Fujica GS645S - Inconsistent decentering

Hi all,

A couple of weeks ago, I got a GS645S off of eBay for quite a good price thanks to the fact that it has the apparently-common "slow rangefinder" issue. That seems like something I might be capable of fixing myself if I feel the need (if anyone else has done it, I'd be thrilled to hear about your experience).

I just took it on vacation with me, which was sort of a bonus since I hadn't expected it to arrive before I left. I shot two test rolls before going, one on Delta 100 and one on Catlabs 80. The Delta 100 roll seemed mostly fine, but the Catlabs showed the camera focusing just a hair past infinity, suggesting I should dial back focus a smidge for landscapes. That's what I did on my trip, and while I'm waiting for the two rolls of color to be processed, I just looked through the one roll of B&W (also Delta 100) I shot and got an odd surprise; many of the shots exhibit softness on the right side, while others don't.

Here's a "bad" example (all of the shots on this roll are pretty boring to boot):

A crop from the "good" right side:

And from the left:

And here's an image without the same issues:

Right side:

Left side:


Weird, right? Both were at f/11, and both were focused at the same point (the dot just before the infinity symbol) - at least as far as I can remember. There's no wobble to the lens at all, which is the first thing that'd come to mind with an issue like this that's sometimes there and sometimes not. So...maybe the pressure plate? It crossed my mind that the Catlabs film showed the past-infinity problem because of some sketchy backing paper the maker might have used, and that I perhaps shouldn't have dialed back focus the whole time. The pressure plate was set to the 120 notch, so everything was working as intended, and I can't see any unevenness to it, but who knows. I'm out of 120 film to test it at the moment.

Anyway, if anybody has a bit of insight, I'd appreciate it!

I would start with some careful measurements of the lens to the film plane or camera body, lens support rigidity, and focusing track smoothness.

• The pics you show, overall, do not have the "bite" I would expect from one of these Fujica 645. Based on my experience with (admittedly different) GS645W. Such softness might as well be caused by the contraption (That Shall Not Be Named Here) used to convey your FILM images to my screen.
  
• Shutter speed? At least one pic seems to have been taken from a boat. Experiment with tripod.
  
• You need to determine whether the rangefinder is wrong, or the distance scale on the barrel (or both). This might help:
  http://rick_oleson.tripod.com/index-123.html

but the Catlabs showed the camera focusing just a hair past infinity

Click to expand...

Not clear what you mean, barring a test on an optical bench with a precision collimator. Do you mean you focus a point at "infinity" with the RF, then look at the distance scale? Then all you can say is they disagree, not which one is wrong. To answer that question, you need to look at what is happening in the film plane, either through a properly planned series of test shots, or using the trick of Rick Oleson (see above). Plus, film plane is defined by the rails, not by the pressure plate; so i don't see how the backing paper of Catlabs film could affect focusing.

shutterfinger said:

I would start with some careful measurements of the lens to the film plane or camera body, lens support rigidity, and focusing track smoothness.

Click to expand...

I'll try to learn how to do all of that!

bernard_L said:

• The pics you show, overall, do not have the "bite" I would expect from one of these Fujica 645. Based on my experience with (admittedly different) GS645W. Such softness might as well be caused by the contraption (That Shall Not Be Named Here) used to convey your FILM images to my screen.

Click to expand...

The two 100% crops from the 'good' shot do match up pretty expectedly vs. border quality of examples I've pored through Flickr for. And the center sharpness is pretty great. More on this in a bit.

• Shutter speed? At least one pic seems to have been taken from a boat. Experiment with tripod.

Click to expand...

Yup, from a boat, so obviously not ideal. I'm waiting on some new film to do some more technical tests.

• You need to determine whether the rangefinder is wrong, or the distance scale on the barrel (or both). This might help:
  http://rick_oleson.tripod.com/index-123.html

Not clear what you mean, barring a test on an optical bench with a precision collimator. Do you mean you focus a point at "infinity" with the RF, then look at the distance scale? Then all you can say is they disagree, not which one is wrong. To answer that question, you need to look at what is happening in the film plane, either through a properly planned series of test shots, or using the trick of Rick Oleson (see above). Plus, film plane is defined by the rails, not by the pressure plate; so i don't see how the backing paper of Catlabs film could affect focusing.

Click to expand...

Unfortunately I haven't even made any attempt to check rangefinder accuracy, since every shot on the rolls I've actually developed has been a landscape. We'll see how the others come out! By 'infinity' I mean turning the focus ring as far as it'll go, and not paying any attention to the rangefinder.

