Panasonic 14 mm f/2.5 ASPH - Lenstip Review

Franka T.L. said:

respectfully I disagree, well Photoshop or Photoshop based product from Adobe do correct those, but almost all others do not ( bar the Mfr's own ) Even Silkypix Studio PRO ( Silkypix is used by Panasonic ) by default do not apply the correction. The like of Picasa, Photo.net ( 2 very common freeware for casual hobbyist ), and even Aperture do not. Capture One and DxO , 2 high end RAW developer also by default do not applied the correction. Even the camera themselves differ , putting the Panasonic lens on Olympus body yield different way of correcting ( or not correcting ) the CA / Vignette ( which surely can be confusing to the M4/3 users ).

Click to expand...

OK, lots of confusion here. Just to be clear, the following converters do auto-correct distortion, at least in their current versions:

Adobe products (Photoshop, Photoshop Elements, Lightroom)
SilkyPix (as supplied with Panasonic cameras)
Olympus's software
Aperture
Capture One and Capture One Pro
DxO Optics Pro

According to our recent survey, this means over 90% of users will get lens corrections completely seamlessly, as long as they use current versions of the software. (Hopefully it should be clear by now that you'll have to update your software to the latest version for raw support whenever you buy a new camera no matter what brand or model, with only a very few exceptions that can output DNG.)

The only software you've named that definitely doesn't apply corrections is Picasa - I simply don't know about Photo.net.

The only difference in corrections when using a lens on a different body brands is that Panasonic bodies additionally correct for lateral CA with Pansonic lenses, while Olympus bodies do not. Neither brand body can correct CA with any existing Olympus lenses (as the parameters apparently aren't stored in the ROM).

There's no evidence of any vignetting correction being applied in the Micro Four Thirds system at all.

--
Andy Westlake
dpreview.com

One argument I uphold: we are sold cheap consumer lenses for their weight in gold.

There is one exception: the Panny 20mm, and the 7-14. The 20mm has natively little distortion, the 7-14 is almost as good as the 4/3 Oly equivalent.

OTH I can confirm that my Oly 17mm (corrected) is almost as good as my Oly 25mm, but not as sharp, although they cost the same.

The Micro 9-18 is more expensive than the classic 9-18, but has worse edges. We all know that the main advantage is in size.

So what would be the conclusion when next year Oly will introduce HG lenses, as announced?

I bet it will be a moderate focal zoom, with as little as possible distortion to be firmware corrected.

Indeed, from my own experience in correcting fisheyes to linear I know that software correction of distortion is not a lossless operation: quite the opposite. You lose in sharpness to such a point that you must crop heavily.

So to me firmware correction is a cheap workaround for the same things that optics engineers were able to achieve perfectly, though at a cost and with bigger groups of lenses.

If I was given a heavily distorted wide, corrected by firmware, I would first compare its price with its Cosina equivalent. After all the latter might be used with adapters on any other brands with a compatible register. So it might be much more future proof.

I assume that small consumer lenses will keep being firmware corrected, but that expensive HG lenses, will have as little native distortion as possible. Not everybody is stupid or clueless, when investing big money.

Therefore a review site must show what the physical distortion is in order that a customer can take an informed decision. Firmware correction is not lossless as some would have it. Stressed pixels, especially at the edges, are a documented problem.

Am.
--
Photostream: http://www.flickr.com/photos/amalric

Every piece of electronic equipment we all use every day makes massive use of software correction, for very sound engineering reasons, so why should its use in camera lenses be singled out for special treatment? Maybe "troll" is a bit harsh, but you are not making a reasonable assessment of the performance; you are making some sort of personal value judgment about Panasonic's engineering approach to m4/3. Comments such as "no software can retrieve lost details, which just haven't been recorded by the detector itself" seem like classic attempts to divert the argument, and remarks like "Sure, corrected image is pretty" indicate to me that you don't have a good understanding of the theory behind digital imaging or you are being deliberately obtuse.

IMHO Andy Westlake is quite correct in his comments about your review, and in my eyes the nature of your response makes me think that, if you represent Lenstip, then my opinion of their reviews is considerably diminished; hence my question.

