Canon EF 24-70mm f4L IS USM

[jeffa4444]

Going by the recommended lenses for the Canon 5DS that Canon published I purchased the EF24-70mm f4L IS USM for mainly landscape use I didnt need f2.8.
This was back in September 2015 and Ive never been entirely blown away by the results so whilst conducting other tests yesterday using one of Canon Cine lenses I mounted this lens on the 5DS and shot some tests on an Essar Test Chart Sphere using a CIPA High resolution chart. The Sphere provides complete even field illumination and the chart is back-lit. Test are carried out in a dark room with the Essar Sphere providing the only light. For video we can show the results on a cinema sized screen (55ft diagonal) in 4K. For this particular test the images were processed in Light Room and viewed on a 27" 4K monitor again in a darkened room.
The test showed two problems a. image shift and b. softness even centre frame at its optimum f8 on a Canon 5DS full resolution raw file at 50mm. The results were marginally better at 70mm and 24mm but certainly the limiting factor in accessing the full resolution the 5DS can acheive. CA were not well controlled.
Maybe this is a rouge lens but Photozone. de had exactly the same issues when they tested the lens a number of years ago so why Canon thinks this lens is suitable for the 5DS / r is beyond me.

 

[Monchoon]

Thanks for the info. Although this doesn't seem to affect me I do enjoy the lens. The only other that I have besides the 24-105 f4 in this range that I have is the Tamron f2.8.

I do have to let two of them go just unsure of which.

 

[j-nord]

I believe the DXO p-mpix measurement of a lens when mounted on a 5DSR is the best way to determine if it's sharp enough. On 20-24mpix bodies, 2 lenses may score the same p-mpix but when mounted on a 5DSR they do not score the same, one is sharper than the other, sometimes by quite a lot. If you look at the 24-70 f4 IS p-mpix, in no way is it sharp enough for use on a 5DSR.

 

[jd7]

:-[

I believe the DXO p-mpix measurement of a lens when mounted on a 5DSR is the best way to determine if it's sharp enough. On 20-24mpix bodies, 2 lenses may score the same p-mpix but when mounted on a 5DSR they do not score the same, one is sharper than the other, sometimes by quite a lot. If you look at the 24-70 f4 IS p-mpix, in no way is it sharp enough for use on a 5DSR.

I have to say I wonder about some of the DxO measurements.

According to DxO, the 24-70 4L IS gets 14 Mpix on my 6D (increasing by almost 50% on a 5DsR). For comparison, 14 Mpix is also the result for a Sigma 50 1.4 EX on a 6D - and that lens is certainly not usually celebrated for sharpness! (Note - the Sigma 50 1.4 EX is Sigma's old model, before the Art. The Art is a different ballgame.)

Further, DxO's Mpix results for the 16-35 4L IS are about the same as for the 24-70 4L IS - on both a 6D and 5DsR ... but the 16-35 4L IS is almost universally praised for its sharpness.

It just seems odd to me that the 24-70 4L IS, the 16-35 4L IS and the Sigma 50 1.4 EX get such similar results.

One more comparison - DxO's results for the 24-70L II say it does do better than the 24-70 4L IS - but perhaps not by that much on a 6D, given the result is only 3 Mpix higher. On a 5DsR, the gap would be much wider - 32 Mpix v 21 Mpix for the 24-70L II over the 24-70 4L IS.

In the end, I'm left not feeling sure what to make of DxO's results ... although if they are correct about the 24-70L II v 24-70 4L IS, it perhaps might explain why I haven't felt blown away by the 24-70L II compared with the 24-70 4L IS when I have shot them on the 6D, while others say the 24-70L II is streets ahead. (That said, I should also say my time with a 24-70L II has been very limited, sadly; and there is no doubt the 24-70L II is good.)

 

[jd7]

Maybe this is a rouge lens

Have you sent it in to Canon and got them to have a look at it?

Edit: Also, at the risk of raising something you may consider obvious, if you have IS on, are you making sure to give the IS system a chance to settle after you activate it before taking the shot? Maybe you are shooting with IS off, so that's not an issue?

 

[j-nord]

:-[

I believe the DXO p-mpix measurement of a lens when mounted on a 5DSR is the best way to determine if it's sharp enough. On 20-24mpix bodies, 2 lenses may score the same p-mpix but when mounted on a 5DSR they do not score the same, one is sharper than the other, sometimes by quite a lot. If you look at the 24-70 f4 IS p-mpix, in no way is it sharp enough for use on a 5DSR.

I have to say I wonder about some of the DxO measurements.

According to DxO, the 24-70 4L IS gets 14 Mpix on my 6D (increasing by almost 50% on a 5DsR). For comparison, 14 Mpix is also the result for a Sigma 50 1.4 EX on a 6D - and that lens is certainly not usually celebrated for sharpness! (Note - the Sigma 50 1.4 EX is Sigma's old model, before the Art. The Art is a different ballgame.)

