30D makes Sigma lenses go bananas

[don_b__south_africa_]

>> My Ford Warrantee does not cover my aftermarket non-ford XM Radio.

 

No, of course not, but if your new Ford started putting out 10 volts instead of 12 on the accessory circuit and, as a result, your old radio wouldn't tune in, you'd have a legitimate beef against Ford. You wouldn't be much soothed by Ford's statement that all Ford-brand radios have stable tuning down to 9 volts, now would you?

[don_b__south_africa_]

Neil, I salute your dedication to manual focus. I regard this new-fangled autofocus stuff with some suspicion myself, and I'd be delighted if my old FD lenses could be made to work on a digital body. Ten points off, Canon, for screwing all your old customers. I'd be even more delighted if digital bodies had proper focusing screens, but, alas, they don't, unless you spend the seriously big bucks.

 

>> Canon does not care because they want you to buy thier own lenses of course. Their angle is that their lenses have no compatitbility issues!

 

If Canon simply didn't care, I wouldn't be annoyed. What irks me is the strong suspicion that Canon keeps in reserve secret and fundamentally nonfunctional parts of the lens communication protocol so that it can invoke them now and again in new bodies to confound Sigma lenses. Sigma is the biggest off-brand seller, is it not?

 

I say nonfunctional because communication between a body and lens is hardly rocket science. Certain, very limited parameters need to be passed both ways, and it should be the easiest thing in the world to reverse-engineer all the functional aspects of such a protocol. I suspect that Canon itself is doing some reverse-engineering -- of Sigma lenses. Aha, says Canon after testing a bunch of Sigma lenses, Sigma is using such-and-such a parameter to do such-and-such arithmetic in a way that we don't -- so let's change the focusing algorithms to mess up Sigma lenses' focusing calculations! Plausible, no?

 

Don

[don_b__south_africa_]

>> Every single Canon camera's manual warns you against using 3rd party lenses / flashes.

 

Yes, well, and my Suzuki motorcycle manual warns me sternly against using any lubricants other than Suzuki-brand oils, which are, of course, three times the price. Does anybody take this kind of self-serving corporate warning seriously?

[don_b__south_africa_]

An even more plausible scenario for what Canon may be doing is this. The company may have taken a decision at the dawn of the EOS system to insert a different set of public-key-encrypted data into the firmware of every EF lens.

 

No third-party lens maker would be able to deduce its function, because it wouldn't have any overt function, and because every lens would have a different encrypted data block, it wouldn't be possible simply to reproduce it in third-party lens firmware and hope for the best.

 

Then, as Canon rolls out new bodies, it can have the algorithms in those bodies interrogate the encrypted data and make interesting decisions about what to do if everything doesn't add up.

 

This wouldn't explain why Tamron and Tokina lenses allegedly have no problems on new Canon bodies (although some say they do), but it should be possible to contrive such altered body algorithms to target a specific maker's -- such as Sigma's -- lenses specifically.

 

Sigma is getting a bad rep because of these difficulties and, if Sigma is Canon's biggest lens competitor, it might be in Canon's interests to keep most Tamrons and Tokinas working perfectly just to make Sigma look particularly bad.

 

A prediction follows from this hypothesis: If Sigma goes down and Tamron overtakes it as the biggest threat to Canon lens sales, one will expect to see Tamron lenses mysteriously failing to work on new Canon bodies.

[neild]

As I said in my first post, I tend to agree that Canon are actively making things difficult for Sigma (no evidence of this, it's just what I think). What amazes me is that there must be so many ways to reconfigure the connections but somehow Canon's own lenses do not become obsoleted by this practice. A masterstroke of design if it was deliberate, whatever you think of its "morals"!

[neild]

Don, your firmware theory may have answered my question as to one possible way Canon could arrange things - I should read more carefully!

[don_b__south_africa_]

>> What amazes me is that there must be so many ways to reconfigure the connections but somehow Canon's own lenses do not become obsoleted by this practice.

 

Indeed, and I think you've hit the nail on the head -- it must be digital chicanery involving encryption. I doubt that those EF lens contacts are anything but a data bus (in other words, there's not one contact for focus, one contact for aperture and so on).

[don_b__south_africa_]

>> I bought Canon system because lenses quality is the most important factor to me. Cameras are almost disposable parts and can be exchange every few years. Mostly quality of photo is limited by glass not a camera.