Apologies if it offended you; but down here in Oz we tend to be plain spoken.

S.

--

• Franka -

I don't see anything wrong with pointing out the true optical nature of the lens being tested--dpreview has done the same thing in their micro-4/3 reviews.

However, their approach has been to analyze the performance of the lens as it was intended to be used, and as it will be used by 99% of the people who might buy it. That is, they focus on the corrected behavior, and the uncorrected behavior is discussed in passing.

The lenstip review is coming at it from the other direction; it makes a big deal of the uncorrected performance, discusses the performance with corrections only minimally, and ends with conclusions whose tone suggests that the lens is quite poor.

Since I'm one of the 99% who will be using the lens as it was intended and who just want a good result in the end, I find the dpreview approach far more useful than the lenstip one. If lenstip wishes to be relevant for my purchasing decisions, I think they should also take the dpreview attitude--review the as-designed performance, and treat the uncorrected performance as a side issue.

no text

siward said:

Every piece of electronic equipment we all use every day makes massive use of software correction, for very sound engineering reasons, so why should its use in camera lenses be singled out for special treatment? Maybe "troll" is a bit harsh, but you are not making a reasonable assessment of the performance; you are making some sort of personal value judgment about Panasonic's engineering approach to m4/3. Comments such as "no software can retrieve lost details, which just haven't been recorded by the detector itself" seem like classic attempts to divert the argument, and remarks like "Sure, corrected image is pretty" indicate to me that you don't have a good understanding of the theory behind digital imaging or you are being deliberately obtuse.

IMHO Andy Westlake is quite correct in his comments about your review, and in my eyes the nature of your response makes me think that, if you represent Lenstip, then my opinion of their reviews is considerably diminished; hence my question.

Click to expand...

I actually find it quite odd that AW has come out and defended Panasonic so vigorously. Dont you? Makes you wonder about the impending 14/2.5 review.

siward said:

Apologies if it offended you; but down here in Oz we tend to be plain spoken.

S.

Click to expand...

Jogger said:

I actually find it quite odd that AW has come out and defended Panasonic so vigorously. Dont you? Makes you wonder about the impending 14/2.5 review.

Click to expand...

1) I haven't defended Panasonic in this thread. I've corrected a serious factual error (claiming the lens is a 31mm equivalent) in Lenstip's review, and pointed out why I think Lenstip's approach to Micro Four Thirds is unhelpful to readers.

2) Nothing I've said is in any way different from what I've been saying, and doing, since I first reviewed a Micro Four Thirds lens. Which, in a nutshell, is that because the overwhelming majority of users will be using distortion-corrected images, we'll review the lenses on that basis.

If you've read everything I've writtten in this thread properly, you'll also have noticed that there is no impending 14/2.5 review.

--
Andy Westlake
dpreview.com

DjarumBlack said:

LOL. I'm a software engineer. You don't think I know what it costs to make and design software? I even design hardware. However, I know that part making quality optics is not just in the design. Its also in the procurement and manufacture of quality glass on a consistent bases. Part of this process gets put into the price of optics. Software, on the other hand, does not have such limitation nor such cost.

Click to expand...

I didn't mean you specifically, I just see many comments in the thread that people don't want to pay a lot of money for the software part even if that is a major factor in final image quality. I have worked several years in both hardware and software development, so I know both can be expensive.

My point is: manufacturers cannot always pass through the real cost to the customer. Sometimes they sell the basic product very cheap, hoping you will buy additional products that allow them to make more money.

A camerabody could have a very fast processor with advanced software that can strongly improve image quality in-camera.; probably even more so when new technologies like variable refractive index materials or liquid lenses are used. Maybe selling the body at real cost + margin would reduce the customer base too much, and selling a body with only 'offline' advanced optics corrections would not be acceptable to the average customer. I such a case lenses could be relatively cheap to build but expensive for the customer, because that is the best way for the company to make money on the product line. As long as the final quality matches the price, I would not have a problem with that.

Andy Westlake said:

If you've read everything I've writtten in this thread properly, you'll also have noticed that there is no impending 14/2.5 review.

Click to expand...

