Canon Lenses. EF or EF-S?

[acidrain]

Looking to get a new zoom lens as my current one doesn't have any IS which is a problem when doing handheld, which I do most the time.

So now looking at lenses I have the option of EF or EF-S ( my camera being a crop sensor at the moment ). One big motivator I've read for EF is that if you do ever want to go full frame, then you don't need new glass but looking at the cheapest canon full frame, which I believe is the 6D, going at 20.5k, I do not foresee ever going full frame then.

The next thing is that apparently EF lenses have better image quality than the EF-S variety but when looking at the price, the EF-S is quite a bit cheaper. Now the question is, does the EF indeed have better image quality, and if so, is the quality difference so much as to warrant the price?

Thanks,

 

[RanzB]

What are you primarily using your camera for?

 

[acidrain]

Primarily landscape and wildlife... but in reality anything which catches my eye.

 

[RanzB]

Primarily landscape and wildlife... but in reality anything which catches my eye.

I meant do you take photographs as a hobby? If so, the image quality or the lack thereof is only an issue to you.

 

[acidrain]

Hobby yes.

Basically I just want to know if i took a shot with an EF lens on a subject and then took the same exact shot with the EF-S lens equivalent, would there be a noticeable difference in quality/sharpness.

 

[bwana]

Hobby yes.

Basically I just want to know if i took a shot with an EF lens on a subject and then took the same exact shot with the EF-S lens equivalent, would there be a noticeable difference in quality/sharpness.

It really depends on the lenses in question. EF lenses aren't necessarily any superior than EF-s lenses until you get to the L series - but even then the waters can get muddied.

 

[Progenix Oj101]

Which lenses are you thinking of? Something like the 17-55mm f/2.8 is so good that pretty much the only thing stopping it from getting an L in its name is Canon doesn't make L series EF-S lenses. It is better than many EF lenses. Conversely, the EF 17-40mm f/4L (flaring) and 24-105mm f/4L (soft) are examples of poor EF lenses that carry the L moniker.

 

[koffiejunkie]

So now looking at lenses I have the option of EF or EF-S ( my camera being a crop sensor at the moment ). One big motivator I've read for EF is that if you do ever want to go full frame, then you don't need new glass but looking at the cheapest canon full frame, which I believe is the 6D, going at 20.5k, I do not foresee ever going full frame then.

The next thing is that apparently EF lenses have better image quality than the EF-S variety but when looking at the price, the EF-S is quite a bit cheaper. Now the question is, does the EF indeed have better image quality, and if so, is the quality difference so much as to warrant the price?

There are some great EF-S lenses and some OK EF lenses and everything above, below and in-between. I'll repeat what I've said here many times before: Unless you know you are buying a full frame body any day now, don't bother constraining yourself. Even the EF-S 55-250mm II (or something like that) that's made of all-plastic will sell 2nd-hand, and it's good lens (particulary considering the cost).

That said, the EF 70-300mm f/4-5.6 IS is a pretty decent lens, and not too expensive.

Which lenses are you thinking of? Something like the 17-55mm f/2.8 is so good that pretty much the only thing stopping it from getting an L in its name is Canon doesn't make L series EF-S lenses. It is better than many EF lenses. Conversely, the EF 17-40mm f/4L (flaring) and 24-105mm f/4L (soft) are examples of poor EF lenses that carry the L moniker.

I've owned two of these (first one got nicked) - they're not soft on any camera I've owned. It's maybe not as sharp as some other lenses, but we'll have to go to much higher pixel densities than we have in the Canon system today, before we can start calling this lens soft. A more reasonable cricism of it is that it has fairly heavy barrel distortion (on FF) and a fair bit of CA on high contrasts - fortunately both are easily corrected in post.

 

[acidrain]

Thanks for all the info thus far guys

It really depends on the lenses in question. EF lenses aren't necessarily any superior than EF-s lenses until you get to the L series - but even then the waters can get muddied.

