Friday, September 30, 2016

Irix 15mm Firefly Review

As I mostly use my Canon 16-35 f/4 IS at the wide end of its zoom range, I began wondering if I might be served better by a prime. The Irix 15mm is one of the few options available and is inexpensive enough to give it a try. While not sold in the United States as of this writing, ebay and Fototip from Poland came through for me.

Build Quality

I would consider the build quality very good.  I bought the firefly version for weight reduction.  It is plastic, but feels solid.  I would place the build quality somewhere between that of the modern plastic Canon L lenses and Canon's non-L lenses.  The Irix Firefly is weather sealed (but not on the front element like the Blackstone version).  Overall I have no problems with the build quality.


There are several features that distinguish the Irix from its closest competition, namely the Samyang 14mm f/2.8 and to a lesser extent, the Zeiss 15mm f/2.8.  First, it has a locking ring for the focus.  This is particularly nice for a manual focus lens where obtaining critical focus can take precious time while shooting.  There was one minor flaw here.  The focus ring would drag the lock ring along with it slightly.  Not a big deal, but annoying.

Second, it has 95mm threads for using proper front filters.  If you want to use a polarizer on a lens wider than 16mm, you are choosing between the Irix and the Zeiss.  That being said, there are many great options slightly longer than 15mm that take more normal 77mm filters.  95mm filters aren't cheap and the thread size isn't common outside of long tele lenses.

Finally, there is an indent in the focusing for infinity focus.  This makes it easier to set fast and precise focus while shooting through the viewfinder or in dark conditions, as long as that focus is infinity.  There is also a panel to adjust the focusing ring for critical applications.  I suspect few people will actually need to take advantage of this, but the few that do will be very appreciative.

Image Quality

I am not going to try to repeat technical measurements reported elsewhere that I have no ability to make.  The Irix has very low distortion for an ultra-wide.  It is mild barrel distortion, which is very easy to correct in software if necessary.  Vignetting is pretty severe wide open and becomes benign by f/8 but doesn't completely disappear.

Vignetting at f/2.5

Vignetting at f/8.0
Sharpness is enough for perfectly usable images at f/2.5 and excellent across the frame at f/8.0.  All shots are 100% crops from a 5Ds.  These are shot in fine detail picture mode and no sharpening has been none in post.

Center sharpness at f/2.5 (left) and f/8.0 (right)
Mid-frame sharpness at f/2.5 (left) and f/8.0 (right)
Extreme corner sharpness at f/2.5 (left) and f/8.0 (right)
Compared to the Canon 16-35mm f/4 IS, the Irix appears to be a little bit worse in the center of the frame and a little bit better on the mid-frame and edges.

Irix 15mm (left) vs Canon 16-35mm f/4 IS @ 16mm (right) both f/8.0.  100% crop taken from mid-frame.
This comparison should be taken with a grain of salt.  The differences were easily within the margin of error of the focusing placement.  I mean only to say here that the Irix is comparable to the Canon, which we already know is an excellent lens.


I am impressed.  This is a lens that will stay in my bag and probably replace my Canon.

-Build quality
-Image sharpness
-Option to use front filters
-Focusing Lock
-Infinity focus indent

-Cost of 95mm filters

Full Set of Sample and Test Images

The Canon 16-35mm f/4 IS is terrible at 35mm and Infinity Focus

I have hiked my Canon 16-35mm f/4 IS hundreds of miles and it has given me many of my favorite images.  I have mostly used it on the wide end since I usually carry a normal range zoom as well.  On a recent trip, I took only this lens and discovered just how poorly it performs at 35mm and focused at infinity.  Here is a center crop at f/8 and 35mm shot on a 5DS:

No problems here.  About 75% to the edge of the frame, however, is a disaster:

I am fairly careful about picking lenses that will perform well enough on a 50 megapixel camera, so I was pretty surprised.  Now what went wrong here?  Lets look at Canon's MTF charts:

Image result for canon 16-35mm f4 mtf

These charts would suggest really good performance across the entire frame, clearly very different that what I was seeing.  It is possible that I simply have a badly decentered copy, but in that case one side of the frame should be better than another, which is also not the case.  Another possibility is that these MTF charts are for a close focal distance and that when focused at infinity, the lens performs entirely differently.  The-Digital-Picture hosts MTF charts for a variety of lenses tested at infinity.  The 16-35mm @35mm and f/8 performs as:

