Rokinon 8mm vs. 10mm vs. Sigma 8-16mm vs. Pentax 10-17mm Ultra-wide Showdown
Distortions
Following angle of view analysis, this section will be dedicated to assessing distortion. Before we get to the direct comparisons, however, there are two things we wanted to direct your attention to as you may have missed it while wading through the many scenes on the previous page.
First, we'll revisit the overlays we are now all familiar with:
 |  |
While creating the individual bands, the one that proved the most problematic (and by far!) was the DA 10-17 FE at 10mm. This is because the Pentax zoom exhibits a bit more than just the expected barrel distortion. Instead, at its widest setting there is a noticeable amount of mustache distortion—referring to how the corners exhibit a bit of a curl—even when compared to the other fisheye.
Second is something that you probably missed despite scrolling through all of the examples on the previous page (we won't hold it against you as it took us a while to notice, too). It has already been widely established that the edges of the frame will bear no resemblance between the different lenses, especially when you pit the two projection types against each other. But what we weren't expecting is that objects in the center will be rendered exactly the same for any given focal length, to include size and between fisheye and rectilinear lenses!
Revisiting several of the side-by-side comparisons from the previous page, the below series highlights those center portions that are identical per focal length. The center portions highlighted in blue refer to the 8mm lenses and that of the 10mm lenses are identified in orange:
 |  |  | |||
 |  |  | |||
 |  | ||||
Beyond a shadow of a doubt, it is clear that the "wide angle magic" takes effect at the periphery of the frame whereas the centers are indistinguishable so long as the focal lengths are the same. Going back to cartography, this is exactly what we would expect to happen: the "center" of the earth (where the equator and Prime Meridian intersect) is the same regardless of the projection type while the same cannot be said for the edges of the map (i.e. edges of the frame).
Our methodology hasn't changed from the previous page:
|
|
nor has the arrangement of the comparisons:
 |  |
Comparison One - Beach Horizon, Center
For our first foray into deliberately assessing distortion, we wanted to check what would happen when the horizon was perfectly centered across all the frames. A Virginia Beach sunset provided the setting for the next few comparisons:
The full set of four RAW images can be downloaded here (102 Mb).
Comparison Two - Beach Horizon, Quarter Up
As you can see in the above, a straight line going through the center of the frame, regardless of the lens, is kept that way. But how does that change if we shift the frame just enough so that linear edge now reaches only a quarter of the way up the frame?
The full set of four RAW images can be downloaded here (105 Mb).
Comparison Three - Beach Horizon, Quarter Up Vertical
As expected, the fisheye lenses couldn't resist warping the horizon, and the same is true if we rotate the camera to portrait orientation:
When taking shots like this, it's especially critical to be aware of all the elements within the frame, especially at the bottom with fisheye lenses since the field of view is so wide.
The full set of four RAW images can be downloaded here (99 Mb).
Comparison Four - Beach Horizon, Bottom
For our final assessment of how the lenses react to the horizon line, we went to the extreme and placed it at the very edge of the frame:
As you can see, the "fishyness" is very high in the top row while even at the extreme bottom, the rectilinear lenses have a minor amount of curvature. This is especially true for the 10mm prime lens that is in fact more tilted—due to difficulty composing the image—than it is distorted.
The full set of four RAW images can be downloaded here (104 Mb).
Comparison Five - Huske Hardware House Upstairs
Our next scene brings us back to this unique restaurant. Making our way upstairs, we instead oriented the camera directly at the railing to stress the lenses with both vertical and horizontal elements across the entire frame:
The full set of eleven RAW images can be downloaded here (308 Mb).
Comparison Six - Cloudy Harbor
In the following scene, we found the comparison to be a good demonstration of two components of distortion between these lenses. First, the large white building on the right side of the frame is stretched significantly on the rectilinear lenses whereas all of its components (windows, doors, etc.) are all in much better proportion to how they appear in real life.
