on a cloudy day so I think I can rule out light level changes.

The images are below with their settings in the water mark.

D71_8462 by Rob Cain, on Flickr

DSC_9172 by Rob Cain, on Flickr
 
Last edited:
The most amazing thing is that you have time to mess about with cameras and lenses with a red squirrel just a few feet away!

But that shouldn't be Rob. You need to do some controlled comparisons. Only thing I would say is the depth-of-field is quite a lot less on the D800 pic, more than I would expect, as a guess, just from the crop factor. If I'm guessing right, then one of your lenses is not actually at a true f/5.6 for some reason.
 
So what I'm reading is that you've fixed the aperture and shutter speed on both but have ISO on auto. The DoF as expected is narrower on the 300mm. Were the metering settings different? Does the lens affect the metering position or metering averaging area perhaps.
I think you would need to repeat the tests with the lenses swapped on the cameras as well.
 
HoppyUK said:
The most amazing thing is that you have time to mess about with cameras and lenses with a red squirrel just a few feet away!
Click to expand...

:agree:


What metering mode?
 
I guess more info is needed like what metering mode, what AF mode (and where was the AF point).

Just from looking at them the first has a massively wider DOF which will potentially affect the metering, won't it?
 
Last edited:
HoppyUK said:
The most amazing thing is that you have time to mess about with cameras and lenses with a red squirrel just a few feet away!
Click to expand...

The squirrel was quite happy eating, it was a case of firing off a couple of frames then picking up the other camera and fire off a few frames. I had already checked what was good exposures so just set the shutter speed and aperture to the same value on both and had auto ISO on.
 
rob-nikon said:
I had already checked what was good exposures so just set the shutter speed and aperture to the same value on both and had auto ISO on.
Click to expand...

But what metering mode?
 
Broric said:
I guess more info is needed like what metering mode, what AF mode (and where was the AF point).

Just from looking at them the first has a massively wider DOF which will potentially affect the metering, won't it?
Click to expand...

I usually only use matrix metering with AF-C set to single point but moveable around the frame. It's mostly to be on the eye on both images. The strange is it's repeated in the f4 images (but I altered the shutter speed by mistake to 1/640 on one of the cameras so the ISO isn't exactly the same but still roughly one stop difference).
 
HoppyUK said:
The most amazing thing is that you have time to mess about with cameras and lenses with a red squirrel just a few feet away!

But that shouldn't be Rob. You need to do some controlled comparisons. Only thing I would say is the depth-of-field is quite a lot less on the D800 pic, more than I would expect, as a guess, just from the crop factor. If I'm guessing right, then one of your lenses is not actually at a true f/5.6 for some reason.
Click to expand...
I was expecting DoF to be different, that was what part of what my blog post was going to be about. I was wondering if the quality of background blur is improved on a FF sensor for the same FoV.

At 5m (roughly at what this was taken at) the DoF on the crop at 200mm should be 14cm, whilst the FF at 300mm DoF should be 8cm. In the images that looks roughly correct as the FF squirrel doesn't have the tail in focus whilst the log is partly in focus in the crop squirrel image.
 
Last edited:
I think I need to repeat this using the same lens on both a crop and FF camera. Shame I don't have a crop body to do it with! Also the test images of bottles don't look as good as a Red Squirrel :(
 
There's, imo, a couple of things going on, the images have different exposures the first shot of the squirrel is noticeably darker in the tail and fur. They're also taken with different lenses, in addition to F-stop lenses have a T-stop rating and will affect how much of the available light is transmitted to the sensor. Two different lenses read by two different metering systems which are seeing two different scenes due to the difference in depth of field would probably account for the difference in iso calculated. The surprising part is it looks like more than a stops difference as the D800 image looks brighter and has the lower iso.

The only way to truly test this as an issue of the sensor would be to go the controlled lighting environment route with the same lens on each camera and even then there's a different processing chip inside so the resultant images could still be affected by that part of the equation.
 
Last edited:
rob-nikon said:
Am I right in thinking an aperture of say f4 or f5.6 should be exactly the same size on all lenses and it should let exactly the same amount of light in?
Click to expand...
Yes,

For a given 'brightness' of scene, the exposure is a combination ISO, f-Stop and Shutter, so if ISO and Shutter are the same, aperture shoudl be as well, regardless of the lens.

