-810443.jpg) |
| The comparison of these two images is indicative of the types of image files I made to compare the full frame 45.44mp Nikon file to an upscaled (to same pixel dimension on the long side) Micro 4/3 20.11mp file from the OM-1 Mark II. The is the full image from the Nikon Z8. It is 45.44mp. (click to enlarge) |
-3.jpg) |
| This is the upscaled 20.11mp image from from the OM-1 Mark II. It has been upscaled to the same linear dimension as the long side of the Nikon file. It is 51.1mp. The difference in file size (45.44mp vs. 51.1mp) is due to the difference in the format—2X3 v. 4X3. The Micro4/3 image has more pixels on the short side of the file. |
On occasion, because we aren’t in the best position, or don't have the right gear with us, to ‘get it right in the camera’ we may need to crop our images significantly. Also, possibly, we want to make a very large print. Sometimes, maybe both. In the past, those two things may not turn out to your satisfaction if one’s camera sensor didn’t have a large number of pixels. I own and use a 20mp OM-1 Mark II as well as a 46mp Nikon Z8. Because sensor size and pixel count are so different in the two cameras, it got me to wondering a couple of things.
The first question I had for myself was, “Can upscaling an OM-1 Mark II 20mp file favorably compare in fine detail and sharpness to a native Nikon Z8 46mp image file?”
The second question I had for myself was, “How does using the 50mp High Resolution Handheld feature in an OM-1 Mark II compare to a native 46mp Nikon Z8 file as well as to an upscaled 20mp > 46mp from the OM-1?”
The only way for me to find out and satisfy my curiosity was to conduct some simple, somewhat non-scientific, tests. Practical comparisons.
I made several identical image files (considering the differences in sensor size and shape) with my OM-1 Mark II and Nikon Z8. The images included vegetation such as grass, leaves, flower blossoms and small bare tree limbs as well as some letters, numbers and structures. I wanted a variety of subjects for comparison. These are kinds of things I regularly photograph so using the same subject matter for comparison sake makes sense.
I had the OM 12-40mm f/2.8 II Pro lens attached to the OM-1 Mark II and the 24-120mm f/4 lens attached to the Z8. Both are excellent lenses. They are my "go to" lenses on both cameras. My intention was to make the tests with lenses I use regularly, not some exotic primes. In other words, again, a practical test with practical choices.
I made raw images at f/4 and f/5.6 with the OM-1 and f/8 and f/11 with the Nikon. From my prior testing I consider those apertures the best for my copy of each lens. I shot images at the full frame equivalent of 35mm, 50mm and 80mm. All images were handheld and manually focused using a 10X magnification. Using autofocus, more or less, tests your autofocus accuracy and I wanted to remove that variable from the test. Both cameras were set for ISO 200, which is the native ISO of the OM. I handheld the exposures because that is the way I normally photograph. I rarely use a tripod due to the excellent image stabilization in both of these cameras. Again, a practical test using the cameras how I normally use them.
Importing the files into Lightroom, I made some initial edits in the files from both cameras in brightness, color balance, highlights and shadows. For the Nikon files, I also added some texture, clarity and adjusted the sharpening as I normally would. At this point I did not make any adjustments in texture or clarity in the OM files. Additionally, I zeroed out the default sharpening. Those adjustments would be applied to the OM files after they returned from Topaz Gigapixel and Topaz Photo AI. I always send files to the Topaz plug-ins without any sharpening, clarity or texture applied. I could have sent the OM files to Topaz as raw files but, in this test, didn’t. I just didn’t think of it. I may go back and send a file as a raw file to see if I can detect any differences. If I do, I’ll update this post.
I then sent the OM files to Topaz Gigapixel AI v.8 as well as a couple to Topaz Photo AI as I also wanted to know if one performed better than the other. In Gigapixel AI I set the upscaling to the same pixel dimension as the long side of the Nikon files—8256 pixels. However, remember the two cameras’ file are different shaped rectangles—the Nikon is a 2X3 ratio and the OM 4X3 ratio so the final image files would differ in total megapixels. I then returned the OM files to Lightroom, added the same amount of texture and clarity to the newly upscaled OM files as I did with the Nikon files but did not add any additional sharpening as that was handled in Gigapixel.
So, what did I find? I'll describe it to you the best I can. I've included a number of 100% crops from both cameras. I don't know how well you will be able to see any differences after Blogger compresses them. Here they are with a description below.