Second, if you take a look at the top right corner of the fisheye images, pay attention to the light pole. Remember how, while discussing the overlays, we said the Pentax zoom seems to be cheating in order to get to the 180 degree field of view? This is a great example showing how the 10-17 really squeezes what it can into the corners, made evident by how compressed that light fixture is as opposed to the much more natural one (unnatural pole curving aside) from the Rokinon FE.
The full set of four RAW images can be downloaded here (98 Mb).
Comparison Seven - Vacant Mansion
A different humble abode than the one on the previous page, this one was shot from the outside. Unfortunately, while hastily culling images prior to importing into Lightroom, the image from the Pentax zoom was deleted! Instead of eliminate the entire series, we wanted you to see how the two rectilinear lenses treated this vista as compared to the Rokinon fisheye:
Knowing what we know now of the two fisheyes and how they compare, it's not too difficult to imagine what the Pentax would have looked using the Rokinon FE and the size of the elements within the Rokinon 10mm's center as a reference.
The full set of three RAW images can be downloaded here (74 Mb).
Comparison Eight - Mansion Pillars
Moving towards the mansion's entrance, we used the tall pillars to frame the following view of the front facade:
The full set of four RAW images can be downloaded here (101 Mb).
Comparison Nine - Binn's Changing Rooms
The following comparison returns us to Binn's Department Store, but this time we found the narrow hallway of its second floor changing rooms to be a great place to see how these lenses dealt with such a tight space. And no, don't worry - no disrobing occurred at this time!
The full set of eleven RAW images can be downloaded here (304 Mb).
Comparison Ten - Beach Picnic Area
The next location is quite similar with its framing and vertical structures, however with a bit more breathing room between those elements as opposed to a narrow hallway:
It appears that regardless of how far those "uprights" are from the camera, in this case door frames and support pillars, they exhibit the same behavior above and below the center line save for "Prime Meridian" of the frame.
The full set of four RAW images can be downloaded here (90 Mb).
Comparison Eleven - Windswept Beach Trees
On a nicer day, we revisited the same area as in Comparison Ten to photograph these trees. Our composition for these four images was extremely deliberate, and as best we could for all four lenses, we arranged them using three parameters:
- Placing the horizon at the bottom third using the K-3's Rule of Thirds overlay in Live View
- Placing the topmost and leftmost branch tips of the tree on the left to the upper left corner of the frame
- Bisecting the distant tree on the right using the rightmost vertical "third line"
And the scene's side-by-side:
The full set of four RAW images can be downloaded here (124 Mb).
Comparison Twelve - World War I Hangar Wings
Back at the Military Aviation Museum, an aircraft from a century ago was also framed in a similar manner as those windswept trees. It was composed so the French tail was kept to the left edge while the top wing of the adjacent aircraft was placed at the upper right corner:
It's interesting to note that the "flat" tail to the left is seen at a sharp angle (i.e. not perpendicular to the lens) because we had to physically shift our shooting position to the right by several meters (and then turn to the left) to place the wing tip directly in the top right due to the wide field of view.
The full set of four RAW images can be downloaded here (87 Mb).
Comparison Thirteen - German Convertible
One last one from the same museum, this beautiful German convertible had a very striking chrome siding running its length which we positioned from the bottom left corner to the center of the right edge of the frame:
The full set of four RAW images can be downloaded here (94 Mb).
Comparison Fourteen - Fort Monroe Cannon
A common usage of ultra wide angle lenses is to emphasize a size and/or distance relationship between two objects. Just outside the moat of Fort Monroe and part of the bunker complex from the previous page are a few cannons that we tested this phenomenon on using the following constraints:
- Placing the opening of the cannon where the center and leftmost grid lines intersect
- Placing the small window where the center and rightmost grid lines intersect
Additionally, for this series the lenses were positioned about a 1.5 feet (.5 m) from the tip of the cannon:
Immediately the 8mm vs 10mm disparity is evident between the two types of lenses when we compare across the rows (fisheye vs fisheye and rectilinear vs rectilinear). We weren't expecting there to be such a difference, however the compression that is inherent with increasing focal lengths is already witnessed as you move from 8mm to 10mm. The most obvious way to see is is the size difference in the cannons' ends as well as how "far away" the armored plating appears to be.