The caveat to that is that you also have a 'T' value for lenses, which is the light transmission value - but for the majority of SLR type lenses, the Aperture and T values are the same I believe (though happy to be corrected by anyone with more detailed understanding :))
 
The more glass in a lens the less light gets through it, and so the greater the difference between the lenses' T-stops (how much light actually gets through) and their f-stops (how much light would get through if there were no loss). Lens coatings also play a part in light loss. Other things being equal (which they're usually not) there will be more glass in a zoom lens than a prime.
 
rob-nikon said:
Am I right in thinking an aperture of say f4 or f5.6 should be exactly the same size on all lenses and it should let exactly the same amount of light in?
Click to expand...
Not quite. The aperture diameter, or more accurately, its apparent diameter viewed from the rear, will be the same for all lenses of the same focal length at the same f setting.

The f No. is a ratio and is given by the focal length divided by this diameter. e.g. a 50mm lens at f8 will have a 6.25mm aperure and at f2.8 it will open up to 17.8mm

A 300mm lens at f8 will have a 37.5mm aperture and at f2.8 it will open up to 107mm. This also explains why fast, long focal length lenses are big!


Steve.
 
Is it me but it looks like the same shot twice, nothing has changed in the pose at all!!!!
 
CraigDHD said:
There's, imo, a couple of things going on, the images have different exposures the first shot of the squirrel is noticeably darker in the tail and fur. They're also taken with different lenses, in addition to F-stop lenses have a T-stop rating and will affect how much of the available light is transmitted to the sensor. Two different lenses read by two different metering systems which are seeing two different scenes due to the difference in depth of field would probably account for the difference in iso calculated. The surprising part is it looks like more than a stops difference as the D800 image looks brighter and has the lower iso.

The only way to truly test this as an issue of the sensor would be to go the controlled lighting environment route with the same lens on each camera and even then there's a different processing chip inside so the resultant images could still be affected by that part of the equation.
Click to expand...
I edited these in Lightroom so there is possible a slight difference. I'm thinking it needs to be tested like you say same lens in controlled light conditions. As I don't have a crop camera or controlled conditions it's not likely to be tested properly.
 
rob-nikon said:
Am I right in thinking an aperture of say f4 or f5.6 should be exactly the same size on all lenses and it should let exactly the same amount of light in?
Click to expand...

Apertures should be the same in theory, but in practise the T.stop may be different (though not much, and nowhere near one stop) or the lens is not stopping down accurately (quite common). If there's a problem there, it could be inconsistent too. Variable aperture zooms can be quite a long way out compared to marked aperture values, depending on focal length setting.

The other likely variable is the light level, changing imperceptively. Shutter speed and ISO should be reasonably close. Or quite possibly you have a compound mixture of small errors all adding up.

I'm assuming the histograms are similar? Same exposure adjustments in LR?
 
HoppyUK said:
I'm assuming the histograms are similar? Same exposure adjustments in LR?
Click to expand...

The RAW histograms are roughly the same positional shape but the crop sensor is capturing higher levels than the FF which is probably expected considering the higher ISO being more sensitive. Having reset both images to RAWs in lightroom I've now noticed the crop image was taken at 185mm not 200mm and I had previously cropped to the same field of view (I had edited these 13 months ago and exported this morning but not remembered exactly what I had done). I think it's a case of small variables (slightly wider focal length of 185mm on the crop so metering more background, different metering modules in the cameras, light changing slightly) making a difference but also the use of auto ISO making the decision on the ISO value.

I'm definitely not using these as examples for a blog post as there are too many small discrepancies to make it a fair comparison. I think it would need retesting with static subjects, in controlled light conditions, using the same lens and no use of auto ISO.
 
•
In the case of these scenes, such differences would puzzle me at first
but seeing that it is within the dynamic range, I wouldn't worry too much.
 
Well this has puzzled me all day! Thinking of the sunny 16 rule the exposure would be the same on a crop or full frame sensor if you had manually entered the same aperture, shutter speed and ISO so this has to have something to do with how the two camera chose the auto ISO value. This has got me thinking that both cameras use different metering chips, the D800 uses 91k pixel RGB metering and the d7100 uses 2016 pixel RGB metering. Has the slightly wider focal length of 185mm on the crop body meant more dark wood background was metered useing a lower spec metering system, causing auto ISO to choose an ISO one stop higher than the d800 chose. The d800 was metering a slightly smaller dark wood background but using a higher spec metering system?
 
rob-nikon said:
Well this has puzzled me all day! Thinking of the sunny 16 rule the exposure would be the same on a crop or full frame sensor if you had manually entered the same aperture, shutter speed and ISO so this has to have something to do with how the two camera chose the auto ISO value. This has got me thinking that both cameras use different metering chips, the D800 uses 91k pixel RGB metering and the d7100 uses 2016 pixel RGB metering. Has the slightly wider focal length of 185mm on the crop body meant more dark wood background was metered useing a lower spec metering system, causing auto ISO to choose an ISO one stop higher than the d800 chose. The d800 was metering a slightly smaller dark wood background but using a higher spec metering system?
Click to expand...