-810443.jpg) |
| Approximately a 100% crop from the Nikon Z8 file. (click to enlarge) |
-3.jpg) |
| The same from the OM-1 Mark II upscaled file. (click to enlarge) |
-810443.jpg) |
| Nikon Z8 file. I don't know if you can see them in Blogger, but notice the tiny antenna shafts above the building's dormers. Also notice the birds and antenna shafts on the roof on the right. (click to enlarge) |
-3.jpg) |
| OM-1 Mark II upscaled file. Notice the same are visible here also. (click to enlarge) |
-810443.jpg) |
| Nikon Z8 file. Notice the tiny bare branches above the tree line on the left. (click to enlarge) |
-3.jpg) |
| OM-1 Mark II file. Notice the same fine, tiny branches are visible as well in the upscaled file. Also, notice the letters on the small size attached to the railings are not quite reproduced as well as the native Nikon file. (click to enlarge) |
-810443.jpg) |
| Nikon Z8 file. (click to enlarge) |
-3.jpg) |
| OM-1 Mark II upscaled file. (click to enlarge) |
The answers from my testing are…..drumroll, please….The differences I could see in detail between the upscaled 20mp > 46mp OM-1 image file and the native Nikon Z8 native 46mp image file are so small and insignificant that I can safely say they are essentially identical. I had to examine the images so closely and only found a couple of differences in tiny detail. The differences are sparse and so minor that I say they are inconsequential.
I could see the same small bare branches in distant trees. I could see the same small antennas on top of the buildings. I could see the same birds on the roof. I could see the same amount of detail in the grasses. I could see the same amount of detail in the pine branch, pine needles and small pine cones in the upper left of the frame. I could read the names on the construction equipment and the sign on the front of the red building. In those regards, I didn't see any differences at all.
I saw no difference in sharpness.
The biggest difference is in the two cameras rendition of color. Nikon and OM’s color rendition is very different—or more properly said, Lightroom’s interpretation of color is very different. I did my best to match the color rendition of both brand's files to match pretty closely.
Another difference I found was that there was some moiré in some of the tiny spindles of the railings along the front of the buildings. A bit of moiré was also found in some of the roof tiles on the red building above the words "Smithfield" across the front. That small amount of moiré can be negated in Lightroom so that didn't bother me. The other difference I saw was in the Keller/Williams sign along the railing in the left center of the image. The upscaled version did not reproduce the letters as clearly as the native Nikon file. A very minor difference. I did try using the "Preserve Text" in Photo AI but saw little difference.
When testing both Photo AI and Gigapixel, I didn't find a significant difference but I think the latest version (8.0) of Gigapixel AI did just a bit better job in retaining extremely fine detail.
All of the differences I just pointed out are so small and, in my view, insignificant. I had to look at the files with the proverbial 'fine tooth comb' to find any differences.
Now this is interesting. Technically, we know that a 20mp file doesn't capture as much fine detail as a native 46mp file. When upscaling the 20mp file to 46mp, more fine detail is not added nor created. The fine detail is just being enlarged through interpolation. That said, I could not find any additional fine detail in the Nikon file than I could see in the upscaled 20mp file. That was a surprise considering the 46mp files should capture more fine detail. That tells me that having those extra pixels is a 'fool's errand.' If I can't see any differences in fine detail between a 20mp file upscaled to a 46mp and a native 46mp file, how large a print would I have to make the 46mp file in order to actually see a difference? Again, this is not a scientific examination conducted with a microscope but a practical test to satisfy my curiosity of what 'practical' differences there are. My answer to me: Little to none.
As far as the 50mp Handheld High Resolution feature in the OM-1, well, it didn’t do nearly as well as the upscaled file. Not even close. If it is ever necessary to enlarge a 20mp OM file or the need to crop in significantly arises, the upscaled file is much better in detail and sharpness. Topaz software does an amazing job.
In the future, I will feel very comfortable using my 20mp OM-1 and superb Olympus and OM lenses and no longer worry that I'm missing out on something or at a disadvantage in sharpness and fine detail that my full frame Nikon files exhibit. If I ever need to crop in significantly or make really big enlargements, I know I can do it with little or no penalty. I call that piece of mind.
For those of you who still carry the mantra that Micro4/3 can’t possibly be any good, you need to go back to photo school and see how good the technology today really is.