The full set of twelve RAW images can be downloaded here (304 Mb).
Comparison Fifteen - Beach Houses
For this scene we placed the camera in the same location atop a stationary tripod and wanted to see what would happen to the dense row of houses:
The full set of four RAW images can be downloaded here (101 Mb).
Comparison Sixteen - Downtown Virginia Beach
Similarly, we took to a public balcony in downtown Virginia Beach for this shot:
What is interesting is how the buildings in the fisheye shots appear to be bending/leaning backwards as that pair's exaggerated barrel distortion squeezes more into the frame.
The full set of eleven RAW images can be downloaded here (257 Mb).
Comparison Seventeen - For Sale
For this beautiful ocean-front home we did our best to capture actual real estate photos as if we were trying to sell the property. As such, all of the positions were slightly different between shots:
The full set of four RAW images can be downloaded here (116 Mb).
10mm vs 10mm - Sigma vs Rokinon
One thing we were curious about was how does 10mm look on each rectilinear lens, especially since we've compared the two at different focal lengths (8mm and 10mm) thus far. Since we had a few scenes that we shot both at across the entire focal length range, we wanted to compared them directly to see how they fared. First, here's a shot of a brick wall with Lightroom's grid superimposed to better see where the how each lens distorts:
Beyond a shadow of a doubt the Rokinon is significantly better corrected in this scene, most immediately apparent along the vertical portion of the wall in the top left corner.
And here are six scenes we've already visited:
 |  |
 |  |
 |  |
Due to the nature of prime lenses being generally more corrected for distortion than comparable zooms, especially for lenses of these wide focal lengths, we gave the Rokinon the benefit of the doubt and expected it to be pretty clear. While this was the trend, it wasn't the rule. What we found instead was that while more often than not we preferred its rendering over the Sigma, there was the rare occasion where a hint more distortion was exhibited on behalf of the prime.
We thought long and hard as to why that might be, especially considering how much newer a lens design the Rokinon is, and then we discovered something we had never seen before. When you focus the Sigma 8-16, it very visibly distorts the image at varying levels as you traverse between the minimum focus distance and infinity, all while not changing the focal length. The below animation of the 8-16mm zoom traversing its focus range while at 8mm only shows this clearly:
While focus breathing is a well-known phenomenon (where the magnification changes as a lens traverses its focus ring, with the smc DA* 60-250 F4 being arguably the worst offender of it), this may be the first documented case of distortion breathing. We tested the other lenses in this review as well as a few other internal focus ones (both zoom and prime) and found that while focus breathing occurred, the Sigma zoom was the only to exhibit the above behavior.
Verdict
It's difficult to make an object comparison because of how different all the lenses are, especially as you cross the rectilinear-fisheye divide. The Rokinon fisheye has a "cleaner" distortion profile as opposed to the Pentax (which can get pretty crazy in the corners) and the Rokinon prime does a fantastic job mitigating distortion, and consistency so. Nonetheless we felt it important to examine those differences and see how they truly compared across the same scenes, and a variety of them.
One last thing we were curious about was how the 8-16mm zoom compared after having it's 8mm setting corrected for distortion. Especially in architectural situations where distortion and misalignment can be very obvious right away, the gap between the pairs' widest field of view narrowed. The Sigma still held the edge because of how much wider it is inherently, however you'll have to work for it in post-processing to truly optimize each scene as you correct for distortion.
Over 40 comparisons later, it's clear the fisheyes' images are "fishy." But can that be corrected? The following chapter has the answer!