Here's the reason I was curious about the Catlabs film:

1. Center crop Delta 100, focused at infinity, f/5.6:

2. Center crop Catlabs, focused at infinity, f/5.6:

3. Center crop Catlabs, focused at the 'pre-infinity dot,' f/5.6:


Interestingly, the Catlabs film seems a bit...warped, with a decidedly non-flat surface. It's obviously not extremely easy to represent that in a photo, but this does a decent job:


So maybe it wasn't the backing paper after all but a flaw in the film itself. Either way, I based my "infinity" focusing on that test roll for this trip, which I probably shouldn't have done! My hope is that my focus point was actually short of the hyperfocal distance and that's what led to the sharpness issues. Obviously, more testing is needed.

Also, that Rick Oleson link is really interesting, and I'm going to give it a try, but I'm not exactly sure what the rangefinder bit is actually testing. Accuracy at infinity?

cptrios said:

I'll try to learn how to do all of that!

Click to expand...

https://www.ebay.com/itm/122870819066?hash=item1c9babe8fa:g:JxMAAOSwJVlgitMz
https://www.ebay.com/itm/143513726896?hash=item216a15afb0:g:9CgAAOSwSKBeLhr3

The lens/shutter should be rigid when the camera is open, no movement in any direction.

cptrios said:

So maybe it wasn't the backing paper after all but a flaw in the film itself.

Click to expand...

Agreed. In a difficult situation, eliminate as many unknowns as possible (e.g. film of unknown or unproven origin).

Also, that Rick Oleson link is really interesting, and I'm going to give it a try, but I'm not exactly sure what the rangefinder bit is actually testing. Accuracy at infinity?

Click to expand...

The second part is testing the "infinity" position of the rangefinder. Personally, I have never used that. In daytime a distant (100m? 500m?) pylon is a good test. To recap, three things must agree: (1) lens focusing an infinity object in film plane; (2) rangefinder showing coincidence for an object at "infinity"; (3) distance scale on lens barrel at the infinity mark.

Also, it might help to advance the film just before taking a pic, rather than immediately after (anticipating the next shot). That is because the memory of the curvature of the film on the spool helps keeping it against he plane defined by the pressure plate. If you advance film by anticipation, the film has time to settle into a plane, and maybe even deform beyond that plane; all this is qualitative and I don't have A versus B evidence, but better stay on the safe side.

Well, I bought a few rolls of Arista 100 just to test it out. First one had some interesting results (even though I utterly botched processing it by initially putting in a 35mm-appropriate amount of developer solution). I know everyone has been extremely eager to hear about them!

- Clearly there was something wrong with that Catlabs film, as infinity focus looks fine.
- The 3 C41 rolls I'm waiting for from the lab are going to be extremely depressing, since based on the Catlabs experience I shot all of my landscapes at the 'dot' before the infinity sign. That might be a fine hyperfocal distance for medium-sized prints, but border softness would be visible at 20x30.
- That "decentering" doesn't exist at infinity - just on distant subjects when focus is as far back as that dot. I should do some brick wall tests to see if it would ever have any real effect on anything.
- I don't like the camera very much, sadly. Turning the camera vertically for horizontal shots is more annoying than I expected. It's a shame, since there isn't much else out there that'll give me 120 IQ in such a small package (and not cost as much as my car). Perhaps I'll keep plugging away at learning to focus my TLR.
- The rangefinder generally seems accurate, despite needing to be re-lubed.
- The imprinting on the Arista film's borders bleeds into the frame way too easily. Super relevant, I know.

Here's a shot at infinity, f/5.6 (original dimensions approx 6600x5000):

Centerish crop:

Left side:

Right Side:


More testing is needed but I'm at least fairly confident that there's nothing terribly wrong with the camera.

cptrios said:

Turning the camera vertically for horizontal shots is more annoying than I expected

Click to expand...

That is something I have a hard time understanding. Nobody complains about having to turn their "normal" camera sideways for portrait shots. But it's a personal thing...

To me the virtue of the Fujica 645's is excellent optics in a relatively lightweight package. Rollei and the like are also relatively lightweight, but (with the exception of super-rare and expensive versions) limit you to "normal" focal length. Their downside is not the allegedly plastic construction (metal chassis underneath) but some reliability weak spots, e.g. in the film advance/counter mechanism (plastic gears).

based on the Catlabs experience I shot all of my landscapes at the 'dot' before the infinity sign. That might be a fine hyperfocal distance for medium-sized prints, but border softness would be visible at 20x30.

Click to expand...