That doesn't help me.

you have a point there, but keep in mind that the environmental impact of "cheap bits of plastic with chips " can be much lower on the production side than those from traditionally manufactured products. Even more so when talking about some types of metal mounts and special glass like fluorite elements

Andy Westlake said:

Where this approach falls over, though, is the moment you attach a new lens whose parameters aren't in the camera's database. You have to load a new profile into the camera with each new lens. Fine if you're near a computer at the time, less so if you're in the field taking pictures.

Click to expand...

maybe in a few years when changing lenses you can get a pop-up message on the LCD saying "new lens found, getting lens profile over Wifi network" or something similar

but maybe the camera manufacturers would prefer if you buy a new body with up-to-date lens profiles instead ...

sacundim said:

We don't know how much research and testing it takes to validate that a set of proposed correction parameters are in fact going to work well enough with real lenses made out of real glass. We don't know what were the initial costs for developing all the software and hardware in question (not just the stuff that goes in the lens and the camera, but also any in-house software and equipment they may have needed to develop to design these lenses). We don't know how much it costs to train people to design, prototype and test lenses in this new fashion.

Click to expand...

Yes, I'm also wondering about the constraints when the system depends strongly on software correction. We know from lens tests that a good percentage of lenses have decentering problems, probably even more so with small lenses (judging from experience with small digicam lenses). This product variation must influence how effective software correction is, and maybe it means they have to use optical designs that are relatively insensitive to decentering or other variations that might skew the processing results from copy to copy.

palindrom said:

I understand arguments for the software correction. It's good in normal life situations as it generates pretty pictures and result in smaller, lighter lenses. Panasonic asks for a load of money for a small, plastic, 5 elements objective. Almost 600 USD in my country. What do I get in that price? Well, one can say that objective + software will give me good results. I say that no software can retrieve lost details, which just haven't been recorded by the detector itself, nor increase tonal range of regions that needed to be brightened by 50%. RAW picture (detector output), is all camera is capable of. The vertical limit of what can you get. Sure, corrected image is pretty, but it doesn't tell anything about the true capabilities of the whole device. Saying that software correction improves picture requires specifying in what aspects, or it's a lie. If it was true, no one would buy µ4/3, Canons and Nikons 135 dSLR because cellphone camera with appropriate software would outperform them, right? Just because something look pretty, doesn't mean that it's overall better. It isn't.

Click to expand...

It's not a discussion whether software correction is good or bad. It simply is a part of m4/3. We as users always deal with corrected images, unless we go out of our way to bypass the correction. At every stage in our photographic workflow we deal with corrected image. For us the corrected image is what the lens is capable of. And this is what we are interested in and what we want to see tested in your reviews.

palindrom said:

Lens itself is and will be the most important element of the camera, the final limit of image quality, and that's why all the tests must be made on the real, physical picture, that it generates. This is also the only scientifically correct approach for lens testing and the reason why tests on lenstip looks that way.

Click to expand...

If you can show that bypassing the correction and then correcting in PP significantly improves the image quality compared to default m4/3 processing, then the tests on uncorrected lens could be interesting. Otherwise, it might be scientifically correct, but completely useless to your readers.

Jon Rty said:

What I'm trying to say is that it's not especially tasking to write a decent distortion profile for a prime lens. Panasonic or DxOlabs are surely better, but even I can make a profile that I'm happy with.

Click to expand...

distortion and vignetting for prime lenses are relatively easy to correct, but with CA and possibly other corrections in future, things get more complicated especially with zoom lenses with a big zoom factor. I have seen some lenses where basic CA corrections (assuming linear increase in CA from image center) don't work.

amalric said:

from my own experience in correcting fisheyes to linear I know that software correction of distortion is not a lossless operation: quite the opposite. You lose in sharpness to such a point that you must crop heavily.

Click to expand...

No one is arguing that software correction is lossless.

What seems to escape you, however, is that the optical correction of aberrations is not lossless either.

So the bottom line is - measure the end result of the engineering solution, and critique that. Compare the losses, detail the resolution. What you will find is that well implemented software correction can produce as good or better images than optical correction.

amalric said:

So to me firmware correction is a cheap workaround for the same things that optics engineers were able to achieve perfectly, though at a cost and with bigger groups of lenses.

