Is there a reason why my FX lenses seem to look less sharp on DX bodies?

Naboo32

Naboo32

Messages
3,278
Name
Andy
Edit My Images
Yes
Although I'm a Nikon user, this is a general question about full frame lenses on crop sensor bodies, so anyone is welcome to join in :).

I've been playing around with different combinations of lenses and bodies lately and think that I can see a trend developing, which I hadn't thought about before. Unfortunately, because these were mostly from experiments at home, I didn't keep any of the images to show here.

In a nutshell, I've noticed that my Tamron 28-75mm f/2.8 (which is pin sharp on my D700), looks a lot softer when I use it on my D90 body. The Tamron 17-50mm f/2.8, which I bought exclusively for use with the D90, is certainly sharp enough on either body.

I have a similar issue with prime lenses. Both of my 50mm primes (a Nikkor f/1.8D and the Sigma f/1.4 EX DG) seem to behave the same way - they don't look as sharp on the D90. It's not just due to the different sensor, as my sharp DX lenses (the 35mm f/1.8 for instance) really do look sharp on both bodies!?

So, is this a known issue :|? Do you lose something by only looking through the middle of the glass :thinking:? Is it a resolution thing, or a focus thing :shrug:? Am I just a paranoid pixel peeper :shake:?

What are your thoughts/experiences?
 
It does sound a bit pixel peepy but it's an Interesting one so I'll be keen to see what others say.

True - cropped bodies only use the 'better' central area of a FF lens so should be great - it may just be that some FF bodies are able to better record the IQ of a decent lens ?

I suspect there's also many variables - were your examples all done under the same conditions and camera settings ?
 
i know this won,t help but i always thought the centre was the sharpest on all lens, it,s usually the edges that are soft. :shrug:
 
I suspect it's to do with the pixel density - purely guessing here, but I reckon the D90 has a higher pixel density than the D700 when you take into account the sizes of each sensor and the pixel counts respectively.

Could also be to do with the strength of the AA filter; on my now sold 450D the 50mm f/1.4 was very sharp but on the 1Ds it is so pin sharp that pictures need virtually no sharpening.
 
Naboo32 said:
Do you lose something by only looking through the middle of the glass :thinking:?
Click to expand...

Exactly the opposite - FX lenses will always be at their best on a DX body because the body is only using the centre of the lens - the sharpest area in any lens.
 
Thanks for all of the replies so far :thumbs:.

As several of you have mentioned, I should be 'losing the soft corners' by putting FX lenses onto a crop sensor - that was what I expected, too :shrug:.

The first time that I noticed this was when I took some shots last Xmas, using a D40 (crop, 6MP sensor) and switching from the kit lens to the (full-frame) Tamron 28-75mm f/2.8. All of the shots with the kit lens were perfectly in focus and quite sharp. With the Tamron on, something was lost :(. I can't decide if it was that the lens wasn't focusing as well (more likely), or if it was accurately focused and just generally a lot softer. Those pics were quite disappointing and ended up being mostly deleted, too.

Anyway, I will try to get a few more 'real world' shots with both types of lenses and look again :|.

The reason that this matters to me, is that I want to use the D90 for traveling (leaving the heavier D700 at home) and I would like to have a faster, long lens than I the 55-200mm f/4-5.6, which I currently have for DX. If I could just carry both Tamrons, that would give me an effective range of 25-110mm (all at f/2.8), which would probably be enough. I'm just puzzled as to why the 28-75mm Tamron (the FX one) isn't as happy on the crop body as it is on the FF one :shrug:!?
 
Naboo32 said:
I'm just puzzled as to why the 28-75mm Tamron (the FX one) isn't as happy on the crop body as it is on the FF one :shrug:!?
Click to expand...

Did you mean it isn't as happy on the full frame body as on the crop? Or have I been reading the thread all wrong. Anyway,

trencheel303 said:
I suspect it's to do with the pixel density - purely guessing here, but I reckon the D90 has a higher pixel density than the D700 when you take into account the sizes of each sensor and the pixel counts respectively.
Click to expand...

I checked and what I suspected was confirmed. Roughly the same amount of megapixels, two very different sized sensors. Funny thing is, my situation was the opposite, I came from a 12MP cropper to an 11MP FF, and the FF camera definitely produces marginally sharper images, BUT, there are other variables at work here. it's not as if the lens magically changes its glass qualities on different cameras, so it might be that the Anti Aliasing thingy is weaker on the D90.
 
