[Edgar_in_Indy]

I think brands just change certain things in their auto focus module. Unfortunately it can mean that third party lenses cease to work appropriately on new gen cameras. I would say it is a side effect of Sigma's reverse engineering rather than a specific goal of Canon or Nikon to make the lenses stop working. Sigma often comes out with updates in these situations, but you usually have to send your lenses in to get them fixed (I remember some issue with certain Sigma lenses not working in live view on Pentax cameras, not the end of the world, but it did require the lenses to be sent in to fix it).

That makes sense of course, but Zygonx implied that Nikon did something on purpose to interfere with Sigma lenses.

[y0chang]

all manufacturers have micro adjustments between the lens electronics to the camera electronics. It tells the camera about the focal length and other factors. Sometimes it will micro adjust lenses due to known focus shifts and other aberations. Newer lenses will factor in much more complicated things, but third parties can't access the camera electronic code so must adjust everything inside the lens (see sigma usb dock). But because they can't talk to each other there will always be some issues. Fortunately (and getting back to the 24-70mm), the 24-70 will be a 1st party lens.

[Rondec]

That makes sense of course, but Zygonx implied that Nikon did something on purpose to interfere with Sigma lenses.

Well, Nikon and Canon and Pentax would never say that's the reason why they make these changes, but I don't suppose they are bothered if it means that more folks buy brand name lenses at a higher rate. They just don't care if some change would brick your sigma lens.

[Edgar_in_Indy]

all manufacturers have micro adjustments between the lens electronics to the camera electronics. It tells the camera about the focal length and other factors. Sometimes it will micro adjust lenses due to known focus shifts and other aberations. Newer lenses will factor in much more complicated things, but third parties can't access the camera electronic code so must adjust everything inside the lens (see sigma usb dock). But because they can't talk to each other there will always be some issues. Fortunately (and getting back to the 24-70mm), the 24-70 will be a 1st party lens.

I'll try to find the story, but the gist of it was that this was something new where the lens provided feedback to the camera. They tried older lenses from the same manufacturer, and the older lenses did not provide the same benefit.

EDIT: Here are some links that may be relevant. It looks like lensrentals.com found the improved AF feedback-loop system was introduced on the 5D III and on a few of the newer lenses, but it did not seem to exist on the 5D II or in older lenses.

http://www.lensrentals.com/blog/2012/08/autofocus-reality-part-3b-canon-cameras

http://www.dpreview.com/articles/5402438893/busted-the-myth-of-open-loop-pha...tion-autofocus

The relevant terminology appears be "closed loop" AF system. In a closed loop system like on the 5D III when used with the newer lenses, the lens provides feedback to camera, and the camera sends updated instructions to the lens, back and forth until perfect focus is achieved AND confirmed.

With an open loop system, on the other hand, the AF sensor (phase detect sensor for a DLSR) takes a look, tells the lens where to go, and the lens goes there and focus is considered to be locked. If the AF sensor is perfectly calibrated, and the lens is perfectly calibrated, and the image sensor is installed perfectly in the camera, and the mechanics and optics of the lens are working how they're supposed to, then the open loop system should work well. But when any one of those things is off a bit, then the camera will think focus is locked, even when the final image is back focused or front focused.

With contrast detect, like on mirrorless cameras, this is pretty much a non-issue.

---

Last edited by Edgar_in_Indy; 10-02-2015 at 08:35 AM.

[Class A]

Which other manufacturer has so poor quality control as to sell repair kits (USB docks) for their lenses?

Pentax, Canon, Nikon, etc.

They all offer cameras that come with a built-in "repair kit". It is called "AF adjustment".

If you understood anything about AF and manufacturing tolerances, you'd knew that the inclusion of an "AF adjustment" facility is not admitting to poor QC. The Sigma dock is just a much better tool to address AF issues. Every lens manufacturer should offer such a dock. The alternative is to send one's lens in to a service centre which will use exactly the same adjustment approach (changing data in the lens). Many lenses cannot be adjusted mechanically so the changing of lens data is the only way.

I'm leaving this discussion now, which has reached a new level of absurdity when plus points such as the option (not necessity) for users to do their own calibrations and six year warranties are used as arguments against a manufacturer.

[goldenarrow]

I'm pretty sure that Sigma calibrates the lens to the body...not the body to the lens. I don't know if you meant to imply that it is the other way around, but that's what it sounded like to me.




But you made a good point that calibration is not all in the lens. Manufacturing tolerances also apply to the camera bodies. I witnessed this first hand when I upgraded from my K-x to my K-30...calibration was drastically different. So in my experience, Pentax is not better than lens manufacturers in this regard. Which is why I found it useful to have my lenses calibrated to my specific body.



In fact, this whole calibration thing is something I hate about DSLR's, and is one of the reasons I have not upgraded to the K-30, and have been looking hard at mirrorless.