Further, DxO's Mpix results for the 16-35 4L IS are about the same as for the 24-70 4L IS - on both a 6D and 5DsR ... but the 16-35 4L IS is almost universally praised for its sharpness.

It just seems odd to me that the 24-70 4L IS, the 16-35 4L IS and the Sigma 50 1.4 EX get such similar results.

One more comparison - DxO's results for the 24-70L II say it does do better than the 24-70 4L IS - but perhaps not by that much on a 6D, given the result is only 3 Mpix higher. On a 5DsR, the gap would be much wider - 32 Mpix v 21 Mpix for the 24-70L II over the 24-70 4L IS.

In the end, I'm left not feeling sure what to make of DxO's results ... although if they are correct about the 24-70L II v 24-70 4L IS, it perhaps might explain why I haven't felt blown away by the 24-70L II compared with the 24-70 4L IS when I have shot them on the 6D, while others say the 24-70L II is streets ahead. (That said, I should also say my time with a 24-70L II has been very limited, sadly; and there is no doubt the 24-70L II is good.)

Ive owned quite a few different lenses in the last few years, I think their measurements for p-mpix is fairly accurate from what I've seen. For example I recently picked up a 135L. It's much sharper than the 24-70 f4 IS and 300 f4 IS. I saw it immediately, I didn't think it would be as obvious as it was. I also side by side compared the 24mm f2.8 (non-IS) with the 24-70 f4. I thought the 24 2.8 was slightly sharper before I knew DXO agreed. I side by side compared my 7D and 6D with several of my lenses including 17-40L and 70-300L. I noticed a considerable hit to detail with the 7D, DXO agrees. I think their individual metrics are pretty good, their reviews and 'overall' scores are pretty useless though. It's always good to get other opinions and compare other tests though.

I'm a bit of a pixel peeper though, my tastes continually move to the sharper end of the scale. I'm currently considering replacing my 24-70 f4 with the 24mm f2.8 IS since I got the 135L which can work as an alternative to 70 for now. I don't see the 24-70ii as worth the money until they throw IS in it.

 

[Larsskv]

:-[

I believe the DXO p-mpix measurement of a lens when mounted on a 5DSR is the best way to determine if it's sharp enough. On 20-24mpix bodies, 2 lenses may score the same p-mpix but when mounted on a 5DSR they do not score the same, one is sharper than the other, sometimes by quite a lot. If you look at the 24-70 f4 IS p-mpix, in no way is it sharp enough for use on a 5DSR.

I have to say I wonder about some of the DxO measurements.

According to DxO, the 24-70 4L IS gets 14 Mpix on my 6D (increasing by almost 50% on a 5DsR). For comparison, 14 Mpix is also the result for a Sigma 50 1.4 EX on a 6D - and that lens is certainly not usually celebrated for sharpness! (Note - the Sigma 50 1.4 EX is Sigma's old model, before the Art. The Art is a different ballgame.)

Further, DxO's Mpix results for the 16-35 4L IS are about the same as for the 24-70 4L IS - on both a 6D and 5DsR ... but the 16-35 4L IS is almost universally praised for its sharpness.

It just seems odd to me that the 24-70 4L IS, the 16-35 4L IS and the Sigma 50 1.4 EX get such similar results.

One more comparison - DxO's results for the 24-70L II say it does do better than the 24-70 4L IS - but perhaps not by that much on a 6D, given the result is only 3 Mpix higher. On a 5DsR, the gap would be much wider - 32 Mpix v 21 Mpix for the 24-70L II over the 24-70 4L IS.

In the end, I'm left not feeling sure what to make of DxO's results ... although if they are correct about the 24-70L II v 24-70 4L IS, it perhaps might explain why I haven't felt blown away by the 24-70L II compared with the 24-70 4L IS when I have shot them on the 6D, while others say the 24-70L II is streets ahead. (That said, I should also say my time with a 24-70L II has been very limited, sadly; and there is no doubt the 24-70L II is good.)

Ive owned quite a few different lenses in the last few years, I think their measurements for p-mpix is fairly accurate from what I've seen. For example I recently picked up a 135L. It's much sharper than the 24-70 f4 IS and 300 f4 IS. I saw it immediately, I didn't think it would be as obvious as it was. I also side by side compared the 24mm f2.8 (non-IS) with the 24-70 f4. I thought the 24 2.8 was slightly sharper before I knew DXO agreed. I side by side compared my 7D and 6D with several of my lenses including 17-40L and 70-300L. I noticed a considerable hit to detail with the 7D, DXO agrees. I think their individual metrics are pretty good, their reviews and 'overall' scores are pretty useless though. It's always good to get other opinions and compare other tests though.

I'm a bit of a pixel peeper though, my tastes continually move to the sharper end of the scale. I'm currently considering replacing my 24-70 f4 with the 24mm f2.8 IS since I got the 135L which can work as an alternative to 70 for now. I don't see the 24-70ii as worth the money until they throw IS in it.

