Quality control issues with the 24-70 L II?

[heptagon]

These two lenses should look exactly the same:
http://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx?Lens=787&Camera=453&Sample=0&FLI=4&API=0&LensComp=787&CameraComp=453&SampleComp=1&FLIComp=4&APIComp=0

At the short end one is better, at the long end the other. But why? They should both look best for all focal lengths and never have passed the quality control.

Maybe the optical design has such low tolerances that Canon can't make it work properly and this was the reason for the long delays.

[Invertalon]

I find one of the samples fishy to be honest. Take a look at this:

http://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx?Lens=787&Camera=453&Sample=1&FLI=2&API=0&LensComp=787&CameraComp=453&SampleComp=1&FLIComp=2&APIComp=2

Same lens, same focal length, one a stop down from the other. The f/4 should be sharper. Why is it so soft compared to f/2.8?

Something seems off.

[LetTheRightLensIn]

I find one of the samples fishy to be honest. Take a look at this:

http://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx?Lens=787&Camera=453&Sample=1&FLI=2&API=0&LensComp=787&CameraComp=453&SampleComp=1&FLIComp=2&APIComp=2

Same lens, same focal length, one a stop down from the other. The f/4 should be sharper. Why is it so soft compared to f/2.8?

Something seems off.

I have to say that I've long said that my results often don't match what TDP gets. Lens testing is really, reallly tricky. You have to be so careful and take many repeated attempts at each shot and then picking the best from that. I've done it a few times and it kinda drives you crazy. I have noticed some odd things on other lenses, like you say, where it will look like an 8 wide open and then suddenly like a 6 the next stop down and then back to an 8 the next stop and then to a 9. Doesn't make sense. Seems like bungled focusing to me.

It's also weird that every single tamron lens they test does horribly. Why? All three tamrons that I have owned have done much better.

That said, there can be copy variation.

[Marsu42]

At the short end one is better, at the long end the other. But why? They should both look best for all focal lengths and never have passed the quality control.

Maybe a good idea not to get a lens from the first batch - given the high demand and high revenue they'll probably sell every lens they've got and lower the qc standards accordingly?

Concerning tdp: The soft lens @70mm certainly shouldn't have been sold, @24mm I cannot see that much of a difference - lesson learned: paying $2000+ for a lens doesn't give you the certainty to have a perfect lens.

The same thing occurs on their 70-300L samples, I was very insecure how to manage to get a "good" copy, but the unfortunate truth is that it's near impossible to tell for a laymen w/o specialized equipment. The only thing you can do is to match the long end of the 24-70 against a known good copy of a 70-200.

[heptagon]

Something seems off.

True, there was some image misordering or something really strange about lens 2...

[bdunbar79]

At the short end one is better, at the long end the other. But why? They should both look best for all focal lengths and never have passed the quality control.

Maybe a good idea not to get a lens from the first batch - given the high demand and high revenue they'll probably sell every lens they've got and lower the qc standards accordingly?

Concerning tdp: The soft lens @70mm certainly shouldn't have been sold, @24mm I cannot see that much of a difference - lesson learned: paying $2000+ for a lens doesn't give you the certainty to have a perfect lens.

The same thing occurs on their 70-300L samples, I was very insecure how to manage to get a "good" copy, but the unfortunate truth is that it's near impossible to tell for a laymen w/o specialized equipment. The only thing you can do is to match the long end of the 24-70 against a known good copy of a 70-200.

Companies don't lower QC standards.  They are written in company SOP and followed like a recipe word for word.  That doesn't make any sense at all.  Have you ever worked in a QC department?

[Marsu42]

Companies don't lower QC standards.  They are written in company SOP and followed like a recipe word for word.  That doesn't make any sense at all.  Have you ever worked in a QC department?

No, I did not. But I wonder how you can say with absolute determination that qc standards are never lowered no matter what the company or market situation is, nowhere, globally. That doesn't make sense to me, if I may say so. But it might very well be that my speculation is wrong and the first samples of 24-70ii will be just like the ones later on.

[bdunbar79]

Companies don't lower QC standards.  They are written in company SOP and followed like a recipe word for word.  That doesn't make any sense at all.  Have you ever worked in a QC department?

No, I did not. But I wonder how you can say with absolute determination that qc standards are never lowered no matter what the company or market situation is, nowhere, globally. That doesn't make sense to me, if I may say so. But it might very well be that my speculation is wrong and the first samples of 24-70ii will be just like the ones later on.

Well, no, I can't do that.  However, for a company as big as Canon, it would be unrealistic that they could take the time to purposefully lower QC standards when they are so large and handle so many manufacturing units.  You'd be surprised at how robotic some of these people are doing the QC testing according to SOP.  We used to joke that a monkey could do it and all you had to do was feed it peanuts every so often.  No offense to QC personnel.

[AmbientLight]

BDunbar is correct here. It takes many years for a manufacturer to establish good quality control procedures.

You don't change those for the worse, if there is no pressing need to do so and Canon is certainly not a company near bankruptcy, so that cost reductions in quality control may become necessary just to be able to manage cash flow.

[kennephoto]

Heck I bet QC has improved since canon has been around for some time now. I know the company I work for it sure has gone up, no company wants to lose customers.

[myocyte]

Quality standards are not reduced over time, but the number of units getting a major inspection off the line can decrease. In order to maintain efficient throughput in manufacturing, not all aspects of the assembly are inspected at 100%. Rather, random sampling is done at various points on the line to try and establish a confidence level in the units being built. Normally, during the first builds, Acceptable Quality Level (AQL) values tend to be stringent so that inspection sample sizes may be larger, and this is done so that the confidence level is greater for an acceptable number of units. If the inspections on those sample sizes falls within the acceptable range with an acceptable deviation from the nominal, inspection sample sizes may get smaller over time due to a reduced AQL on a well-established procedure.

I'm sure there is a standard final quality assurance procedure that is done 100% as the final step, but it is probably not as detailed as a complete inspection based on the AQL sampling of a manufacturing lot. The same also applies to the components in a lens (e.g. optic components, lens housing, etc). So, it's completely believable that there is some unit-to-unit variation since not all components are inspected at a 100% level (a lot of validation and six-sigma work is probably done in parallel to nail down the process for the manufacturing of the components too).

[DB]

Deja-Vu all over again, just like the 24-70 mark 1, everyone will be talking about whether there's is a 'soft' or 'good' copy

Having said that, according to the digital picture tests in the previous links, the 70mm @ 2.8 looks about the same as the 24mm @ 5.6

QC is all about tolerances - not every product has to be 100%, so 99% = "QC Passed"

Then there is the LensRental argument that both bodies and lenses have tolerances, so AFMA becomes an imperative...and so on

[marekjoz]

Read http://www.lensrentals.com/blog/2012/09/a-peak-inside-the-canon-24-70-f2-8-mk-ii
You'll see what a big progress they've done to keep better quality in 24-70 II vs 24-70 I and where those differences you showed could come from.

[Axilrod]

Could the AF have been off on the soft sample?  I noticed the 24-70 II I used was backfocusing slightly.

Also, I wouldn't always trust their test results.  Some of the Zeiss lenses they have on there have terrible test results but when I've used them they were absolutely amazing.

[Invertalon]

Anybody know if they focus via live-view for these tests (I would hope so)? It would be weird relying on the AF of the camera, so I am 99% sure they do use live-view.