 

Canon makes some brilliant glass but it also makes some horrible glass, and has peculiar gaps in its range. Sigma makes a lot of horrible glass but also some gems, such as the 70-200mm/2.8 and the 30mm/1.4. It is blind brand loyalty to suggest that only Canon glass is capable of producing top-quality pictures.

[mbb]

What is good for some it is not necessary good for others. Trying to kill market for superior Canon lenses will either drive their (Canon) price up to keep production or drop the quality in order to lower price. I am not interested in neither situation so any way to eliminate parasites is good. Especially that reverse engineering is very immoral to me. Protecting own investment and research is not. Supporting thieves trying to make easy money on somebody's hard and expensive development work is also questionable as a moral position.

 

One loves Sigma lenses, satisfied with their optical quality - nothing wrong with that. But one should buy off-brand camera to match it if do not want to take any risk of compatibility.

 

That said I am not a Canon freak (even that I am using their equipment exclusively), there are just happened to make the best tools I need. As everybody else I am trying to save money if possible on some cost but never when compromise the photo quality. As an example I am using off-brand batteries, dozen of them, on my own risk. Saved money goes back to buy more and better glass anyway. So yes I am supporting those who copied their shape (batteries) and cutting profit for Canon. Only excuse is that batteries are not invented (except for shape) by Canon but their cameras and lenses are. Same with camera bags etc. I do not have to show Canon name outside, some places I event prefer not to.

 

And just an extra benefit (very important to me) I usually have my stuff back from Canon service in about a week, only once took about 2 weeks. Some of my friends were waiting over a month to get their Sigma lenses serviced.

[don_b__south_africa_]

Thieves and parasites! That's rather extreme. Look, there are degrees of reverse engineering, and some are certainly immoral, such as copying patented techniques and processes.

 

But, by your standards, it would be immoral, for example, to measure the output of a Ford's electronic ignition system so that you could design a compatible tachometer. And I think that's absurd.

 

And what "investment and research" of Canon's are off-brand lens makers actually exploiting? They're taking a signal from the camera's metering and focusing systems, certainly, but that's inherent in building a modern lens. For the rest, they're conducting their own perfectly respectable R&D on lens design, coatings, etc etc. And that's surely where the serious R&D is concerned in making a lens.

[yves_jalbert]

<p>

Wow! This thread is becoming a monster of it's own (laughs).

</p>

 

<p><i>

"The company may have taken a decision at the dawn of the EOS system to insert a different set of public-key-encrypted data into the firmware of every EF lens."

</i></p>

 

<p>

Conspiracy theories now?!?! (LOL). While I won't rule out the possibility that Canon made a personnal, regular, business decision at risk of getting flamed, I would like to bring back everybody's attention to Bob Atkin's comment which is all true. He's not inventing anything here and this sudden realization is not something new at all. I think you can forget about encryption keys inserted in lenses 15 years ago.

</p>

 

<p>

<i>

"Tamron and Tokina lenses do not suffer from these incompatibility issues, yet Sigma lenses have been plagued with them for at least 10-15 years. It's not just digital. I remember WAY back when a Sigma 400/5.6 lens that had been working just fine on an EOS Rebel and EOS 630 wouldn't work at all on an Elan!"

</i>

</p>

 

<p>

And about Don's prediction...

</p>

 

<p><i>

"A prediction follows from this hypothesis: If Sigma goes down and Tamron overtakes it as the biggest threat to Canon lens sales, one will expect to see Tamron lenses mysteriously failing to work on new Canon bodies"

</i></p>

 

<p>

My personnal opinion? This one has been discussed over and over at repetition for weeks in this forum. Either we get the CIA on it (hey, sure why not? - laughs), or lay it down to rest.</p>

 

<p>The market gives us the possibility to buy third party lenses which it's manufacturer claims will work fine on original Canon equipment (some brands work better than others, some not, just as in any manufacturing sector). Or we can buy the original that was designed and built by the manufacturer to be used on the original body. It's a personnal individual choice we make and it depends on how much we are willing to invest in our equipment.</p>

[don_b__south_africa_]

Well, Yves, a conspiracy requires more than a single party to it, so I don't think I have a conspiracy theory here.

 

I reckon it's perfectly possible that Canon, irked by loss of lens sales to off-brand makers in the FD period, decided to put something into the new EF lenses -- which now contained their own little computers, after all -- with which to make life difficult for third-party lens makers.