I sold a Sigma 100-300mm f4 to someone on the forum last year, and on my K-30 it worked beautifully with perfect focus. But the buyer, who had a K-5, claimed that it was too far out of calibration to correct on his body, and forced a return through PayPal, even though I had discussed possible camera body calibration issues with him before the sale. I then resold it on eBay as "for parts or repair" at a significant discount since I didn't want to deal with a second return, and the second buyer used it on his K-3 without any issues and was delighted with it and left gushing feedback. So yeah, I'm pretty sore on the subject in general, and it seems to me that Pentax bodies are a moving target in terms of calibration.

Sorry to hear of the 100-300 experience with various Pentax. Another reason I am keeping my lens setup limited to 5 or maybe 6 lenses. I may ask Sigma if they change anything on the body. Not sure how they can do that without knowing how for dozens and dozens of different bodies and models and manufacturers. It must be that they tune the lens to the body sent to them and be done with it.

The 24-70 pentax coming out is a Tamron lens. Tamron is also "guilty" of lenses with focus problems as well as decentering problems. So, sigma is not alone.

Thanks again for your thoughts.

[reh321]

I'll try to find the story, but the gist of it was that this was something new where the lens provided feedback to the camera. They tried older lenses from the same manufacturer, and the older lenses did not provide the same benefit.

EDIT: Here are some links that may be relevant. It looks like lensrentals.com found the improved AF feedback-loop system was introduced on the 5D III and on a few of the newer lenses, but it did not seem to exist on the 5D II or in older lenses.

LensRentals.com - Autofocus Reality Part 3B: Canon Cameras

Busted! The Myth of Open-loop Phase-detection Autofocus: Digital Photography Review

The relevant terminology appears be "closed loop" AF system. In a closed loop system like on the 5D III when used with the newer lenses, the lens provides feedback to camera, and the camera sends updated instructions to the lens, back and forth until perfect focus is achieved AND confirmed.

With an open loop system, on the other hand, the AF sensor (phase detect sensor for a DLSR) takes a look, tells the lens where to go, and the lens goes there and focus is considered to be locked. If the AF sensor is perfectly calibrated, and the lens is perfectly calibrated, and the image sensor is installed perfectly in the camera, and the mechanics and optics of the lens are working how they're supposed to, then the open loop system should work well. But when any one of those things is off a bit, then the camera will think focus is locked, even when the final image is back focused or front focused.

With contrast detect, like on mirrorless cameras, this is pretty much a non-issue.

When I played softball, I would head where I thought the ball was going, and kept adjusting my path until I caught the ball ... or flubbed the catch. I was amazed to see Yogi Berra turn around as soon as the ball was hit, run to some spot, then look up and catch the ball; years as being catcher apparently allowed him to determine the path of the ball as soon as he saw it leave the bat, so when Elston Howard replaced him as catcher, he was able to become a good solid outfielder.

When manually focusing, I make a change, and based on the results make another change, continuing that process until I can see that the object is in focus; in retrospect it is sort of like how I played outfield. I had always assumed that auto-focus worked the same way. Are you saying that it doesn't, that it works more like Yogi Berra playing the outfield?? If so, it explains the front/back focus problems that people with expensive cameras are always talking about. I always wondered why the camera couldn't tell that it had gotten it wrong; you seem to be saying that it is so sure it is right that it doesn't check. Do less expensive cameras, like my K-30 work the same way? I can't say that I have ever taken a picture that appears to me to be out of focus.

[bxf]

years as being catcher apparently allowed him to determine the path of the ball as soon as he saw it leave the bat

Nah, that's not it. With Yogi, it was Deja Vu all over again.

[Edgar_in_Indy]

When manually focusing, I make a change, and based on the results make another change, continuing that process until I can see that the object is in focus; in retrospect it is sort of like how I played outfield. I had always assumed that auto-focus worked the same way. Are you saying that it doesn't, that it works more like Yogi Berra playing the outfield?? If so, it explains the front/back focus problems that people with expensive cameras are always talking about. I always wondered why the camera couldn't tell that it had gotten it wrong; you seem to be saying that it is so sure it is right that it doesn't check. Do less expensive cameras, like my K-30 work the same way? I can't say that I have ever taken a picture that appears to me to be out of focus.

Yes, there is an AF sensor in your DSLR separate from the image sensor. So autofocus is performed using this other sensor, which is not always in perfect calibration with the actual image sensor. As mentioned above, my K-x and K-30 required different lens calibrations, probably due to these differences. Click on the image for a link to the article where I grabbed the image:



It is also my understanding that most DSLR AF systems have been open-loop. They are like Yogi Berra in your example. He sees where the ball should be, and he goes there and hopefully he was correct. In a closed loop system, there is a continuous AF cycle where the camera and lens keep checking focus and making adjustments until it is confirmed to be correct. This would obviously be a more elaborate system, and it's apparently what Canon has designed into the Mark III and some of the new high-end lenses.

