Is m43 flat? The side by side comparisons feat. the Lumix GX8
I wrote not long ago an article on why I stopped using m43 cameras: flat rendering. Some has asked me to switch to Panasonic as their cameras render images differently from Olympus despite following the same standards and using the same lenses. A friend of mine has compared using his Leica M lenses on a Sony A7ii vs. a Leica SL and the results in depth are day and night. I hope the GX8 does the same to lenses I have deemed flat (perhaps because of Olympus).
A call for testing
Many of us asked for truer side by side comparisons. I have acquired the latest in amazing m43 gear:
- The amazing Lumix GX8: New sensor with hopes of better results.
- Leica 15 (9 elements): my favorite walk-around prime lens for m43
- Lumix 20 (7 elements): the only low-element count fast prime lens of the m43 lineup
- Lumix 42.5 1.7 (10 elements): the newest m43 telephoto prime
I’ve decided to use the lenses I know will pop on my fullframe Nikon D750.
- Nikkor AF 28mm 2.8D (6 elements): a low element count wide angle street lens
- Voigtlander 40mm f2 Ultron (6 elements): a low element count normal lens to match the lumix 20
- Nikkor AF 85 1.8D (6 elements): a low element count telephoto lens
- tripod + timer: same angle, pure sharpness/stability
- f8 (half aperture and half ISO values for m43 meaning f4 on the Gx8): wide depth of field (dof), to make sure elements are in focus
- focussed on the bunny: round head, best hint of pop.
- matching frames (either d750 or gx8 shots were cropped at 4:3 aspect ratio to match frames): better comparisons
- matching colors in raw on lightroom: better comparisons
While I am sure the 28 (6) and the 85 (6) will out-pop the 15 (9) and the 42.5 (10), my interest lies on the 20 (7) vs 40 (6) faceoff. This test is also done on the sensor/processor combo that shapes the new direction of m43. I hope it will provide new rendition to the previous flat Olympus files of my OMD-em10. Once again, I encourage viewers to pay attention to shapes, not details nor colors nor sharpness. The differences are subtle but they will matter on repeated uses over a wide range of scenarios. The shapes, depth, distance and geometry of objects will reveal the depth or lack thereof of each lens/camera combo. If you don't see the differences, simply disregard this article and be happy using whatever you are using. Here we go.
10 vs. 6 : the 42.5 vs. the 85
With Gx8 on the left (forgive the intro picture) has more contrast and colors but is flatter in depth and geometry, not by much.
By looking at them side by side (lumix left, nikkor right), you can already pick up that the depth and distance between the bunny’s head and the red book: it is closer in the lumix. The rabbit head is rounder in the right side but we can look in closer to see.
Here’s the close up.
Pay attention to how the left cheek of the right rabbit is closer to you than the left.
Then pay attention to the depth between the very same left cheek and the brown book behind it.
If you have followed the instructions well, you would have seen that the low-element count Nikkor definitely pops more than the lumix, but the differences are more subtle than expected.
9 vs 6: the 15 vs the 28
The results are similar to the previous comparison: m43 flatter, fullframe with classic primes more deeper. Lets see if the same tell-tale signs are visible.
So far the left cheek to red book depth and distance remain the same: Lumix being closer. I have highlighted other books behind the bunny to help evaluate the depth found between the head and the books. Lets look at the head.
In this comparison, I have tried my best to shrink the nikon side in photoshop to match the size of the lumix head. This is very hard to match since the lumix is a 2x crop into a 15mm and the nikon is a fullframe coverage of a 28mm lens. (forgive the labeling)
The cheeks are showcasing a similar behavior again, but it is much more subtle this time. It can be due to the fact that the lens has 1 less element than the previous comparison. Lets keep looking.
In this highlight, the depth and distance between the bunny ears and the books behind is closer in the lumix shot. There’s somehow an unequivalence of depth of field despite both lenses technically showcasing an f8 capture. I also notice that the ears are much more detailed and sharper in the lumix shot, perhaps due to modern optics and coatings. Nevertheless, the m43 image is still flatter, but I’m beginning to believe it’s a glass issue instead of a sensor issue.
7 vs 6: the 20 vs the 40
This is where I really want to see if the m43 can achieve great depth rendition. Results have been very close with this new camera, especially with the leica 15 (9 elements). With 1 glass element of difference, I am deeply curious to see how the classic 20 will match up to the 40, both amazing and legendary lenses.
As I suspected on my naked eye, the images come close: you’d have to really peep the telltale signs to really see the difference, it is at this point quite satisfying "enough" for users of m43 I believe.
Highlighting the bunny and the ENTIRE row of books shows that the lumix bunny is STILL flattened closer to the books behind. Lets look at the bunny head (ONE LAST TIME).
Predictably so, bunny head is rounder on the nikon side, but this could simply be due to 1 element of difference. Lets look at the ears.
There, the ears are closer to the background in the lumix shot.
The glass element theory holds true here: the more glass elements a lens has, the flatter the perspective and depth will become. Lenses of 9 or 10 elements have indeed recorded flatter images in relation to those who have 6 or 7. The m43 system CAN produce images with great depth rendition. It just requires low-element count lenses and Panasonic's understanding of depth rendition in analyzing the recorded sensor data. By low-element, I believe there are minimums: my Nikkor Series E 135mm f2.8 lens is quite the champion at 5 elements; most "decent" lenses have 6. Like you have witnessed, 9 would be turn any lens into a corner to corner flatness device.
What to do now?
If you shoot m43, you also require a camera with a sensor/processor combo that can read the lenses’ output well. The GX8 is that camera because I have never seen such pop from my OMD EM10 images and I have produced close to 10000 keepers and done countless successful commercial jobs with such a camera despite its flat rendition. This leads me to recommend using this list of “low-element count” lenses on the GX8 (it's my first Panasonic m43 so I don't know how the 12mp or 16mp LiveMOS sensor behaves).
- Lumix 14mm 2.5 (6 elements)
- Zuiko 17 2.8 (6 elements)
- Lumix 20mm 1.7 (7 elements)
- Lumix 25mm 1.7 (8 elements) <- This one maybe, there are no image samples on the internet that can tell if it’s a flat lens or a depth lens since 9 elements on m43 lead to flatness.
I’d be curious to achieve a few more tests. If you have such gear and are willing to talk about your results, please share them up! Here's a list of suggestions using the GX8+20mm combo as "depth-rendition/pop reference":
- Compare depth rendition of the combo against higher element count lenses of the m43 lineup.
- Compare the files of the combo against a recent Olympus m43 camera.
- Compare the files of the combo against m43 cameras of the 12mp, 16mp and 20mp eras using the 20mm.
Panasonic is the saviour of m43?
Seems Panasonic got it right with the GX8, but is missing on Olympus color profiles in Lightroom. There needs to be more lenses like the 14 and the 20 (element count of 7 and below), or a DFD refresh of them (for faster hunting). When the 20 isn’t hunting, it acquires focus very fast. I fear the demand for max aperture sharpness and bokeh will drive both manufacturers to release more corrected lenses with more elements or is it purely a case of cost-cutting or budget-matching? Just like any system, m43’s depth rendition is a slave to element count, unless they produce a camera with a sensor/processor that can cope with the amount of glass and provide a pleasing result.
If I am to get a m43 again for leisure, it surely would be the GX8, but that’s for another article.