I thought we'd established that the two images were very similarly exposed, if not identically? Any adjustment in post processing must be factored in. If not, then it could simply be a metering error, and not that much of a puzzle.
 
HoppyUK said:
I thought we'd established that the two images were very similarly exposed, if not identically? Any adjustment in post processing must be factored in. If not, then it could simply be a metering error, and not that much of a puzzle.
Click to expand...
It may be easier for me to add screen shots.

Unedited RAW images

RAW Images Unedited by Rob Cain, on Flickr
Left D7100 Right D800

There is the difference in FoV due to it actually being taken at 185mm on 70-200 f2.8 VR2 (may have a little lens breathing too) and not an exact FoV.

Histograms

D800 Histogram by Rob Cain, on Flickr

D7100 Histogram by Rob Cain, on Flickr

I'm starting to thinking metering error makes sense as there are darker areas in the crop image that could have fooled it into raising the ISO.
 
Last edited:
mike weeks said:
Do not forget that the metering is optimised to the camera so things like the cameras dynamic range will effect how the camera meters.

Mike
Click to expand...

Thats another interesting point, both DR and metering are better on the D800 so perhaps it has just metered the scene better and auto iso has then chosen a lower ISO value.
 
mike weeks said:
Do not forget that the metering is optimised to the camera so things like the cameras dynamic range will effect how the camera meters.

Mike
Click to expand...

Thats another interesting point, both DR and metering are better on the D800 so perhaps it has just metered the scene better and auto iso has then chosen a lower ISO value.
 
The metering may be out, but that's kind of by the way - it's the histograms that show what's actually going on. And they make the difference greater I think - the D800 image has abut 1/3rd stop more exposure than the D7100, confirmed by the histogram and by looking at the images So factoring that in, you're looking to explain a 1.3 stops difference when, in theory (and in answer to your original question) there should be none.

That's a big change, and if there are equipment errors, like aperture accuracy etc etc, then some controlled tests would quickly confirm that. I still suspect the apertures may be out, but that's mainly down to the increased DoF shown by the D71000 shot that, guessing, looks like more than the format change would deliver. In other words, your D7100 is actually shooting at closer to an effective f/8.

If it's not that, or not only that, then on balance I think changing light is the most likely culprit.
 
HoppyUK said:
The metering may be out, but that's kind of by the way - it's the histograms that show what's actually going on. And they make the difference greater I think - the D800 image has abut 1/3rd stop more exposure than the D7100, confirmed by the histogram and by looking at the images So factoring that in, you're looking to explain a 1.3 stops difference when, in theory (and in answer to your original question) there should be none.

That's a big change, and if there are equipment errors, like aperture accuracy etc etc, then some controlled tests would quickly confirm that. I still suspect the apertures may be out, but that's mainly down to the increased DoF shown by the D71000 shot that, guessing, looks like more than the format change would deliver. In other words, your D7100 is actually shooting at closer to an effective f/8.

If it's not that, or not only that, then on balance I think changing light is the most likely culprit.
Click to expand...

That's a good explanation, unfortunately I can't test any further as these were taken over a year ago and don't have them any more.
 
As I understood it, the ISO value is just a nominal figure and not an exact measurement. Furthermore larger sensors allow for larger pixel size (usually) so are more sensitive to light, perhaps requiring a lower ISO value for the same exposure. It would be difficult to compare crop sensors to full frame and be sure their "sensitivity" would be set the same.
 
TimmyG said:
As I understood it, the ISO value is just a nominal figure and not an exact measurement. Furthermore larger sensors allow for larger pixel size (usually) so are more sensitive to light, perhaps requiring a lower ISO value for the same exposure. It would be difficult to compare crop sensors to full frame and be sure their "sensitivity" would be set the same.
Click to expand...

Yeah I read an article in Amateur Photographer by Proffesor Bob Newman which stated that sensitivty is not changed just the algorithim for processing the information is changed. Different sensors and sizes differ in sensitivty and it why some sensors give thier best results at different ISO setiings ie maybe ISO 200 not ISO 100.

So they have different sensors that perform differently.
 
The T stop for the two lenses is within .1 (3.3/3.2) of each other, not enough to matter. I have the 70-200 VRII and it breathes pretty badly inside 10m... (It's around 135mm at MFD/200mm).
A shorter lens at the same aperture will have greater DOF, the sensor factor was largely negated by the change in FL (i.e. same subject distance). 180mm/crop will be ~ 2x the DOF of 300mm/FF at the same distance. But it's still pretty shallow I would guess (10-15ft?). However, DOF says noting about the OOF parts (character/size/what's included) and IMO that's the most significant part of the apparent DOF difference. http://howmuchblur.com/#compare-1x-300mm-f5.6-and-1.5x-200mm-f5.6-on-a-0.9m-wide-subject
I see a bit of a histogram shift due to composition and BG. I don't see a significant difference in exposures. I would agree that the D810 image is about 1/3 brighter but it seems to be mostly in the highlights... My best guess, the lighting changed and became slightly harder/stronger.

