GFX for Astrophotography - Fujifilm GFX 50S II Review
Landscape astrophotography is one of those demanding niches of landscape photography, where a very specific set of traits for gear are important. For that reason, it’s hard to find reviews that specifically go into this online. I’ll add pretty pictures taken with this camera later. This is a test to see if it can take pretty pictures of the night sky potentially. I’m convinced. Here’s why.
I use affiliate links to pay for the bills. These reviews take a lot of work, so I have to do this. If you buy any gear through these links, I get a small commission. A better way of saying thanks is getting an e-book or processing video.
Straight answer
To answer your question: is the medium format Fuji GFX 50S II a good camera for astrophotography?
Yes. Without reservation, but there’s a lack of lenses to make this eco-system the best choice. Having said that, the shadow recovery ability of the GFX 50S II might offset the loss of light in the current lens line-up.
The idea for testing the Fujifilm GFX 50S II for nightscapes came in preparation of a workshop to Iceland. So I researched the web for a rabbit hole here: https://www.cloudynights.com/topic/754363-anyone-using-a-fujifilm-gfx50100-or-100s
It turns out there’s a pretty deep warren behind that rabbit hole, so let’s go.
Defining landscape astrophotography
I’ve defined landscape astrophotography or “nightscapes” extensively before: we’re shooting landscapes shot at night, sometimes using advanced processing techniques such as stacking, focus stacking or compositing to some degree. They can feature constellations, celestial events or the aurora, but typically feature the Milky Way as the main subject over a significant portion of landscape. Different than deepsky astrophotography, nightscapes don’t really use long lenses or equatorial mounts to build up the faintest light over really long exposure times. Instead, we’re typically bound by the rotation of the Earth and want to include a sense of wonder and recognition.
What we need to test
Cameras need a different set of test parameters than lenses. Here’s a testbed for testing and checking if a camera is a good candidate for nightscapes:
Does the camera feature a “star-eater” on-sensor noise reduction algorithm?
Is there any significant banding, patterning or forming of concentric rings with this camera?
What’s the dynamic range at typical ISO’s for landscape astrophotography?
Does that dynamic range scale linearly, meaning: is the camera ISO-invariant?
Choice of lenses: does this camera support lenses that are good candidates for shooting at night?
How is the signal noise performance at high ISO’s?
How is the thermal noise performance at long exposures?
Does that lead to “amp glow”, where colored noise is more obvious in one area (usually the bottom) than another?
Is there a difference between using electronic, mechanical or first curtain electronic shutter if applicable?
Questions 1-5 can be answered by checking the right websites online.
1 Does the GFX 50S II feature a “star-eater” on-sensor noise reduction algorithm?
Mark Shelley is the guy to ask. It appears the 50S II has not been tested for this, but the previous generation has. I have not found out online if the GFX 50S II star eater is a thing. So here’s what I see:
Fujifilm GFX 50SII - Laowa 19mm f/2.8 Zero-D GFX - f/2.8 25s ISO800 - Full moon - cloudy - No processing, noise reduction turned off completely.
“Nice picture, Daniel. Why the fuck would you test a camera like this on a ‘night’ like this?”
Well, it doesn’t really matter. There’s no need to take pretty pictures to test anything. All those lab tests elsewhere aren’t real world photos either. Here’s a crop of that image:
Crop of the image above.
I can clearly see darker areas or pixels here and there in the background that to me indicates there’s some amount of noise suppression going on, that leads to fainter stars being filtered out. That’s what a “star eater” is. Here’s another, deeper crop to show the noise profile.
Another crop of the same image from above.
2 Is there any significant banding, patterning or forming of concentric rings with the GFX 50S II?
Yes, there is. Although it’s not that bad. This image is a screenshot from photonstophotos to illustrate the patterning formed from thermal noise. I can’t share a screenshot because it’s copyrighted.
There’s no mention of concentric rings forming when stacking online, which is a good thing. Although I suspect no-one has tested this with the GFX 50S II. Concentric rings show up and are an issue when you’re stacking several dozens of images for the purpose of showing fainter stars and nebulae. And we don’t use those techniques for nightscapes.
3 What’s the dynamic range at typical ISO’s for landscape astrophotography?
ISO 400: 10.16 log2(EV)
ISO 800: 9.23 log2(EV)
ISO 1600: 8.28 log2(EV)
ISO 3200: 7.29 log2(EV)
ISO 6400: 6.28 log2(EV)
Source: https://www.photonstophotos.net/Charts/PDR.htm
4 Does that dynamic range scale linearly, meaning: is the Fujifilm GFX 50S II ISO-invariant?
Yes, the GFX 50S II is ISO invariant. More so than full frame camera’s that also use Sony-made sensors. What I mean by that is that the whole ISO range follows a nice straight line. I can’t share a screenshot because it’s copyrighted.
I see a slight bump at ISO 200. Not enough to say that this camera uses a dual gain architecture like most Nikon cameras do. And it’s certainly not useful.
Recommendation: On the GFX 50S II shoot at an ISO lower than you need to expose the photo, then raise the exposure as needed in post. This will introduce the exact same noise as if you would exposing correctly in-camera, but it will allow much greater dynamic range.
5 Choice of lenses: does the GFX 50S II support lenses that are good candidates for shooting at night?
Again, this is research that others have already done: https://alikgriffin.com/fujifilm-g-lenses-the-ultimate-list
What I would add is that these wide-angle lenses are good candidates for landscape astrophotography:
AstrHori 12mm f/2.8 Fisheye
TTArtisan G 11mm f/2.8 Fisheye
Laowa G 19mm f/2.8 Zero D
Note that these are the focal lengths on GFX medium format. Divide these focal lengths by 1.25 to get an indication of what the view would look like on full-frame 35mm.