Click to expand...

You are loading your argument with pejorative statements here;

"optics engineers were able to achieve perfectly" - no lens is perfect, and optical correction of distortions creates other distortions.

"firmware correction is a cheap workaround" - You are placing the value on lenses and cheapening the value of software, but without any proof or qualification. If the 'cheap workaround' produces the same quality image at the end, then it is a better engineering solution than the heavy, expensive old fashioned solution.

amalric said:

One argument I uphold: we are sold cheap consumer lenses for their weight in gold.

Click to expand...

Frankly, I couldn't care less if Panasonic used the bottoms of old Coke bottles as lenses, then applied software correction, as long as the end result is a good image, it matters not a jot.

Of course, that is a ridiculously extreme example, but really, what is it we are doing here?

Yet more pejoratives from you - "cheap consumer lenses" - when the lenses are well made, small and light, and produce very good images.

Metal and glass has been replaced with high impact plastic and software - get over it FFS!

--
http://www.flickr.com/photos/30225435@N00/

technic said:

maybe in a few years when changing lenses you can get a pop-up message on the LCD saying "new lens found, getting lens profile over Wifi network" or something similar

Click to expand...

It would happen silently unless Microsoft implemented it, then you'd have to click a zillion "ok" buttons in endless dialogs before anything useful actually happens.

Anyhow, lens-eating fungus is bad enough without adding Windows-eating worms and viruses as well.

technic said:

but maybe the camera manufacturers would prefer if you buy a new body with up-to-date lens profiles instead ...

Click to expand...

Or just do the sensible thing and keep it in the lens.

--
John Bean [GMT]

Jon Rty said:

To be honest the DPreview test of the Sony 16mm doesn't look much worse than the Lenstip 14mm test, especially considering that the Lenstip MTF curves were measured before the distortion correction. But as the methold and numbers are different we'll have to wait for either DPreview to review the 14mm, or Lenstip to review the 16mm to be able to compare. Both are good in the center but nothing to rave about in the corners.

Click to expand...

judging from the image galleries the Sony 16mm is worse outside the center at larger apertures; but for its much higher price the Panny 14mm is a bit disappointing as well.

Lenstip is right in explaining the importance of pancakes for this type of camera in their test (important for many users, not for everyone of course). The whole idea of compact camera, big sensor and compact lens seems to run into a bit of trouble, at least for WA range (which is more important to me than standard lenses). I wonder what the announced m43 12mm Olympus will be like (is this a pancake design?).

palindrom said:

I understand arguments for the software correction. It's good in normal life situations as it generates pretty pictures and result in smaller, lighter lenses.

Click to expand...

"pretty pictures"? Just say good or great images, don't patronise.

palindrom said:

Panasonic asks for a load of money for a small, plastic, 5 elements objective. Almost 600 USD in my country. What do I get in that price? Well, one can say that objective + software will give me good results. I say that no software can retrieve lost details, which just haven't been recorded by the detector itself, nor increase tonal range of regions that needed to be brightened by 50%.

Click to expand...

Don't cry about lens prices, if you don't want the lens, don't buy it. The market will decide the outcome.

Regarding your other points, these things can be measured AFTER the engineered solution has been fully gone through, and the results can speak for themselves.

Software cannot retrieve lost detail, but neither can sensor retrieve lost detail from optical corrections.

Measure the whole system as it is intended to be used, please!

palindrom said:

Sure, corrected image is pretty, but it doesn't tell anything about the true capabilities of the whole device.

Click to expand...

What? Do you have a 'pretty' measurement? Really, this patronising tone is demeaning and puts you and your site in a very bad and unprofessional light.

--
http://www.flickr.com/photos/30225435@N00/

nikclick said:

Software cannot retrieve lost detail

Click to expand...

not yet... In the lab one can compute a lot of image detail from a few defocused images, when the lens characteristics are known (which could be defined after lens assembly). Software can also be used to sharpen images that were blurred due to camera shake (requiring knowledge of lens characteristics and camera movement for best result). One day some of this may be available in-camera.