I have done similar (informal) tests and have not found more than a .3 difference between cameras using REI for their ISO determination, crop or FF. I believe there is a DSLR or two out there that rates it's ISO notably differently than the rest, but not w/in the Nikon line.
(I have found significant exposure differences between compacts using SOS against cameras using REI)
 
Last edited:
TimmyG said:
As I understood it, the ISO value is just a nominal figure and not an exact measurement. Furthermore larger sensors allow for larger pixel size (usually) so are more sensitive to light, perhaps requiring a lower ISO value for the same exposure. It would be difficult to compare crop sensors to full frame and be sure their "sensitivity" would be set the same.
Click to expand...

alfbranch said:
Yeah I read an article in Amateur Photographer by Proffesor Bob Newman which stated that sensitivty is not changed just the algorithim for processing the information is changed. Different sensors and sizes differ in sensitivty and it why some sensors give thier best results at different ISO setiings ie maybe ISO 200 not ISO 100.

So they have different sensors that perform differently.
Click to expand...

There is some truth in that, but it doesn't explain a 1.0-1.3 stops difference, especially with cameras of the same brand and sensor/processor of roughly similar generation. ISO with digital can be a bit of a movable feast (basically, some manufacturers simply lie to get high numbers for marketing) but there should be a firm basis for it and all those things mentioned above are factored in. That is, a mid-grey subject should be recorded in the middle of the histogram with standard JPEG settings. When printed out correctly, with no post-processing adjustments, that mid-grey should be the same tone as the original subject if you put them side by side, and all other tones will fall in line.

That is technically 'correct' exposure, though that's not necessarily the same as optimum exposure for a particular subject. For squirrel shots here for example, from the histograms, they could both take another stop or so of exposure to extract the last bit of quality - particularly shadow detail in the fur/tail, with a bit less noise. When shooting Raw, there is at least one stop of highlights headroom to the right even when the blinkies are flashing.
 
Fair enough. I'm going with changing ambient light levels between shots then. Seems to me this would be something to investigate under controlled lighting conditions.
 
Some interesting reading on this thread, for someone like me who doesn't get involved to much in the tech side of cameras.

Anyway...........

I had a thought a few months ago when I was purchasing my 70-300 lens for my Nikon. I was looking at the Nikon 70-300vr which has a front element filter size of 67mm and the Tamron 70-300vc lens which has a front element filter size of 62mm, so in theory the Nikon lens should let in about 7.5% more light than the Tamron due to it being more surface area, but the apertures for the lenses are roughly the the same...Tamron f4-5.6 and Nikon f4.5 -f5.6.

The above is based on the filter thread size for the lenses, so the glass will be a little smaller and with this thought, would this affect the above sampling or similar tests?

I just checked the current versions of the Nikon lenses Rob is using and the 70-200 has a filter size of 77mm and the 300mm has a filter size of 52mm?? or are the speccing a drop in filter for the prime as 52mm seems very small

70-200 spec > https://www.nikonusa.com/en/nikon-products/product/camera-lenses/af-s-nikkor-70-200mm-f%2f2.8g-ed-vr-ii.html#tab-ProductDetail-ProductTabs-TechSpecs

300 spec > http://www.nikonusa.com/en/nikon-products/product/camera-lenses/af-s-nikkor-300mm-f%2f2.8g-ed-vr-ii.html#tab-ProductDetail-ProductTabs-TechSpecs
 
Last edited:
TimmyG said:
Fair enough. I'm going with changing ambient light levels between shots then. Seems to me this would be something to investigate under controlled lighting conditions.
Click to expand...

Just as a very general observation, I've often tried to compare the same shot on two different lenses. Sometimes I do it when I'm outside and all set up for a shot, and have some time to kill. So many of those tests have been degraded by a stop or so of lighting difference I didn't notice at the time that I've decided wherever possible to do such tests indoors at night with my own controlled artificial light.
 
the way i understand it is it's something to do with overall light at a given aperture
in other words a full frame sensor with an aperture of F2.8 will gather more light than a crop sensor at the same aperture because of the bigger area of the full frame sensor or am i barking up the wrong tree with these thoughts ?

it would go some way to explaining the differences in ISO
 
You must log in or register to reply here.
Back
Top