The reason that mirrorless cameras are able to avoid the lens calibration issue is because autofocus is performed on the image sensor itself. There is no separate AF sensor that has to be in perfect calibration with the actual focal plane. This is the so called "contrast detect" method of autofocus. In the past, it was markedly slower than phase detect systems, but with some of the latest mirrorless cameras that is no longer the case. Some mirrorless cameras are even said to be able to focus better in low-light conditions than DSLRs. And it seems reasonable to assume that they will keep getting faster in future models.

[bxf]

In a closed loop system, there is a continuous AF cycle where the camera and lens keep checking focus and making adjustments until it is confirmed to be correct

So is this not what is happening when we see a lens going back and forth a few times trying to obtain focus, as is sometimes the case?

[Digitalis]

sold a Sigma 100-300mm f4 to someone on the forum last year, and on my K-30 it worked beautifully with perfect focus. But the buyer, who had a K-5, claimed that it was too far out of calibration to correct on his body

As you know I also own a Sigma 100-300mm f/4 and I have never had any serious AF issues with it when I used it on my K10, K7 and K5IIs.

[pacu]

Phase detection is not an one time reading and calculation. At least in the case of Pentax K-r and Pentax K-30 that I know well, they make adjustments sometimes while focusing in single mode. (My other SLR is manual and for borrowed ones from other companies I cannot be sure).

[reh321]

Yes, there is an AF sensor in your DSLR separate from the image sensor. So autofocus is performed using this other sensor, which is not always in perfect calibration with the actual image sensor. As mentioned above, my K-x and K-30 required different lens calibrations, probably due to these differences. Click on the image for a link to the article where I grabbed the image:

It is also my understanding that most DSLR AF systems have been open-loop. They are like Yogi Berra in your example. He sees where the ball should be, and he goes there and hopefully he was correct. In a closed loop system, there is a continuous AF cycle where the camera and lens keep checking focus and making adjustments until it is confirmed to be correct. This would obviously be a more elaborate system, and it's apparently what Canon has designed into the Mark III and some of the new high-end lenses.

The reason that mirrorless cameras are able to avoid the lens calibration issue is because autofocus is performed on the image sensor itself. There is no separate AF sensor that has to be in perfect calibration with the actual focal plane. This is the so called "contrast detect" method of autofocus. In the past, it was markedly slower than phase detect systems, but with some of the latest mirrorless cameras that is no longer the case. Some mirrorless cameras are even said to be able to focus better in low-light conditions than DSLRs. And it seems reasonable to assume that they will keep getting faster in future models.

I wonder if my tendency to use mid-range aperture, like f/8, rather than wider apertures where DoF would be more skinny, is why I tend not to have problems with AF getting it wrong.

[bxf]

I wonder if my tendency to use mid-range aperture, like f/8, rather than wider apertures where DoF would be more skinny, is why I tend not to have problems with AF getting it wrong.

Works for me, but I would also add that AF inconsistencies appear to be more serious under some (artificial) lighting conditions.

[IchabodCrane]

Yes, there is an AF sensor in your DSLR separate from the image sensor. So autofocus is performed using this other sensor, which is not always in perfect calibration with the actual image sensor. As mentioned above, my K-x and K-30 required different lens calibrations, probably due to these differences. Click on the image for a link to the article where I grabbed the image:



It is also my understanding that most DSLR AF systems have been open-loop. They are like Yogi Berra in your example. He sees where the ball should be, and he goes there and hopefully he was correct. In a closed loop system, there is a continuous AF cycle where the camera and lens keep checking focus and making adjustments until it is confirmed to be correct. This would obviously be a more elaborate system, and it's apparently what Canon has designed into the Mark III and some of the new high-end lenses.

The reason that mirrorless cameras are able to avoid the lens calibration issue is because autofocus is performed on the image sensor itself. There is no separate AF sensor that has to be in perfect calibration with the actual focal plane. This is the so called "contrast detect" method of autofocus. In the past, it was markedly slower than phase detect systems, but with some of the latest mirrorless cameras that is no longer the case. Some mirrorless cameras are even said to be able to focus better in low-light conditions than DSLRs. And it seems reasonable to assume that they will keep getting faster in future models.

Nice explanation, Edgar. One point to note is all AF systems are closed-loop. If they weren't, they would have almost no chance of "nailing" focus. What you might be thinking of is PDAF systems are able to estimate the focus parameters by knowing how far out of focus they are at any time (the Yogi Berra example of running to where he thought the ball would land). This is what has made PDAF systems so fast. Historically, CDAF systems couldn't do this which is the primary reason they (again, historically) were slower focusing. The notable exception is Panasonic has recently developed a CDAF system where it also can estimate final focus parameters using some algorithm that looks at how blurred the image is before focusing starts. So far, it only works with some of their newer lenses.