These are the 2 lenses I'm looking at:

Canon EF 70-300mm f/4-5.6 IS USM Lens - R5000

OR

Canon EF-S 55-250mm f/4-5.6 IS II Lens - R1600

Ok, while the the EF-S isn't exactly the same, it is R3400 cheaper. Yes i Do see the one is USM and the other not but my understanding of USM is really only a quieter zoom, quicker which isn't really a necessity for me as I mostly do manual zooming.

 

[Progenix Oj101]

There are some great EF-S lenses and some OK EF lenses and everything above, below and in-between. I'll repeat what I've said here many times before: Unless you know you are buying a full frame body any day now, don't bother constraining yourself. Even the EF-S 55-250mm II (or something like that) that's made of all-plastic will sell 2nd-hand, and it's good lens (particulary considering the cost).

That said, the EF 70-300mm f/4-5.6 IS is a pretty decent lens, and not too expensive.



I've owned two of these (first one got nicked) - they're not soft on any camera I've owned. It's maybe not as sharp as some other lenses, but we'll have to go to much higher pixel densities than we have in the Canon system today, before we can start calling this lens soft. A more reasonable cricism of it is that it has fairly heavy barrel distortion (on FF) and a fair bit of CA on high contrasts - fortunately both are easily corrected in post.

I played with one which was visibly soft wide open without pixel peeping when compared to a 70-200. I'm not alone, I just did a Google search for "Canon 24-105 soft" - you'll you'll find plenty of discussions/forums with people complaining about how soft the lens is.

 

[bwana]

Thanks for all the info thus far guys



These are the 2 lenses I'm looking at:

Canon EF 70-300mm f/4-5.6 IS USM Lens - R5000

OR

Canon EF-S 55-250mm f/4-5.6 IS II Lens - R1600

Ok, while the the EF-S isn't exactly the same, it is R3400 cheaper. Yes i Do see the one is USM and the other not but my understanding of USM is really only a quieter zoom, quicker which isn't really a necessity for me as I mostly do manual zooming.

I don't use either of those however you should note that the 55-250 does not have full time manual focusing which means if you don't switch it over to manual you're going to stand a good chance of shredding the gears in there.

IIRC my first telephoto zoom was the 70-200 f/4 - great lens. Lightweight and sharp.

Despite the misgivings some seem to have my 24-105 is also a decent lens. I got one while I was debating whether or not to get my 24-70 f/2.8 sigma repaired for the umpteenth time… in the end I was happy enough with the 24-105 so I didn't.

 

[fvdbergh]

The next thing is that apparently EF lenses have better image quality than the EF-S variety but when looking at the price, the EF-S is quite a bit cheaper. Now the question is, does the EF indeed have better image quality, and if so, is the quality difference so much as to warrant the price?

All other things being equal, it helps to visualise what really happens when you use an EF lens on a crop body. The lens projects a picture that is wide enough to cover the 36 mm "full frame" sensor, but on a crop body you only use the centre 24 mm section of this image (and similarly in the height direction too).

This means that if an EF lens "looks soft in the corners" on a full frame body, it might look relatively better on the crop sensor because you are not using that part of the projected image (it is "cropped", i.e., the light is absorbed by the matte black coating inside the crop-sensor body).

So as you can see, this will depend entirely on the specific lens. Some EF lenses might only "go soft" right in the extreme corners --- these lenses may look great on a crop sensor body.

If you want to really get into this discussion, though, you also have to keep in mind the design parameters for the lens. Example: I have a Nikon 105 mm f/4 macro lens from about 1977 --- obviously this is a "full frame" lens. Being a macro lens, its sharpness is pretty even right up to the corners (of a crop sensor, since this is all I have). I also have a Sigma 17-50 mm f/2.8, which is a crop-only lens (i.e., EF-S). This lens is significantly sharper than the macro lens, especially in the centre of the frame (do not quote me on the extreme corners, cannot give a definitive answer there at the moment). In the centre, the actual resolution of the 105 mm macro lens is only about 75% that of the Sigma.