 

As I said in an earlier post, the information exchanged between lens and body for functional purposes must be be quite trivial and very easy to decode, certainly for a big company with Sigma's resources. If Sigma does keep on getting it wrong, either Sigma has a bizarre blind spot or something hidden in the Canon system is periodically being wheeled out to scupper the current generation of Sigma lenses.

 

In fact, it seems less a possible explanation than a probable one. Nothing paranoid involved.

[henrik.ploug]

The only thing Canon needs to build into their cameras are some kind of recognition-algorith. This should be able to distinguish between Canon-lenses and non-Canon-lenses.

 

If the lens is made by Canon, the camera will send correct signals to the lens. If the lens is not made by Canon, the camera will send send incorrect information to the lens.

 

But I don't think, that this is very likely. I have seen both Sigma- and Canon-lenses misfocus. So my guess is, that the AF-system just isn't as reliable, as we like to think.

[don_b__south_africa_]

Henrik, yes, a recognition algorithm is what would be required. However, it would have to be encrypted and appear different for each lens or else it could simply be copied. I tried earlier to give an over-elaborate account of how such a scheme might work.

 

To me, though, logic is increasingly suggesting that such a scheme might well exist. The occasional Canon lens might misfocus a bit, but by and large they are rock solid on Canon bodies. Sigma, on the other hand, has a terrible reputation for lenses being obsoleted by new Canon bodies. This fact is not disputed by most people.

 

And to labour the point slightly, I find it stretches credulity to think that Sigma is simply a stupid company that cannot reverse-engineer body-lens communication. What has to be transmitted, anyway? Surely nothing more than distance info from the lens and aperture info and focus-motor instructions from the body. Not rocket science, as I have said.

[henrik.ploug]

Half a year ago I made this test with my Sigma 30mm and my Canon 50mm:

 

http://www.photo.net/photo/3755990

 

As you can see the Sigma is a lot sharper than the Canon wide open. Later I found out, that this was because of backfocus. See this thread:

 

http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=00ERXl

 

So I had my 50mm sent in for calibration. When it returned, I was told, that there was nothing wrong with the lens. Then I had my 350D sent in for calibration (with the 50mm). When they returned, I made this test:

 

http://www.photo.net/bboard/uploaded-file?bboard_upload_id=30201984

 

As you can see, it is now the 30mm Sigma that is soft wide open and the 50mm Canon that is sharp. And my new 17-55mm f2,8 IS is sort of in between.

 

My conclusion was, that the AF on the 350D is not very consistent, especially not at f1,4. So I sold the 350D and my two f1,4-primes and bought a 30D.

 

Now I just have the 30D and my new 17-55mm f2,8. I haven't made any test with the setup yet, but my impression is, that the 30D AF is more consistent than the 350D AF. And that the 17-55mm f2,8 doesn't require as much AF-precision as the f1,4-primes.

 

The bottom line is, that I think we expect to much from AF.

[andrew robertson]

" To Canon, I am a nobody. If you buy Sigma lenses then you are also a nobody in Canon's

eyes - why should they care what you think?"

 

Actually, Neil, you bought a Canon camera, didn't you? That counts for something, and

they are happy to support their own products. Canon has always been more than willing

to answer technical questions, but officially they, like most other manufacturers, only

warrantee their own products. This isn't a case of the Canon Corporation or its officers

and employees having feelings of anger against you.

 

"What irks me is the strong suspicion that Canon keeps in reserve secret and

fundamentally nonfunctional parts of the lens communication protocol so that it can

invoke them now and again in new bodies to confound Sigma lenses."

 

Don, you can allay your suspicion. Canon does in fact keep their entire lens / body

communications protocol(s?) completely secret. They have licensed this technology to no

other company. Sigma and Tamron and Tokina and Phoenix and so on have no

fundamental legal right to compel them to do so either. Canon is not a monopoly.

 

We don't really know how the Canon EOS / EF interface works. It may be a lot more

complicated than we may imagine it to be. It is entirely possible and even likely that

instead of this being an evil plot against Sigma et al, the lens / body issues are a side

effect of the vast complexity of the system. With a total of 53 bodies, 127 lenses, 4

teleconverters, and 4 extension tubes available, the amount of possible interactions

between these highly computerized components is huge. Canon also periodically adds

functionality to the lenses and bodies, and as the 'user' is never supposed to interact

directly with a lot of the onboard software Canon designs their software to be mostly

invisible to the user. All he or she has to know is if they only stick to Canon stuff, it all

works flawlessly.