Shooting outside, you must have used an f-number such that your focusing error was well within the DOF. And there is no reason why it should be more apparent at the edges than at the center (except if the optics has field curvature). And, I feel that the images that you show do not do justice to the Fuji optics. Unless mistaken, you did not state which s*****r model you use. Whatever, your images could use a reasonable amount of sharpening. Not to create artificial sharpness (that is impossible) but to compensate the falloff of the s*****r's MTF at the highest spatial frequencies; Suggest starting with a radius of 1.4, increase amount cautiously and back off when results look artificial or edge effects appear.

To document my statement about the image quality that a Fuji 645 camera can deliver. Fujica GS645W. Reala 100. Tripod. Commercial photo store dev+s**n (Noritsu). Location: Humberstone nitrate plant, north of Chile.

bernard_L said:

That is something I have a hard time understanding. Nobody complains about having to turn their "normal" camera sideways for portrait shots. But it's a personal thing...

Click to expand...

I think for me it's simply that I don't often shoot in portrait orientation. A quick filter of my LR library says that only about 25% of my shots are in portrait, and a lot of those are also likely to be parts of eventual stitched panoramas. So holding the camera in portrait orientation for 75% of my shots is a bit awkward. I particularly don't like the location of the strap lugs in this context!

Shooting outside, you must have used an f-number such that your focusing error was well within the DOF.

Click to expand...

Actually, I'm not so sure about that. Even at f/11, hyperfocal distance on this camera should be a bit over 7m. The last marked distance on the scale on this camera is 5m, so it's tough to know exactly where I focused on the 'bad' shots. Almost definitely shorter than hyperfocal for f/8 and under. That's also coupled with the fact that I'm rarely satisfied with infinity sharpness at calculated 'hyperfocal' distances when pixel-peeping.

And there is no reason why it should be more apparent at the edges than at the center (except if the optics has field curvature).

Click to expand...

Yep, that was my main concern, and I'll need to do more testing to figure it out. I'm sure the lens does have some field curvature, but how much, I don't know. I think that one side being weaker than the other definitely indicates a centering/alignment issue, but once I'm even sure that one side is sharper than the other in those situations, I'll then want to make sure the issue is actually relevant to real shooting.

And, I feel that the images that you show do not do justice to the Fuji optics. Unless mistaken, you did not state which s*****r model you use.

Click to expand...

I remain reasonably confident that I'm getting the expected IQ out of the lens when I hit focus. A high-res Flickr search doesn't show me any samples that are noticeably better than my 'good' ones, and plenty that are worse. Unfortunately, many of them are either at substantially lower resolutions (12mpx vs my 32) or done on flatbeds, which never do film/lens sharpness justice. I've really learned to hate flatbed scanners in all of this time researching buying film gear!

As for my scanner (it took me a moment to realize that was what your 'bleeped' words were!), I'm using a NEX-7 with an old Macro Takumar 50/4. Definitely not an ideal setup, but more or less fine. All of the samples I've posted are from two-frame stitches, resulting in a 6600x5000 image. When I do them in a single frame (getting something like 5300x4000), the 'problems' are barely noticeable. I would ideally like to be able to squeeze more resolution out, though! If I can't consistently get a satisfactory 6600x5000 scan, I'll probably count it as a disappointment.

For what it's worth, here's a stitched-scan shot that was focused well, and a center crop:


Personally, I'm happy enough with that!

The scans do look unnaturally soft to me. Maybe due to being rezzed down? How do the negs look on a light table w/ a loupe? I owned a few examples of your camera and felt it was a PITA to have to turn the camera in order to make horizontal images, which is the default on most cameras. The Fuji glass is very clean and sharp, but a little lacking in the way of "character" compared to the old glass like Heliars, Planars, and the Summicrons in 35mm. By pushing Tri-X w/ Y and R filters I was finally able to coax what I considered to be some really good shots from your camera. But it was easier to just buy an old Zeiss 6x4.5 folder. It had that old classic look to the negs. Still had to turn the camera sideways, but not a big deal because it was so tiny.

momus said:

But it was easier to just buy an old Zeiss 6x4.5 folder. It had that old classic look to the negs. Still had to turn the camera sideways, but not a big deal because it was so tiny.

Click to expand...

Can I ask which one? Folders are very appealing to me but it seems like anything with the sort of IQ I'm looking for is out of my budget range.