I think the AA on Nikon's is VERY week - the sharpness between Canon equiv and Nikon equiv I find amazing.
 
trencheel303 said:
Did you mean it isn't as happy on the full frame body as on the crop? Or have I been reading the thread all wrong. Anyway,



I checked and what I suspected was confirmed. Roughly the same amount of megapixels, two very different sized sensors. Funny thing is, my situation was the opposite, I came from a 12MP cropper to an 11MP FF, and the FF camera definitely produces marginally sharper images, BUT, there are other variables at work here. it's not as if the lens magically changes its glass qualities on different cameras, so it might be that the Anti Aliasing thingy is weaker on the D90.
Click to expand...

Well, yes, a few people here have misinterpreted what I wrote (probably due to my long-winded writing style ;)).

I know from experience that the two sensors (a big 12.1MP on the D700 and a smaller 12.3MP on the D90, IIRC :thinking:) produce slightly different looking images, so I was taking that into account. For my money, the D700 sensor has a much sweeter, more balanced look and can withstand greater degrees of sharpening and of course has far superior high ISO handling capabilities.

No, the issue here is that although all of my FX and DX lenses behave in a similar way on the D700 (albeit in 'crop mode' for the DX lenses), when I try out all of my FX and DX lenses on the D90, the DX lenses perform just as well as they do on the D700, but with all of the FX lenses (apart from the 300mm f/4, which is pin sharp on both), there seems to be a softness to the images - almost as if they are not properly in focus :shrug:!?

My suspicion, given that all of my test shots (and the Xmas party shots that I mentioned before) were taken indoors, is that it might be more to do with the AF struggling a bit (although I can't see why this should be an issue exclusively with FX lenses :thinking:). If the camera was front-focusing a fraction, then I wouldn't really know, as there is nothing in front of the subject to look at and compare :|.

Perhaps I should have done some more research before posting this and spreading confusion, but I just thought that there might be a 'scientific' explanation, which I hadn't heard before.
 
lawrenceots said:
I think the AA on Nikon's is VERY week - the sharpness between Canon equiv and Nikon equiv I find amazing.
Click to expand...

Possibly Lawrence, I've never owned a digital Canon (not a good one anyway), so I wouldn't know what to make of that.

Still, it doesn't change the fact that I can get great results from DX camera/DX lens, FX camera/FX lens and FX camera/DX lens, but the DX camera/FX lens combo just doesn't seem to be in the same league :shrug:.

If nobody else is experiencing this, then this whole thread is probably best ignored and I'll go away and play a bit more.

Thanks again to everyone who contributed :thumbs:.
 
Naboo32 said:
<snip>

Perhaps I should have done some more research before posting this and spreading confusion, but I just thought that there might be a 'scientific' explanation, which I hadn't heard before.
Click to expand...

There is an explanation - it's the lens MFT. The cropped format is asking the lens to perform at a higher resolution level, where the contrast is inevitably lower. That's what you're seeing. In terms of sensor format, basically bigger is better. End of story really. Full frame is better than crop, medium format is better than full frame and so it goes on. The question is where do you stop, and what is 'good enough' for what you want to do.

There are lots of variables going on, but the main thing is that for a given framing and given print size, the cropped format image has be enlarged more. In terms of area, the difference between full frame and crop format is about 2.5x (the number of pixels is relatively less important than their physical size - pixels are a bit of a red herring TBH). It is therefore calling on the lens to deliver a higher level of resolution, and as resolution goes up, so contrast goes down. That's an inevitable fact of physics and while lens designers can tweak it and the best lenses will hold on to a good level of contrast longer (mainly because they start out at a higher level, eg by using expensive high refractive index glass, fluorite etc) some decline is inevitable as resolution increases.

Here is an MTF graph (Modulation Transfer Function) for the Canon 50 1.8 http://www.usa.canon.com/consumer/controller?act=ModelInfoAct&fcategoryid=152&modelid=7306 Lots of interesting stuff going on there but for this purpose just look at the solid black lines on the left hand axis. The thick one shows the % of image contrast 'transfered' to the sensor at 10 lines per mm (the 'modulation' level) and sits at about 82%. That's very high. When the resolution is increased to 30lpmm, the thin solid black line drops to 58%, which is pretty average.