 

"Yes, well, and my Suzuki motorcycle manual warns me sternly against using any

lubricants other than Suzuki-brand oils, which are, of course, three times the price. Does

anybody take this kind of self-serving corporate warning seriously?"

 

Well, if your bike suddenly gave 'err99' and didn't start, maybe people would. You

certainly don't fill its gas tank with diesel or lamp oil or imitation vanilla flavor and expect

to drive it away do you?

[andrew robertson]

Henrik, you should go over to the nature photography forum and talk to people who

regularly achieve perfect AF right on the eyes of flying birds etc. There's more to AF than

'press the button, kacham!'. AF is just a computer program that reads in data from some

optical sensors. I can AF within system tolerances on any number of moving or stationary

subjects with all my lenses. You should also be aware that the AF sensor indicators in the

viewfinder are often not the same exact size as the actual AF sensors themselves.

[don_b__south_africa_]

>> "We don't really know how the Canon EOS / EF interface works. It may be a lot more complicated than we may imagine it to be. It is entirely possible and even likely that instead of this being an evil plot against Sigma et al, the lens / body issues are a side effect of the vast complexity of the system. With a total of 53 bodies, 127

lenses, 4 teleconverters, and 4 extension tubes available, the amount of possible interactions between these highly computerized components is huge."

 

Any sensible design engineer -- and we assume Canon employs sensible engineers -- would, when faced with the prospect of a system with so many components, design an interface to be rigorously standard and as simple as possible.

 

And, indeed, it seems Sigma is capable of making its lenses function flawlessly on all 53 Canon bodies. It's only when the 54th body comes out that Sigma lenses go haywire on it. Which smacks, if you prefer to call it such, of an evil plot.

 

Not that I think it's evil in any way. It's just a reason to wonder if Nikon isn't perhaps a better option if you want the widest possible choice of lenses.

[don_b__south_africa_]

>> "My conclusion was, that the AF on the 350D is not very consistent, especially not at f1,4. So I sold the 350D and my two f1,4-primes and bought a 30D."

 

Henrik, you may be right about the 350D. One of the reasons I upgraded to a 30D was because my most prized lens, the 24-70/2.8L, was a little iffy about focus on the 350D. Nothing serious, mind you -- it was just a little bit off, randomly either back or front, more often than I would have liked. Tests on a friend's 20D and a colleague's 1D showed that the lens focused perfectly on those bodies. And now it focuses perfectly on my new 30D. So far, so excellent.

 

Ironically, the Sigma 30/1.4 always focused bang-on on the 350D, even at 1.4, where the DOF is sliver-thin. I expected it to do just as well on the 30D, but instead it front-focuses by miles -- or by about 20ft over a 60ft distance. This is not "misfocusing". This is something seriously amiss.

[henrik.ploug]

ASndrew: "Henrik, you should go over to the nature photography forum and talk to people who regularly achieve perfect AF right on the eyes of flying birds etc. There's more to AF than 'press the button, kacham!"

 

I'm aware that it takes skill to get autofocus right, if you are shooting a flying bird. But shooting a flat brick wall with the centerpoint AF is not rocketsurgery. It is pretty simple to see if the lens is backfocusing, frontfocusing or - hopefully - something in between.

[rob_murray]

Not to worry, the Sigmas aperture is likely to break soon anyway so just send it in then for a total rebuild and hope they still have parts for it.

[neild]

Andrew, my point was merely that Canon want us to buy Canon products. This is not unreasonable on their part - I would not expect them to be sympathetic towards me if one of my non-Canon lenses somehow upset my Canon camera. As consumers, we have the right to buy off-brand things if we can find them (and want them), but we do so at our own risk of course.

[yakim_peled1]

>> And Canon, in fact, makes no lens comparable to the 30mm/1.4, unless you count the 35L, which is a little too rich for my blood.

 

I think that the 28/1.8 USM is a comparable lens to the 30/1.4 and is FF compatible to boot. Price is actually a little cheaper.....

 

Happy shooting,

Yakim.

[don_b__south_africa_]

>> I think that the 28/1.8 USM is a comparable lens to the 30/1.4 and is FF compatible to boot. Price is actually a little cheaper.....

 

Yes, but f/1.4 is nearly a stop faster than f/1.8. However, I think Canon has just about convinced me to get rid of the Sigmas, and I suppose the 28/1.8 is the best Canon option for a normal lens on a crop camera.

[dogbert]

Has anyone else experienced a problem with the Sigma 10-20 and 30D combination? I have a Sigma 10-20 and was thinking of upgrading to the 30D.