My final word is: Don't take my word for it. Conduct your own experiments to satisfy yourself. I try to be very careful as to how I set up tests so they are practical and not a laboratory exercise, reduce the number of variables, are repeatable, and provide answers to my own questions. That way, I can know of what my cameras and lenses are capable.
Join me over at my website, https://www.dennismook.com.
Thanks for looking. Enjoy!
Dennis A. Mook
All content on this blog is © 2013-2024 Dennis A. Mook. All Rights Reserved. Feel free to point to this blog from your website with full attribution. Permission may be granted for commercial use. Please contact Mr. Mook to discuss permission to reproduce the blog posts and/or images.
Thank you for doing this! I have been wondering about this, having an OM-1. And also wondering how upscaling with Gigapixel 8 compared to Photo AI 3. (I have Photo Ai 3 and an older Gigapixel, debating whether upgrading Gigapixel would be worth it, given Photo AI 3]. Looks like taking regular 20mp shot and using Photo AI 3 upscaling will meet my needs. Much appreciated! Best wishes for the holidays!
ReplyDeleteRick
Thank you Rick. The only advantage I saw in Photo AI that I didn't see (it may be there but I didn't see it) is the 'Preserve Text' feature. That said, I didn't see much of a difference using it versus not using it. The biggest difference I found in preserving text was the choice of which upscaling algorithm you choose. Some upscaling algorithms upscale better than others and some preserve text better than others. Whenever I use either Photo AI or now Gigapixel AI v.8, I very carefully pick the best algorithm and also the best slider applications. I almost never just take what the program recommends. ~Dennis
DeleteLots of variables to take into account. As long as iso is low, cropped sensors hold their own but jump to 3200 and differences start to show up even after using software to denoise. It is not necessarily a game killer but there is a noticeable difference at least to my eye.
ReplyDeleteLarry, I can’t argue that the differences become more apparent as ISOs rise. However, I seldom shoot at high ISOs so any perceivable differences are not something I normally worry about. When I do have to shoot at higher ISOs (usually birds or wildlife), I’m very happy with the results I get from running my files through DXO and/or Topaz plug-ins. Just a comment about how I’ve changed the way I look at photos (mine and others’) now. I no longer seek perfection in photos. I used to not like my photos if everything about them was not perfect—the best they could be. I looked for flaws instead of looking at my photos for content and pleasing aspects. I now look at photos in their entirety instead of at 100% and examining them with a fine tooth comb—the way they are supposed to be seen. I find digital image perfection sterile, lacking aesthetically pleasing characteristics and soulless, if that can be said. If there is some noise, a bit less than absolute sharpness or bokeh that is not perfect, so be it. It’s about the content not the technical perfection of an image. I’m writing a couple of blog posts about this but I’ve not fully fleshed out my thoughts. When I do, I’ll post them. ~Dennis
DeleteExactly! I used to worry if the image would not be perfect at 20x30, but came to realize I don't print 20x30. :)
Delete👍🏻 ~Dennis
DeleteEven the Enhanced raw in Adobe Camera Raw allows me to make billboards and wall wraps with lots of fine detail preserved out of 20 and 25 mp m4/3 files. I laugh when people say the format is too small, then I show them the results in a 10 ft tall wall wrap and they scratch their heads.
ReplyDeleteI’m with you Mike. I think most criticism about the format comes from those who have never used it extensively. ~Dennis
DeleteExtensively, that is the key word. Most reviewers use a camera for a week or two and make a declaration. Honestly, after 12 years of Lumix cameras, it took me months of continuous using, tweaking, reading, tweaking, watching, tweaking, learning, and tweaking some more before I came to grips with how to best deploy the autofocus on my OM-5 cameras. One has to take the time to learn something new. Refining one's workflow to best take advantage of the small format also takes time. Everyone wants it effortlessly perfect out of the box.
DeleteYou nailed it, Mike. I continually find little tweaks I can make to eke out better performance in my cameras. I once wrote to a very prominent and highly respected YouTuber after watching one of his videos reviewing a new camera. He was critical of several aspects of it not having certain features or not doing what he thought it should do. Since I owned the camera and had read the entire manual, he was wrong on every count. If he had read the manual before pontificating or giving his ‘expert’ opinion, he would have known the camera could do what he thought it couldn’t. I’m sure that many photographers took the camera out of consideration for purchase because of his lack of thoroughness. That’s a shame. Never heard a reply to my comment on his site, of course. ~Dennis
DeleteVery interesting post. My standard outfit has been an OM-1 and Z7 for 2+ years now. I carry them in the same bag when out with camera. Usually the OM-1 has the 40-150 f2.8 attached, the Z7 with 14-30 or 15 shift lens.
