Luminous Landscape Forum
Equipment & Techniques => Medium Format / Film / Digital Backs – and Large Sensor Photography => Topic started by: ErikKaffehr on June 13, 2019, 02:26:19 am
-
Hi,
The Fuji GFX 100 holds a lot of promise, offering a modern Sony CMOS sensor at a reasonable 10k$US price point. Therefore I eagerly awaited DPRs studio test image. It seems that there is some serious shortcoming of that test image.
(http://echophoto.dnsalias.net/ekr/Articles/TMP/GFX100_MTF.png)
Measuring and plotting MTF indicates that sharpness is worse than on the GFX 50S.
So, what is the problem?
- It could be a bad sample of the lens. I would consider this to be a possible explanation.
- Correct focus has not been achieved.
- Camera vibration?
- IBS causing unsharpness?
From the EXIF, it seems that electronic shutter was used.
I have mailed Rishi Sanyal at DPReview a few days ago, no response yet.
Best regards
Erik
-
There is thread on Preview where they discussed this.
It seems they performed several tests but had a hard time getting better results.
I'll do my own test in 10 days... ;)
Cheers,
Bernard
-
Hi Bernard,
Could yo post a a link to that thread? I cannot find it...
I am really looking forward to your findings!
Best regards
Erik
There is thread on Preview where they discussed this.
It seems they performed several tests but had a hard time getting better results.
I'll do my own test in 10 days... ;)
Cheers,
Bernard
-
Hi Bernard,
Could yo post a a link to that thread? I cannot find it...
I am really looking forward to your findings!
Best regards
Erik
I think this was the thread: https://www.dpreview.com/forums/thread/4399061
And, yes, there is something wrong with the captures, as there is a clear sheen visible over bright edges which additionally look soft.
-
Hi,
I have seen the start of that thread. But I think they miss the issue, there is something fundamentally wrong with that test image.
Just to say, the original test images in the GFX 50S was shot on the 63/2.8 at f/8. Later they tested the GFX 50R with the 120/4 macro at f/8 and that lens yielded slightly worse results.
Shooting the GF 120/4 macro at f/5.6, which is less diffraction affected with a sensor having twice the pixels should give better results.
So, there must be a flaw, either in the lens or in the protocol used.
Best regards
Erik
I think this was the thread: https://www.dpreview.com/forums/thread/4399061
And, yes, there is something wrong with the captures, as there is a clear sheen visible over bright edges which additionally look soft.
-
Hi,
I have seen the start of that thread. But I think they miss the issue, there is something fundamentally wrong with that test image.
Just to say, the original test images in the GFX 50S was shot on the 63/2.8 at f/8. Later they tested the GFX 50R with the 120/4 macro at f/8 and that lens yielded slightly worse results.
Shooting the GF 120/4 macro at f/5.6, which is less diffraction affected with a sensor having twice the pixels should give better results.
So, there must be a flaw, either in the lens or in the protocol used.
Best regards
Erik
I saw that as well. I chalked it up to the fact DP Review often goofs on focusing these lenses or perhaps chooses a bad copy of a lens (or both!). This does happen and it is noticeable. Not sure if that was the case here, but the degree of softness in the GFX 100 image does not appear to me to provide a realistic expectation of what the Fuji GFX 100 system will actually deliver.
Steve Hendrix/CI
-
Hi Steve,
I am not in the business of bashing DPR. or anyone else, but I have done some MTF calculations and the results make no sense to me.
My take is that DPR may have some homework to do...
Just to say, I have run some tests on my Hasselblad system with some feedback from Jack Hogan. That was a sobering experience. I learned quite a few bits from that.
The way I see it, the images they have are below reasonable expectations, so they need to evaluate the images and identify possible issues.
Just to say, had it not been for Jim Kasson's testing, I would say that it is just another mediocre lens. But, Jim has found it is a very good lens, so I would say DPR needs to have another look.
On the other hand, DPR used to be a supplier of reasonably good test images. But, they are a bit short of being fully transparent on protocols.
Best regards
Erik
I saw that as well. I chalked it up to the fact DP Review often goofs on focusing these lenses or perhaps chooses a bad copy of a lens (or both!). This does happen and it is noticeable. Not sure if that was the case here, but the degree of softness in the GFX 100 image does not appear to me to provide a realistic expectation of what the Fuji GFX 100 system will actually deliver.
Steve Hendrix/CI
-
I have seen city scape samples shot with the 120mm f4 on the GFX100 at f8 that were some of the sharpest and CA free images I have ever seen.
So some at least some copies of that lens are outstanding.
Focus shift could be a possible cause of the softness? Or some issue with shutter induced shocks even with EFC?
Cheers,
Bernard
-
I have seen city scape samples shot with the 120mm f4 on the GFX100 at f8 that were some of the sharpest and CA free images I have ever seen.
So some at least some copies of that lens are outstanding.
Focus shift could be a possible cause of the softness? Or some issue with shutter induced shocks even with EFC?
Cheers,
Bernard
The softness persists thru out the iso range, so shutter related is not likely.
I find the differences in corner performance highly suspect, especially the right side.
