Sigma 24mm f/1.4 DG HSM Art Coming in October

[100]

It's no possible they leave that lens to chance. And they still gave him a non working lens, so does that give me high hopes for a retail lens? Nope..

Did you actually read the review?

Bryan wrote things like:
With some inconsistency showing in tripod-based One Shot mode AF and with AF being only moderately fast, I was not expecting great performance from this lens in AI Servo AF mode. I was, however, pleasantly surprised that this lens delivered much better results than I expected. I spent a day at the track, capturing close to 1,000 photos of runners in action with the 1D X and 5D III behind the lens.

There is a lot of grey between black and white.
There are inconsistencies, that’s a shame. We don’t get a percentage though. Sure, 4 out of 10 static shots shown in the review were out of focus, and that’s substantial but anyone who knows anything about statistical research knows that you need a lot more shots before you can draw any meaningful conclusions.

I imagine Sigma has done a lot of testing themselves before handing out lenses to reviewers, but it’s always possible they missed something. Camera bodies get tested for light leaks and still the first 5D3’s had one.
Maybe Sigma can iron out the inconsistencies Bryan discovered with a firmware update maybe not, time will tell.

 

[Loswr]

Seriously? An appeal to what's relative? AFMA would have shifted the spectrum of focus variability, perhaps into a more accurate range. Pure and simple. Yes, it would not change focus variability (assuming all other things constant), but it would change focus accuracy and objective (not relative) sharpness. So, don't be juvenile and try and trivialize its mention. I would think you'd endorse disclosure of methodologies, but maybe your background isn't that rigorous.

Yes, seriously. Maybe in your world, AFMA stands for Auto-Freakin'-Magical-Adjustment...but in the real world, it's a systematic correction, and a systematic correction will not compensate for a random error. Maybe your background doesn't facilitate an understanding of those basic concepts.

"Shift the spectrum of variability?" Yes, it would have shifted it..for better or worse. It might change objective sharpness, it might not. If no adjustment was needed, or if it was already performed (as it most likely was), further adjustment would reduce objective sharpness. In no way would application of AFMA make the four OOF shots sharper while not also affecting the six equivalently sharp shots. If AFMA had been applied such that the most egregiously blurry shot (#3, IIRC) was sharp, you'd have nine OOF shots instead of four.

The point is that the results show inconsistent AF. Pure and simple.

As for your innuendo and aspersions, I'm curious...in which field is your advanced degree, and how many peer-reviewed scientific journal articles are listed on your CV?

 

[Loswr]

There could also have been an earthquake (or minor tremors) rumbling through ...

;D

Maybe. Being from California, I know what earthquakes feel like, and have been through several major ones. Bryan lives in Pennsylvania, where like New England, what gets reported on the evening news as an earthquake causes our house to shake less than our young kids running around upstairs...

 

[jdramirez]

There could also have been an earthquake (or minor tremors) rumbling through ...

;D

Maybe. Being from California, I know what earthquakes feel like, and have been through several major ones. Bryan lives in Pennsylvania, where like New England, what gets reported on the evening news as an earthquake causes our house to shake less than our young kids running around upstairs...

I live in pa. It has been windy lately... And cold... so there might be a shiver factor.

Here's a dumb question... south this type of inconsistency... how do you afma the lens? Just find a mean, mode or median and pray.

 

[BLFPhoto]

Here's the thing, Neuro...I would call that 10 shot sample a REALLY shitty result for Sigma. Worse than many of their older lens' results. And you're buying that without questioning the validity of the test and data? I have lurked longer than I've posted, and frankly, I've come to expect more rigor from your responses. You've got a great academic take on things, and a lot of insight that the more casual Canon user doesn't have, so I'm a bit perplexed here. Please spare us the harping on precision vs accuracy. Clearly those of us carrying on the discussion have demonstrated that we are aware of the difference. I once wrote a defense white paper on the different test methods and protocols required to measure accuracy vs precision. I'm pretty read into the difference. I'm also well versed in the absolute test control and statistical requirements to actually assess levels of precision. I simply don't see that in this review or write up. It certainly wouldn't stand up to peer review of the test results.