Just for comparison, here's a link to the same MTF graph for the Canon 300L 2.8, which is the sharpest lens I've ever seen. In a perfect lens, all the lines on the graph would run right along the top from left to right and one on top of the other, and this one comes close! http://www.usa.canon.com/consumer/controller?act=ModelInfoAct&fcategoryid=154&modelid=7317 You can immediately see why a lens of this fundamental quality takes even a 2x extender pretty well, because it's got so much high resolution and contrast potential.

Just on those graphs, and Nikon and Sigma publish something very similar, they were designed for film and the measures at only 10 and 30 lpmm are way lower than digital demands, eg 18mp from the Canon 7D translates to over 100lpmm. However, it's generally true to say that if you know what a lens will do at one resolution level, good, fair, poor etc, you can predict that it will also perform relatively in the same way at higher resolutions, just at a lower level.

There are a lot of other factors going on, but I think the main one is the lens MTF with lower contrast being delivered to the more demanding cropped sensor.
 
HoppyUK said:
There is an explanation - it's the lens MFT. The cropped format is asking the lens to perform at a higher resolution level, where the contrast is inevitably lower. That's what you're seeing. In terms of sensor format, basically bigger is better. End of story really. Full frame is better than crop, medium format is better than full frame and so it goes on. The question is where do you stop, and what is 'good enough' for what you want to do..
Click to expand...

That is exactly what I was trying to say up there ^^^^ but i can't explain stuff for saving me...
 
trencheel303 said:
That is exactly what I was trying to say up there ^^^^ but i can't explain stuff for saving me...
Click to expand...

Yes, you touched on it with reference to pixel density, but it's fundamentally the bigger full frame sensor working the lens less hard.

Pixel numbers and pixel density confuse the issue a bit I think. For example, if you take a full frame camera with less pixels than a crop camera (eg a 5DMkI vs 7D) you would still get better IQ from the full framer.
 
My experience is very similar. All my lenses appear sharper on 1Ds mkII compared to 40D. This is for 5 or so reasons.

1) They are decent lenses and the corners are still better than the centre of something like 18-55mm.

2) FF has a lower pixel density, hence requires lower resolving power of the lens. In my case FF has larger overall resolution.

3) AA filter is weaker on pro (FF cameras). This is particularly obvious with my Canons.

4) Focusing is better on pro cameras. 40D would produce slightly sharper image when focused using liveview.

5) Diffraction limit kicks in earlier on crop cameras.

You will find that DX lenses are specifically optimised for the 1.6x cropped area. Nevertheless, Tokina 12-24mm is sharper on 1Ds at 19-24mm (covers full frame).
 
HoppyUK said:
There is an explanation - it's the lens MFT. The cropped format is asking the lens to perform at a higher resolution level, where the contrast is inevitably lower. That's what you're seeing. In terms of sensor format, basically bigger is better. End of story really. Full frame is better than crop, medium format is better than full frame and so it goes on. The question is where do you stop, and what is 'good enough' for what you want to do.

There are lots of variables going on, but the main thing is that for a given framing and given print size, the cropped format image has be enlarged more. In terms of area, the difference between full frame and crop format is about 2.5x (the number of pixels is relatively less important than their physical size - pixels are a bit of a red herring TBH). It is therefore calling on the lens to deliver a higher level of resolution, and as resolution goes up, so contrast goes down. That's an inevitable fact of physics and while lens designers can tweak it and the best lenses will hold on to a good level of contrast longer (mainly because they start out at a higher level, eg by using expensive high refractive index glass, fluorite etc) some decline is inevitable as resolution increases.

Here is an MTF graph (Modulation Transfer Function) for the Canon 50 1.8 http://www.usa.canon.com/consumer/controller?act=ModelInfoAct&fcategoryid=152&modelid=7306 Lots of interesting stuff going on there but for this purpose just look at the solid black lines on the left hand axis. The thick one shows the % of image contrast 'transfered' to the sensor at 10 lines per mm (the 'modulation' level) and sits at about 82%. That's very high. When the resolution is increased to 30lpmm, the thin solid black line drops to 58%, which is pretty average.

Just for comparison, here's a link to the same MTF graph for the Canon 300L 2.8, which is the sharpest lens I've ever seen. In a perfect lens, all the lines on the graph would run right along the top from left to right and one on top of the other, and this one comes close! http://www.usa.canon.com/consumer/controller?act=ModelInfoAct&fcategoryid=154&modelid=7317 You can immediately see why a lens of this fundamental quality takes even a 2x extender pretty well, because it's got so much high resolution and contrast potential.