You should be aware that the way DXO are testing lenses, the slower lenses they test, the worse are the results. F/4 lenses and slower seems to perform especially bad in DXO testing. This is because they measure sharpness from pictures taken in low light (150 lux) at the fixed shutter speed of 1/60 sec. This results in higher ISO´s, the slower the lens is, and less again sharpness. This is why, according to DXO, the nifty fifty (50mm f/1.8) is sharper than the 600mm f/4 L.

DXO can to be very misleading, if you dont study how they achieve their results.

 

[jd7]

You should be aware that the way DXO are testing lenses, the slower lenses they test, the worse are the results. F/4 lenses and slower seems to perform especially bad in DXO testing. This is because they measure sharpness from pictures taken in low light (150 lux) at the fixed shutter speed of 1/60 sec. This results in higher ISO´s, the slower the lens is, and less again sharpness. This is why, according to DXO, the nifty fifty (50mm f/1.8) is sharper than the 600mm f/4 L.

DXO can to be very misleading, if you dont study how they achieve their results.

Really?!? That seems insane. I know Dxo is testing camera plus lens as a system, and that that means the ability of the camera sensor to resolve detail therefore makes a difference to the mpix result, but I didn't know they let ISO vary too. That means that for slower lenses, the sensor's noise performance could be having a substantial impact.

That would explain the mpix results mentioned above, and similar results for the 100-400L II, which doesn't get an especially good mpix result even though most people seem to think it is very sharp.

 

[Larsskv]

You should be aware that the way DXO are testing lenses, the slower lenses they test, the worse are the results. F/4 lenses and slower seems to perform especially bad in DXO testing. This is because they measure sharpness from pictures taken in low light (150 lux) at the fixed shutter speed of 1/60 sec. This results in higher ISO´s, the slower the lens is, and less again sharpness. This is why, according to DXO, the nifty fifty (50mm f/1.8) is sharper than the 600mm f/4 L.

DXO can to be very misleading, if you dont study how they achieve their results.

Really?!? That seems insane. I know Dxo is testing camera plus lens as a system, and that that means the ability of the camera sensor to resolve detail therefore makes a difference to the mpix result, but I didn't know they let ISO vary too. That means that for slower lenses, the sensor's noise performance could be having a substantial impact.

That would explain the mpix results mentioned above, and similar results for the 100-400L II, which doesn't get an especially good mpix result even though most people seem to think it is very sharp.

Yes, really. I'm on my phone, so it's a little hassle to find DXO's own explanation about this, but it's there. One should also keep in mind that DXO results don't tell you how a lens perform when stopped down. For much of my daylight shooting, that's more important than wide open performance.

 

[j-nord]

You should be aware that the way DXO are testing lenses, the slower lenses they test, the worse are the results. F/4 lenses and slower seems to perform especially bad in DXO testing. This is because they measure sharpness from pictures taken in low light (150 lux) at the fixed shutter speed of 1/60 sec. This results in higher ISO´s, the slower the lens is, and less again sharpness. This is why, according to DXO, the nifty fifty (50mm f/1.8) is sharper than the 600mm f/4 L.

DXO can to be very misleading, if you dont study how they achieve their results.

Really?!? That seems insane. I know Dxo is testing camera plus lens as a system, and that that means the ability of the camera sensor to resolve detail therefore makes a difference to the mpix result, but I didn't know they let ISO vary too. That means that for slower lenses, the sensor's noise performance could be having a substantial impact.

That would explain the mpix results mentioned above, and similar results for the 100-400L II, which doesn't get an especially good mpix result even though most people seem to think it is very sharp.

Yes, really. I'm on my phone, so it's a little hassle to find DXO's own explanation about this, but it's there. One should also keep in mind that DXO results don't tell you how a lens perform when stopped down. For much of my daylight shooting, that's more important than wide open performance.

Thanks for pointing that out, it took me quite a while to dig around and find the exact statement. I was somewhat aware of it but I never read it for myself. They actually average the p-mpix of the lens across every f-stop and focal length(in the case of zooms). Since zooms typically have a weak spot or 2 and are often f4+ they can take a decent hit in p-mpix. @50mm + f4 on the 24-70 f4 IS is pretty soft and DXOs charts show this, certainly that will impact the overall p-mpix, maybe more than it should. Also, DXOs charts use an upper limit of >12pmix (bright green) which means they are pretty useless for comparing very sharp lenses or results on a high m-pix body.

 

[jeffa4444]

Personally from a professional perspective I think DXOMark testing methods are flawed and Ive pointed this out to them.
When I tested this ONE example of the EF 24-70mm f4L IS USM lens it was tested with IS turned off because it was mounted on a test stand which is screwed to the floor. My results mirrored those of Photozone.de with the two lenses being years apart in manufacture. All the other lenses I tested that day (EF 16-35mm f4L IS USM, EF 100mm f2.8L IS USM, Canon CN-24mm T1.5) showed just how badly this lens resolved now the Canon CN-24mm T1.5 you would expect to be much better given the cost differential but the comparison to the EF16-35mm f4L IS USM you would not expect such a huge difference but there was.