Therefore, I'm not satisfied that this is a valid result, let alone indicative of the general quality of the lens. I have followed TDP for a very long time and agree his reviews and methods are some of the most complete out there...the standard. Y which most are judged, as you note. But here and on other reviews, I think there is a need for much further disclosure of the exact configurations under test, including all camera settings. Clearly there is room for a lot of speculation on why he got this sequence of shots, and not all of it is to Sigma's detriment. Incomplete data can be worse than no data if errors are induced in the analysis.

If the Sigma proves to be a lemon, so be it. But I'm not hanging my hat on that appellation based on this review. Let's not get too sold on the strong TDP review brand that were blind to gaping holes in the data. Question everything. Over on my triathlon forums where we debate aerodynamics and friction to the same degree we here on the canon forum disect lenses, I have a friend who often states, "In god we trust. All others must bring data." I'm not a god guy, but I understand what he's getting at.

Meanwhile, I'm curious to watch the development of more Art series lenses like the 24..

 

[tron]

Now, for this lens o be useful even without being AF perfect it will have to be a superset of Samyang 24 1.4

1. NO Coma
2. No decentering issues.
3. Very good center wide open with good corners.

If the above are met then I would tolerate some minor AF inaccuracies and consider it a successful 24 1.4 lens.

If they are not met then it will face very strict competition form Canon's 24 1.4L II lens and only a really low price will make people prefer it.

P.S Did I mention NO COMA?

 

[jdramirez]

I have a friend who often states, "In god we trust. All others must bring data."

Love the line. I doubt I'll steal it because I require evidence from all deities as well.

And even the most well respected scientists have been wrong.

Watson and Crick got right what... what's his name for wrong... what is his Damn name? Christo de jesus!

 

[Loswr]

...how do you afma the lens? Just find a mean, mode or median and pray.

The best way is how Reikan FoCal does it – take many shots of a test chart, varying the AFMA value applied, plot the sharpness (measured by image analysis) against AFMA value, fit a curve, and the peak of the curve is the optimal AFMA setting. (The details of the test setup are covered lots of places.) Before FoCal came along, I had done that with a MATLAB script, but it needed TIFs, had no GUI, etc. - FoCal is easier.

 

[pdirestajr]

I think you guys are missing the big "A" on the side of these lenses... for ART! Perhaps the focus issues are for artistic effect!

 

[Loswr]

I would call that 10 shot sample...

10 shots shown.

[quote author=TDP Review]
The first 5 and last 5 images from this particular test are presented below and are representative of the larger test group.
[/quote]

Regardless, the real issue is this is n=1 lens. Maybe this lens was the best copy Sigma had made, hand picked by the CEO for a prominent reviewer. Maybe it was a lens randomly selected by some workaday employee. Maybe it was hand picked as an act of revenge by some unfortunate employee as his last official act on the day he received termination notice. We don't know.

Through many years and many reviews, Bryan has consistently demonstrated honesty and integrity (albeit with a bit of a 'rose-colored glasses' viewpoint). I trust that he did the testing accurately and well. If (best written IF) those results, and Lenstip's findings, turn out to be representative of production lenses, it'll be an issue. For now, it's anecdotal data, and should be considered as such.

 

[jdramirez]

SoI was afma'ing my 85 f1.8 and I noticed that the plane of focus didn't go in the direction I thought it would... but that wasn't all over the place like this suggests.

All things considered, I do want consistency in my auto focus... I don't want to treat it like it is a manual focus lens, and I hope the test is flawed.

Simply not using a wired shutter release might make a difference... though that is grasping at straws.