Just on those graphs, and Nikon and Sigma publish something very similar, they were designed for film and the measures at only 10 and 30 lpmm are way lower than digital demands, eg 18mp from the Canon 7D translates to over 100lpmm. However, it's generally true to say that if you know what a lens will do at one resolution level, good, fair, poor etc, you can predict that it will also perform relatively in the same way at higher resolutions, just at a lower level.

There are a lot of other factors going on, but I think the main one is the lens MTF with lower contrast being delivered to the more demanding cropped sensor.
Click to expand...

Yet another award-winningly, precise yet understandable post by HoppyUK (Richard, it really is time that you wrote that book we talked about before :D)! Thank you very much.

This does actually explain a lot and seems to confirm my suspicions. Although some people might see the differences which I've experienced and say, "That's not an issue for me", I'm not 'some people' ;). No, I think that the message is clear here - I shouldn't use my less sharp FX lenses on my DX body.

It perhaps also explains why my sharpest lens (I think), the Nikkor 300mm f/4, looks just as sharp and contrasty on either body (although the images from the D90's sensor have a more 'pixelated' look, IMO :|).

Knowing that there is actually one possible reason why my FX lenses appear to perform worse on a DX sensor is very useful to me. I was going through my lenses this week, looking to weed out some of the less used ones, but was testing most of them on the D90 and was a little bit surprised to see that both my 'nifty fifty' and my Sigma 50mm f/1.4 were only producing the same levels of sharpness (and marginally less contrast :() as my DX Tamron zoom. On the D700, they really sing (especially the Sigma :))!

On another topic, the Canon 300mm f/2.8 in your link looks like an amazing piece of kit :naughty:. I've seen a few pics posted on TP which have the same kind of 3D quality that the sample image in that link had, and most of those pics were from high end Canons with prime "L" glass. Very impressive stuff!

Thanks again for the explanation and the reassurance :thumbs:.
 
daugirdas said:
My experience is very similar. All my lenses appear sharper on 1Ds mkII compared to 40D. This is for 5 or so reasons.

1) They are decent lenses and the corners are still better than the centre of something like 18-55mm.

2) FF has a lower pixel density, hence requires lower resolving power of the lens. In my case FF has larger overall resolution.

3) AA filter is weaker on pro (FF cameras). This is particularly obvious with my Canons.

4) Focusing is better on pro cameras. 40D would produce slightly sharper image when focused using liveview.

5) Diffraction limit kicks in earlier on crop cameras.

You will find that DX lenses are specifically optimised for the 1.6x cropped area. Nevertheless, Tokina 12-24mm is sharper on 1Ds at 19-24mm (covers full frame).
Click to expand...

Thanks for your input, daugirdas :). It's nice to know that I'm not alone with my suspicions ;).
 
Naboo32 said:
Yet another award-winningly, precise yet understandable post by HoppyUK (Richard, it really is time that you wrote that book we talked about before :D)! Thank you very much.

This does actually explain a lot and seems to confirm my suspicions. Although some people might see the differences which I've experienced and say, "That's not an issue for me", I'm not 'some people' ;). No, I think that the message is clear here - I shouldn't use my less sharp FX lenses on my DX body.

It perhaps also explains why my sharpest lens (I think), the Nikkor 300mm f/4, looks just as sharp and contrasty on either body (although the images from the D90's sensor have a more 'pixelated' look, IMO :|).

Knowing that there is actually one possible reason why my FX lenses appear to perform worse on a DX sensor is very useful to me. I was going through my lenses this week, looking to weed out some of the less used ones, but was testing most of them on the D90 and was a little bit surprised to see that both my 'nifty fifty' and my Sigma 50mm f/1.4 were only producing the same levels of sharpness (and marginally less contrast :() as my DX Tamron zoom. On the D700, they really sing (especially the Sigma :))!

On another topic, the Canon 300mm f/2.8 in your link looks like an amazing piece of kit :naughty:. I've seen a few pics posted on TP which have the same kind of 3D quality that the sample image in that link had, and most of those pics were from high end Canons with prime "L" glass. Very impressive stuff!

Thanks again for the explanation and the reassurance :thumbs:.
Click to expand...

Thanks for those kind comments Andy. You're most welcome :)
 
You must log in or register to reply here.
Back
Top