What this means in practice: For printing, say, and A4 photo, the magnification factor for the EF lens on a full-frame camera is only about 7.75 times (to scale the image on your sensor to an A4 size page). To print A4 from a crop-sensor camera you have to magnify the image by 11.62 times. This means that if you want the same sharpness in your printed A4 photo, the crop-sensor camera must have a lens that 1.5 times sharper (1.5 times the resolution).

Since a lens design is always a trade-off between many characteristics (centre sharpness vs. corner sharpness, lens distortion, chromatic aberration, etc.), it is reasonable to assume that not all lenses are designed for maximum sharpness. So how sharp must a lens be? Well, if you are going to print, say, A4, then you know how sharp the lens must be for a given sensor size. The trouble is that the target for "acceptable sharpness" on a full-frame lens would be 1.5 times lower than that of a crop-sensor lens to produce the same apparent sharpness in a printed A4 photo.

Bottom line: Theoretically you need a lens designed for a crop sensor body to achieve maximum sharpness (in the same size printed photo). In practice, there may well be many EF lenses that produce better overall image quality (specifically, sharper image corners) when used on a crop body, but this will probably only apply to the better quality lenses (think L). Sigma and Tamron do a pretty good job (Sigma more so) of designing good lenses for crop sensors, so I would definitely include their lenses in a comparison before blindly buying an EF L-class lens only on the assumption that this should produce better image quality.

 

[garyc]

These are the 2 lenses I'm looking at:

Canon EF 70-300mm f/4-5.6 IS USM Lens - R5000

OR

Canon EF-S 55-250mm f/4-5.6 IS II Lens - R1600

Here are some tests on those two lenses:

http://slrgear.com/reviews/showproduct.php/product/1370/cat/11
http://slrgear.com/reviews/showproduct.php/product/1440/cat/11

Their conclusion seems to be that they are good lenses for the money but not great.

 

[Progenix Oj101]

How about I sell you my 55-250 IS II for R1200 including courier?

Link - http://carbonite.co.za/f42/canon-55-250mm-4-5-6-lens-79248-post566728/

 

[swift412]

Test chart comparison: http://www.the-digital-picture.com/...meraComp=736&SampleComp=0&FLIComp=3&APIComp=3

Stopped down to f/8, they're pretty close on the wide end. On the long end, the 70-300 is quite a bit better in the corners. But then again, you aren't going to be using the lens to take photos of test charts, are you?

 

[koffiejunkie]

I played with one which was visibly soft wide open without pixel peeping when compared to a 70-200. I'm not alone, I just did a Google search for "Canon 24-105 soft" - you'll you'll find plenty of discussions/forums with people complaining about how soft the lens is.

OK, let's be clear:

1. Every zoom lens Canon ever made (with the possible exception of the new 24-70 f/2.8) is soft compared to a (Canon) 70-200. But then, the 70-200 f/4 is "soft" compared to the 70-200 f/2.8 II. It doesn't mean much.

2. Internet = vocal (and often misinformed and opinionated) minority. Google any lens and "soft" and you'll find someone saying it's soft (even the aforementioned 70-200mm f/2.8 II). Like I said, I've owend two (and I've rented one on a number of occasions when I didn't own one). All have been sharp, even wide open. As sharp as any of Canon's 70-200 models? Of course not. But to call them "poor" EF lenses, as you did, is frankly nonsense. If the one you played with is soft, it means that one copy is faulty/needs adjustment - not that that lens in general is soft.

 

[BigAl-sa]

...

What this means in practice: For printing, say, and A4 photo, the magnification factor for the EF lens on a full-frame camera is only about 7.75 times (to scale the image on your sensor to an A4 size page). To print A4 from a crop-sensor camera you have to magnify the image by 11.62 times. This means that if you want the same sharpness in your printed A4 photo, the crop-sensor camera must have a lens that 1.5 times sharper (1.5 times the resolution).