Aside from this selection, there are a number of adapters to fit basically every conceivable lens on this system, but be careful with full-frame lenses. They are probably worse in the corners than they are on full-frame, as you will add more area. That will result in more vignetting and worse artifacts in the corners.
I’ve tested the Laowa G 19mm f/2.8 Zero D:
Laowa 19mm f/2.8 GFX Coma Test
“Coma” is a catch-all word to describe multiple lens aberrations in practice. Actual coma is one of those aberrations. How well does the Laowa 19mm f/2.8 perform wide-open? Well not so stellar:
Bright lights (stars and distant lights on the horizon) appear stretched and distorted with lots of structure.
Here’s the other extreme corner, where I brought up the exposure 2.85 stops up from ISO 100 at 15 seconds, showing a loss in sharpness as a result.
Stars to the right side of the frame.
Star in the top left corner (not the extreme corner).
6 How is the GFX 50S II’s signal noise performance at high ISO’s?
This is the part where I test bringing up the exposure in post to see if the noise is acceptable to me.
ISO 400 brought up 2 stops, to the equivalent of ISO 1600.
I used these noise reduction settings in Adobe Camera Raw. No sharpening (don’t do that for stars).
A more extreme sample?
Turning night to day. Fujifilm GFX 50SII - Laowa 19mm f/2.8 Zero-D GFX - f/8 30s ISO400 - Full moon.
5 stops of exposure added in Adobe Camera Raw.
For the image above, I’ve added this sharpening and noise reduction.
I do see the grass turning purple here and there in the immediate foreground. So I wouldn’t use this if I don’t have to. My foregrounds don’t look like this either. But here’s a crop to let you see the signal noise:
That white grain is the signal noise. Aside from that, the color starts falling apart. But again, this isn’t a technique I would use for my foregrounds.
The conclusion here is that the GFX 50S II’s noise performance is on par with any modern full-frame high mega-pixel camera like the Canon R5, Nikon Z7II, Z8 or the Sony A7RIV. It’s also very comparable to the Hasselblad X1D II 50C, which is perhaps less surprising as these cameras are comparable in a number of ways. If you were to compare apples to apples, these are the cameras to compare.
Now this is 5 stops added in post on top of ISO 400 is the equivalent of ISO 12.800 - a setting I have never used on any camera, for any landscape, ever. I just don’t see the need in landscape photography. To get things brighter and with less noise, we’ll test how well the GFX 50S II performs with extended shutter speeds next.
7 How is the thermal noise performance of the GFX 50S II at long exposures?
Thermal noise is a very different beast compared to signal noise. Hot pixels will show up and the noise is likely more colorful, leading to unsightly shadows. Let’s go.
The dark corner of my studio. Fujifilm GFX 50SII - Laowa 19mm f/2.8 Zero-D GFX - f/8 4 minutes ISO 100.
All noise reduction and sharpening left off.
I don’t see any stuff that’s not supposed to be there at 4 minute exposures. I would feel very comfortable using this or even 8 minute exposures in the field. The added bonus is that the GFX 50S II allows automated exposure times for longer than 30 seconds by a simple additional turn of the shutter speed dial. How come not more cameras have this feature?
8 Does that lead to “amp glow”, where colored noise is more obvious in one area (usually the bottom) than another?
While I have not seen amp glow in the long exposure, I have seen it in the field test for bringing up the ISO 400 to 12.800. Here’s an even more exaggerated processing of that bottom area.
The bottom part looks slightly more magenta than the rest already does.
Here are my settings in Adobe Camera Raw. Again, I would never do this in an actual photo, but to illustrate what this camera does under extreme conditions…
I mean, this is the converted raw file; untouched at ISO 400 and 30 seconds. It’s almost completely black. here’s the histogram that goes with it:
Barely any information in the shadows. It think shadow recovery on the Fujifilm 50S II is… decent. Lol. Especially considering this is a 50MP camera.
9 Is there a difference between using electronic, mechanical or first curtain electronic shutter of the GFX 50S II if applicable?
The GFX 50S II has a mechanical shutter, so we can test all three modes. The reason for testing this is to make sure no additional on-sensor processing takes place that could make your photos look like crap.
I don’t see any difference at all between any of the shutter modes.
Recommendation: shoot using the electronic shutter whenever possible. This saves shutter actuations, battery life and potentially lowers camera shake leading to slightly unsharp images (be sure to use a 2 second exposure delay or a remote and always shoot from a sturdy tripod). There is no downside with the GFX 50S II in the resulting images.
Conclusion: Would I recommend using the Fujifilm GFX 50S II Medium Format camera for landscape astrophotography?
Without reservation: absolutely. Here’s why:
The GFX 50S II is as ISO invariant as cameras come. That means (even) more dynamic range when shooting underexposed with a lower ISO setting.
The large sensor allows for large photosites and a high pixel pitch, leading to less interference between pixels. But you could deduce that from its specs. What I found in real world scenarios is:
There’s a low level of signal noise at high ISO settings, considering this is a 50MP camera.
There’s no visible thermal noise at shutter speeds of 4 minutes, so your foregrounds will be clean as a whistle.
The dynamic range is extremely high and it shoots 16-bit raw files full of color information.
It doesn’t matter if you prefer electronic or mechanical shutters or a combination of both - they perform equally well on the GFX 50S II.
There’s a huge amount of post-processing leeway in these raw-files.
As for lenses, I would probably not go for the Laowa 19mm f/2.8 Zero-D GFX. It seems like it’s a full-frame lens with a G-mount. I knew it was too good to be true. I don’t mind the crazy vignetting, but the star quality is just sub par. It’s not among the worst, but I’ve definitely seen better lenses for nightscapes.
“The medium format Fujifilm GFX 50S II is highly recommended for nightscapes (landscape astrophotography).”