Credit to the-digital-picture and Olaf Optical Testing supported by LensRentals

There we go, mystery solved.  The 16-35mm @35mm and f/8 is simply bad in the tangential MTF at infinity focus and it is readily apparent in photographs.  This is not the first time I have been screwed by Canon's lenses being optimized for closer focal distances.  From a marketing perspective, this makes sense.  The vast majority of optical testing is done by taking images of test charts.  The moral of the story is that one measurement tells you very little about a lens 

Thursday, June 25, 2015

5DS vs 5DS R Accutance, Resolution, and Artifact Comparison.

   Canon's new high resolution cameras, the 5DS/R have reignited the debate over the virtues and cost of having an optical low pass filter (OLPF).  Thus far the posted samples appear to show a clear sharpness and resolution advantage for the R version.  I have been wondering how much of this apparent advantage is real resolution and how much is simply a difference in accutance that can be corrected for in post processing.  With DPreview posting studio samples from the cameras, there is now a high quality comparison available.  I have opened the raw files directly into Lightroom and only adjusted the sharpening as specified below.

   The first images are the raw files loaded directly into Lightroom with the default sharpening setting of Amount = 25.  The crop from the studio scene has been chosen deliberately to emphasize both resolution and color artifacts.  Neither effect should be as strong in most real-world images.  Clearly the SR version has higher accutance and displays stronger color artifacts as expected.  Interestingly, the S version isn't completely artifact free implying that the OLPF isn't strong enough to completely eliminate aliasing.
Left: 5DS Sharpening = 25.  Right:  5DS R Sharpening = 25.
   The second set of images has a bit of sharpening applied to the S version.  Clearly a little sharpening goes a long way and increases the accutance and thus the apparent sharpness significantly above the SR version.
Left: 5DS Sharpening = 50.  Right:  5DS R Sharpening = 25.
    In this third set of images, has equal sharpening applied to both images.  The SR again leads in accutance but has become over-sharpened.  To come back to an appropriate level for a print, the sharpening needs to be backed off.
Left: 5DS Sharpening = 50.  Right:  5DS R Sharpening = 50.
   This final set of images has the sharpening set such that the accutance looks roughly equal to my eye.  Unfortunately this is rather subjective.
Left: 5DS Sharpening = 50.  Right:  5DS R Sharpening = 35.
The differences between the two images are now pretty subtle.  Upon close inspection there are some apparent details that are clearly visible in the SR version that do not show up as well or as clearly in the S version.  The SR also shows stronger color artifacts, as expected.  In the end the difference between the cameras is splitting hairs.  The S version seems to tolerate sharpening better, but also needs a bit more of it.

Personally, I am willing to give up a slight amount of detail in exchange for not having to deal with as much aliasing or as many color artifacts.  As has been demonstrated by the popularity of AA-less cameras over the last several years, others may find the SR version better for their needs.

The Image That Convinced Me to Anti-Alias

    As a Canon user, I watched cameras from Sony and Nikon outresolve my Canons by virtue of higher pixel count and lack of an anti-aliasing filter / optical low pass filter (OLPF).  I had assumed that Canon kept the OLPF around for their professional customer base but that I, as a landscape photographer, would not encounter the fine mesh patterns that the OLPF is needed for.  This sample image from Canon changed all that:
If we zoom into the reflections on the side of the hippo, there are severe color artifacts.
Crop from above.  Note the red, green, and blue color artifacts scattered in the reflections.
    The level of detail and sharpness that the AA less 50MP sensor provides is incredible, but the color artifacts render the image severely damaged.  A different demosaicing algorithm could potentially eliminate the artifacts, but not without significant loss of real detail as well.  The potential for color errors is not limited to man-made subjects, but can occur in images of any kind.

    It is worth noting the conditions in which color artifacts are most likely.  The subject must have high contrast details at a high spacial frequency and the optical system must be good enough to properly resolve those details.  Specifically, the lens must be perfectly in focus, at an aperture wide enough to limit diffraction, and sharp enough at said aperture.  The hippo shot was with a 500mm f/4 L IS II @ f/4 and so certainly meets those criteria.