ReplyDeleteMy standard display sizes are 4K electronic and on 13x19" prints. The image outputs between the two formats at these sizes are indistinguishable. I have wider lenses including a 15mm shift lens for the Z7 so it is often my choice for architecture. The OM-1 is a much faster camera for active subjects and I have a better choice of telephotos for it. I photograph active wildlife very early in the morning so ISO's are typically in the 1600-25,600 range. I do not see any advantage in the noise level from the full frame Z7 for this use. I find noise levels are determined by pixel pitch rather than sensor size.
My results, all though this period, in a daily use mode, have been consistent with your findings. I was a proponent of the OM high resolution but do not find any real advantage now, and there are disadvantages. I resize the standard RAW file in Photoshop after using Topaz Denoise.
Thank you for your thoughtful comment. Very informative. I get many comments and private emails from photographers who use both Micro4/3 and full frame, especially Nikon. More than I would have thought. Again, thank you. ~Dennis
DeleteFirst of all, thank you very much for your great test! I really liked it. However, I don't agree with your results. I think you can still see very clear differences. The Olympus images look significantly oversharpened in comparison and the fine details are missing. This is particularly noticeable in smaller labels, which you can still see for the most part with the Nikon images and not at all with the Olympus. The Olympus images are still good, but I wouldn't like them if I had to choose between the two.
ReplyDeleteI like and use Olympus very much, but only for larger Tele and full Format for the shorter ones.
Sheers,
Patrick
First of all, thank you very much for your great test! I really liked it. However, I don't agree with your results. I think you can still see very clear differences. The Olympus images look significantly oversharpened in comparison and the fine details are missing. This is particularly noticeable in smaller labels, which you can still see for the most part with the Nikon images and not at all with the Olympus. The Olympus images are still good, but I wouldn't like them if I had to choose between the two.
ReplyDeleteI like and use Olympus also, but only for Tele shots and the Rest with full frame
Sheers,
Patrick
I agree with most of your arguments. I remember using the Canon 7D and 1Dx, where I encountered similar challenges. However, I still managed to capture some great pictures with those cameras. Even earlier, I printed photos as large as 90x60 cm with only 6-8 MP and significantly less dynamic range.
ReplyDeleteWhile I primarily use Sony cameras (A7R5 and A9M1) for my landscape photography, I must admit that the Olympus system offers some remarkable tools. I began using Olympus two years ago, alongside my Sony gear, specifically for macro photography. Features like 10 fps with flash, 20 fps without flash with focus bracketing are fantastic—almost groundbreaking. In contrast, Sony remains limited to about 3-4 fps, even with their newer bodies, which feels far too slow for my needs.
Other features, such as “Live Composite” mode for photographing lightning or fireflies, are incredibly simple to use—almost to the point of feeling like cheating. The handheld high-resolution mode allows you to take photos at very high ISOs without significant noise issues. The system’s speed—60 fps and even 120 fps for several years now—is another standout feature, all while producing RAW images.
For these and other features, it’s perplexing why other brands lag so far behind. And yes, the image quality at 100% is excellent. It’s only at 200% magnification that you might notice upscaling artifacts.
As photographers, we often scrutinize prints—even large prints—from an unnaturally close distance, which exacerbates perceived flaws. If we break this habit, the difference becomes almost negligible. I must admit, I’m still working on improving in this regard.
Interestingly, none of my family or friends (non-photographers) ever mention issues like noise, limited dynamic range, or lower resolution. They don’t even notice these aspects.
So, yes, I agree—this is an excellent review.
Thank you for your extensive comment and thoughts. Although I will look at an image file at 100% in certain editing steps, never at 200% as I find that counterproductive to reality, I’ve adopted the practice of looking at my images, and others’s images, in their entirety, which is the way they were envisioned and crafted to be viewed. I would think almost all photographers compose and make their images to be viewed in their entirety, not consumed in tiny portions. That reduces the differences among cameras, etc., for the most part (not all circumstances) to a moot point. ~Dennis
Delete