 

[BLFPhoto]

The best we lay testers can do is automated testing using products like FoCal. The tool allows you to not only perform repeatable cycles of testing, but allows you to specify variables such as how the lens is de focused between shots. Multiple test runs, altering the variables can help pinpoint possible performance limitations or at least give you further insight into the character of your copy of a particular lens. Over time you can also compare results to note changes or anomalies in the performance that may indicate a need to travel back to the manufacturer for service. One of the key parameter settings for Focal is for mirror lock up, probably alleviating your need for a wired release to control vibrations.

I have no experience yet with the LensCal product. Maybe someone has a comparison.

 

[BLFPhoto]

Regardless, the real issue is this is n=1 lens.

On this we are in 100% agreement.

Let me also be very clear that I'm not questioning his honesty or integrity. Rather, in my line of work, I don't accept anyone's work without seeing the data to back up the conclusions. Otherwise people get dead. Now we're not risking bodily injury here, even though these new 50s aren't our dad's old compact, lightweight primes. But we are talking about a serious knock on a new product that seems somewhat out of character with the current paradigm. If I were Sigma, and I was looking to build on my newly improving reputation, I would not let a lens anywhere near the street that was such a laggard in focus precision, especially on current flagship cameras of the big OEM players. Certainly I would not release it with the fanfare this lens is getting. I'd slap a Quantaray silkscreen on that puppy and let it die a death in kit lens hell.

Me? I'm just thinking I'd love to dig in on the at sample shoot and do some root cause analysis. I guess that's the engineer and test geek in me, though.

 

[jcleare]

Can't wait for a 24-70 2.8 art !

 

[sanj]

I think you guys are missing the big "A" on the side of these lenses... for ART! Perhaps the focus issues are for artistic effect!

Brilliant! Many believe that some Canon lenses are soft for artistic effect. I am so confused these days that I am adding Coke to my scotch. Milk into green tea.

 

[epsiloneri]

Me? I'm just thinking I'd love to dig in on the at sample shoot and do some root cause analysis. I guess that's the engineer and test geek in me, though.

Sounds like you'd do a good job testing the AF thoroughly if you got the chance. Please share your findings if you do.

 

[epsiloneri]

Now, for this lens o be useful even without being AF perfect it will have to be a superset of Samyang 24 1.4

1. NO Coma
2. No decentering issues.
3. Very good center wide open with good corners.

Judging from the image quality tool at TDP, the Samyang 35/1.4 seems optically quite the match to the Sigma 35/1.4A, with both better than the EF 35/1.4L. The Samyang 24/1.4 though looks quite a bit worse than the EF 24/1.4L II wide open (though COMA is hard to infer since not explicitly tested). This hopefully means that there is room for improvement for a future Sigma 24/1.4A over the corresponding Samyang. (and yes, the EF 24/1.4L II unfortunately has terrible coma as I know from first-hand experience)

 

[jdramirez]

Can't wait for a 24-70 2.8 art !

Nope... but the rumored Sigma 24-70 f/2 is very intriguing.

 

[Viggo]

I'll just wait until my copy arrives, and test myself. But I'll try it in store and if my results matches those Bryan had, I'm not buying it... Plain and simple. There are endless discussions over the Sigma AF, and when people refuses to realize there is an issue it's no point. If there are great or horrible results, I'll be sure to post my findings and dilbert won't believe them.

 

[jdramirez]

I'll just wait until my copy arrives, and test myself. But I'll try it in store and if my results matches those Bryan had, I'm not buying it... Plain and simple. There are endless discussions over the Sigma AF, and when people refuses to realize there is an issue it's no point. If there are great or horrible results, I'll be sure to post my findings and dilbert won't believe them.

I forget where I heard it (other than everywhere)... but once is an accident, twice is a coincidence, and three times is a trend. It all boils down to sample size. Maybe Bryan did everything perfectly a just got a bad copy. Maybe the UPS guy played soccer with the box before it got to his front door. But if you and 14 others come to the same conclusion... then what we have here is a cheaper otus with AF usable only at f/4 or f/5.6... which REALLY makes the lens less desirable.

And there is nothing wrong with questioning scientific fact (which this test would not be considered).