...

Not sure what you're on about here. You're trying to gloss over ppi of the sensor, dpi of the print and lpi of the lens in a short paragraph. The images from most modern dslr cameras have more than enough pixels to fill an a4 page at a satisfactory resolution.

 

[bwana]

Not sure what you're on about here. You're trying to gloss over ppi of the sensor, dpi of the print and lpi of the lens in a short paragraph. The images from most modern dslr cameras have more than enough pixels to fill an a4 page at a satisfactory resolution.

At 300dpi you want roughly 8.6mp to print a 8" by 12" (302mm x 203mm) "A4", which is what you typical lab spits out.

 

[fvdbergh]

Not sure what you're on about here. You're trying to gloss over ppi of the sensor, dpi of the print and lpi of the lens in a short paragraph. The images from most modern dslr cameras have more than enough pixels to fill an a4 page at a satisfactory resolution.

Perceived image sharpness (or equivalently resolution) is affected strongly by the lens, especially on modern sensors that have sufficiently small photosites so that they are effectively limited by the lens resolution. (So I agree with you on that point. Read on to see why I simplified the whole thing down to magnification factors).

Norman Koren (the person behind Imatest) has this to say: "Since a line pair corresponds to two lines of this size, the corresponding spatial frequency is 6.88 lp/mm or 175 lp/inch. Assume now that the image was printed from a 35mm frame enlarged 8x. The corresponding spatial frequency on the film would be 55 lp/mm. " http://www.normankoren.com/Tutorials/MTF.html

To paraphrase Norman: 20/20 vision can resolve down to about 6.88 lp/mm (line pairs per mm) at roughly the distance and field of view as would be used when looking at an A4 printed photo. Since a line pair requires two pixels, this works out to about 4086 pixels on the long edge, or about 11 MP on a 3:2 aspect ratio, and corresponds to roughly 350 dpi in print.

Working backwards, we take our 6.88 lp/mm resolution in print across the A4 page (297 mm) and map that to our sensor (36 mm wide for full-frame), which gives us a magnification of 1/8.25, or a required lens resolution of 56.76 lp/mm (similar to Norman's figure, but adapted to A4). In other words, we must resolve ~57 lp/mm on the sensor+lens system to obtain sharpness that matches 20/20 vision.

By similar reasoning, we can map 6.88 lp/mm to an APS-C (DX, 1.5 crop factor) at 24 mm wide (roughly) using a magnification of 1/12.375, so we need about 85 lp/mm on the sensor+lens system.

In both cases, this translates to about 4100 pixels across the long edge (i.e., our hypothetical 11 MP sensor). We can model this more accurately by going into demosaicing and OLPF (for sensors that have them), but these effects can be included if we measure MTF50 using something like Imatest, which evaluates the resolution of the imaging system (which includes both lens and sensor MTFs).

You will agree that both the sensor and print resolutions requirements are sufficiently addressed using current sensors and printers (11 MP sensor, 350 dpi printing). For this reason, I left those effects out of the discussion to focus on the part relevant to the original discussion: A lens that produces a sharp A4 print from a full-frame camera may produce a less sharp print when used on a crop sensor.

The remaining variable that controls this is the magnification factor (8.25x for full-frame, 12.375x for crop). The lens's resolution (say in the centre of the image to keep things simpler) is fixed: a given lens may give us only 44 lp/mm (using my ancient 105 mm f/4 macro lens as example). This translates to 5.33 lp/mm in A4 print when used on a full-frame sensor, but only 3.55 lp/mm when used to make an A4 print from a crop sensor. (keep in mind that because of the oversampling inherent to the slanted edge method used to measure MTF in Imatest, or MTF Mapper, the lens' resolution can be measured accurately even if it exceeds the apparent resolution of the sensor used to measure the lens with).

Of course, the 20/20 vision resolution limit is a fairly high standard; my guess would be that about 3.5 lp/mm or more would already start to look decent.