   I am now grateful to be presented the choice of an OLPF and am asking myself "is the small increase in resolution worth the chance that a portfolio quality image would be ruined by aliasing or artifacts?"  Sadly for the pixel-peeper in me, I think not.

Monday, June 22, 2015

A Reasonable Backpacking Tripod

While hand-holding can work just fine for most general daylight photography, a tripod is critical as the light fades and the best photographic opportunities arise.  A tripod is also necessary for many techniques used to enhance final image quality including but not limited to HDR, nodal point panoramas, and focus stacking.  Unfortunately for our purposes, tripod stability is highly inversely correlated with tripod heft.  Thus my goal was to find the absolute lightest setup that will reasonably hold a full frame DSLR with an attached f/2.8 zoom.

In addition to an excellent strenght/weight ratio I look for twist leg-locks and an Arca-style quick release for speed of deployment and a bottom hook for hanging weight.  My solution consists of the following:

Oben CT-2331 - 26.4oz (
Really Right Stuff BH-25 - 4.1oz  (
Sunwayfoto DDC-37 - 1.55oz (*

*This Arca-style clamp has an M5 mounting hole that needs to be drilled out for compatibility with 1/4-20.

After removing the mounting platform from the tripod and screwing the head directly onto the center column, the total weight comes to 1 lb, 12 oz.  This gives the tripod the following specs:

Height without center column extended -   47"
Height with center column extended-         56"
Folded length-                                          21"
Weight-                                                   1.75 lbs
Manufacturer weight rating-                       8.8 lbs / 4 kg

Using the typical rule for tripods of:

Practical weight rating = Manufacturer weight rating / 2

this tripod should hold the 4 lb weight of my 5DS and 24-70 F/2.8.  There is only one way to find out for certain and is a good excuse to play around with camera equipment.  Lab test!  For the following test I shot a test chart with the 5DS and the 24-70 f/2.8 ii @ 70mm f/5.0, ISO 100, 3.2s exposure.

This is an 8:1 crop.  We can see that critical sharpness has been achieved.  However, I did have to use a cable release or >0.5s mirror lockup with the on camera shutter release to achieve stability.

Note that there are some subtle color shifts in the image.  These are imaging artifacts and not real despite the anti-aliasing filter on this camera.  Intriguing, but the subject of another post. 

The number of tripod and ballhead choices out there are absurd compared to even ten years ago.  I could not have researched every possible combination so please share what you have come up with!

Thursday, June 18, 2015

I Shoot Canon ... For Historical Reasons

I will make repeated references to the equipment I am using in various tests and experiments.  This will invariably some sort of Canon DSLR and lens combination.  I expect the reader to have the following instantaneous befuddlement: "If this guy is so interested in maximizing image quality and reducing weight for non-moving subjects, WTF is he doing using Canon DSLRs???"  Good question.

The answer is a mix of cost considerations and historical reasons.  I began shooting with Velvia and a Canon Elan 7e way back when.  My lens collection grew from there to where I find myself now with a beautiful set of Canon glass.  I currently shoot with a 5DS, 24-70 f/2.8 II, 16-35 f/4 IS, 70-200 f/4 IS and some assorted primes and older bodies.  It is simply not economically viable, and in some aspects not possible, to switch systems and maintain such a high level of image quality.

I also want to diffuse any notion that I am religious about my choice of gear.  There are at least a dozen camera systems out there that when used with good technique and vision, will produce phenomenal results.  If I were to start from scratch today, I would likely end up in the Sony A7xx/FE mount universe.  The cameras are light with superb image quality.  The few available lenses are all of high quality.  But the equipment I have is also fantastic weight difference of a pound just isn't enough for me to switch.

Searching for the Gigapixel Camera Phone

Two of my favorite hobbies, backpacking and photography complement each other exceptionally well.  Except in one department, weight.  Image quality tends to scale with camera size.  Bigger sensors have lower noise and higher per-pixel resolution.  Larger sensors require heavier lenses.  Heavier lenses demand beefier tripods.  Conversely, every ounce taken off of the pack is that much more enjoyment on, and mileage of the trail.  This blog is simply a documentation of my efforts of optimization and studies in compromise.