OP, I just read this thread from the first post. All of your example shots are horribly soft. You seem to be shooting at relatively large apertures and to be focusing very badly. For example, in some of your building examples the buildings were much closer than far, far away but you focused at infinity. And you talked about hyperfocal. Both are classic recipes for poor focus. Focus on the main subject, shoot from tripod (your examples are so fuzzy that I'm sure motion blur can be discerned in them). And your example color shot with Catlabs film is just one giant blur. Can't tell anything from it.

cptrios said:

Can I ask which one? Folders are very appealing to me but it seems like anything with the sort of IQ I'm looking for is out of my budget range.

Click to expand...

The Super Ikonta A is small (have one). It has a Tessar lens and coupled rangefinder. Later models have Albada viewfinders, providing in principle more accurate framing. All these features, however, are more important on paper than in practice. Tessar lens: Other cameras with triplets (e.g. a Nettar with Novar lens) provide results of (IMO) equal quality to Tessar, at apertures f/8 or smaller. Rangefinder: most pictures I take are at distances 2m or more, where estimate is good enough; with the Nettar, to take a pic at less that 2m, I bring out the aux rangefinder; total cost of rangefinder-less camera plus aux rangefinder is significantly less than coupled-rangefinder camera, especially in the current state of the market. Albada v/f: after some practice, I learned how to use the crappy viewfinder (eye placement) and do not have bad surprises with the as-shot framing.

But we are outside the scope of the OP's question. Sorrry about that.

I've shot many rolls of slide film with a pre-war Super Ikonta A with Tessar lens.
Bitingly sharp even focused at infinity. I rarely shoot at an aperture smaller than 5.6-8.

bernard_L said:

But we are outside the scope of the OP's question. Sorrry about that.

Click to expand...

No worries, I'm the one who asked!

Dan Fromm said:

OP, I just read this thread from the first post. All of your example shots are horribly soft. You seem to be shooting at relatively large apertures and to be focusing very badly. For example, in some of your building examples the buildings were much closer than far, far away but you focused at infinity. And you talked about hyperfocal. Both are classic recipes for poor focus. Focus on the main subject, shoot from tripod (your examples are so fuzzy that I'm sure motion blur can be discerned in them). And your example color shot with Catlabs film is just one giant blur. Can't tell anything from it.

Click to expand...

A lot of what you say is basically in line with my posts. At some point I'm going to try to get a better visual representation of how messed up that Catlabs film is! Out of curiosity...do you really find these two center crops to be horribly soft? I'll do an LR export rather than a screenshot.


I guess you might be able to squeeze more detail out of Delta 100 (first image), but I can't imagine doing much better with Fomapan (2nd).

cptrios said:

A lot of what you say is basically in line with my posts. At some point I'm going to try to get a better visual representation of how messed up that Catlabs film is! Out of curiosity...do you really find these two center crops to be horribly soft? I'll do an LR export rather than a screenshot.

Click to expand...

Yes, terrible. I've been digitizing old 35 mm slides (ISO 100 Ektachromes, ISO 25 Kodachromes) with a D810, 55/2.8 MicroNikkor @ f/8 or f/11, bellows and slide holder. I don't always focus well on the slide to be digitized, and then the results are, um, dreadful. Your examples remind me of my dreadfuls.

Dan Fromm said:

Yes, terrible. I've been digitizing old 35 mm slides (ISO 100 Ektachromes, ISO 25 Kodachromes) with a D810, 55/2.8 MicroNikkor @ f/8 or f/11, bellows and slide holder. I don't always focus well on the slide to be digitized, and then the results are, um, dreadful. Your examples remind me of my dreadfuls.

Click to expand...

Ok, I know I shouldn't let one person's opinion on the internet affect me, but I trust posters' knowledge on here by default and you've made me feel rather gaslit! I worry, since those two crops are more or less in line with all of my results with my film cameras scanned with the NEX-7 setup. And they look fine to me!

Nevertheless, I just went back and did an 8-shot stitch that resulted in a 108MP, 9000x12000 image (I stopped down the lens to f/11 because I worry about DoF that close, and unfortunately I've noticed it perform slightly worse than at f/8. The extra magnification should more than make up for that though). I then reduced it back down to the 6600x5000 of the 'original' and made a similar crop:


I'm quite happy with that, and if this represents a low-quality 645 shot then I'm a lot more excited about the potential of the format!

Another thought. For many reasons, most images on film look soft when enlarged > 10x. How much are your scans enlarged from the original?

Well, I went and did another test Arista roll, and if it had been my first, I might not have even thought there was any kind of issue. These two: 1 2 came from there, and while I do think one end of the frame (just at the very edge and at the corners) is softer than the other, it's probably not intrusive enough that I wouldn't use the camera if I enjoyed using it more. But I don't, so if anyone wants it, it'll be in the classifieds soon!