What's further interesting is the nearly complete lack of moire.
So, it's either a lens issue, or maybe a badly placed ir filter layer on the sensor, or goofy ibis behavior (badly parked in off position).
-
IMHO, the question that remains is; Where did they focus on, and could it be that "curvature of field" from the lens played a role?
Cheers,
Bart
-
.. could it be that "curvature of field" from the lens played a role?
It's not a flat field lens ?
-
It's not a flat field lens ?
Yes, it was my understanding also.
Cheers,
Bernard
-
Hi Bernard,
I got some feedback from Rishi Sanyal, science editor at DPR. He said that they missed focus. So they are going to reshoot, but they decided to wait with the reshoot until they have the final firmware.
Best regards
Erik
I have seen city scape samples shot with the 120mm f4 on the GFX100 at f8 that were some of the sharpest and CA free images I have ever seen.
So some at least some copies of that lens are outstanding.
Focus shift could be a possible cause of the softness? Or some issue with shutter induced shocks even with EFC?
Cheers,
Bernard
-
Imaging-resource also have their samples up.
-
I got some feedback from Rishi Sanyal, science editor at DPR. He said that they missed focus. So they are going to reshoot, but they decided to wait with the reshoot until they have the final firmware.
I can understand that. Then it will also be interesting to do a comparison between an ACR conversion and a Capture One conversion, all based on the same properly focused (on the slanted edge) Raw-file.
Unfortunately, the "Imaging-Resource" test scene doesn't include slanted edges, and it exhibits more depth of Field, so some parts of the image are somewhat out of critical focus.
Cheers,
Bart
-
I can understand that. Then it will also be interesting to do a comparison between an ACR conversion and a Capture One conversion, all based on the same properly focused (on the slanted edge) Raw-file.
Unfortunately, the "Imaging-Resource" test scene doesn't include slanted edges, and it exhibits more depth of Field, so some parts of the image are somewhat out of critical focus.
Cheers,
Bart
They do include table top focus cards which are quite telling.
-
This may be a sign, not of a problem (although there seems to be a problem here as well), but of the increasing sensitivity of higher and higher resolution cameras to external conditions.
I remember that this happened when the D800(e) came out - it ate our lenses for lunch, it was incredibly tricky to focus and it required new levels of vibration discipline. Most Nikon shooters had 12-16 MP bodies at the time (there was the D3x, but it was somewhat exotic). Apart from the D3x and the month-old 16 MP D4, FX was 12 MP only, and the 16 MP (DX) D7000 used only the center portion of the lens.
Now that cameras at that resolution and beyond are (relatively) common, several things have made them easier to deal with. Lenses have gotten quite a bit sharper, AF has gotten more precise, and VR (Nikonese - substitute your favorite term) has gotten really good. A Z7 or an A7rIII isn't nearly as tricky to handle as the first D800 class cameras were... resolution previously the province of 4x5" film is now handholdable in many conditions...
The GFX 100 is a 100 MP camera that encourages you to go out and shoot with it. It practically screams "take me off my tripod and see what I can do handheld" - but it has close to the resolution of 8x10" film... 100 MP backs from Phase One and Hasselblad are mostly studio beasts that are used much like 8x10" cameras always were - they're big, heavy, monstrously expensive and have mirror slap that vibrates the whole camera in your hand. We're still just beginning to learn what 8x10" level resolution in a more flexible package might look like. Both technique and technology will evolve, but it'll take time.
-
100 MP backs from Phase One and Hasselblad are mostly studio beasts that are used much like 8x10" cameras always were - they're big, heavy, monstrously expensive and have mirror slap that vibrates the whole camera in your hand. We're still just beginning to learn what 8x10" level resolution in a more flexible package might look like. Both technique and technology will evolve, but it'll take time.
How you perceive a particular camera is, of course, subject and is your business.
But objectively, most of our Phase One customers use their cameras outside a studio context often or exclusively. They are not "studio cameras" though obviously the can do great work in a studio.
-
I can only speak for the 120 with the 50MP camera, however it is one of the best lenses I have seen. Matching all my phase lenses and probably being better and matching some of the best Leica and Zeiss glas I have used.
-
This may be a sign, not of a problem (although there seems to be a problem here as well), but of the increasing sensitivity of higher and higher resolution cameras to external conditions.
I remember that this happened when the D800(e) came out - it ate our lenses for lunch, it was incredibly tricky to focus and it required new levels of vibration discipline. Most Nikon shooters had 12-16 MP bodies at the time (there was the D3x, but it was somewhat exotic). Apart from the D3x and the month-old 16 MP D4, FX was 12 MP only, and the 16 MP (DX) D7000 used only the center portion of the lens.
Now that cameras at that resolution and beyond are (relatively) common, several things have made them easier to deal with. Lenses have gotten quite a bit sharper, AF has gotten more precise, and VR (Nikonese - substitute your favorite term) has gotten really good. A Z7 or an A7rIII isn't nearly as tricky to handle as the first D800 class cameras were... resolution previously the province of 4x5" film is now handholdable in many conditions...
The GFX 100 is a 100 MP camera that encourages you to go out and shoot with it. It practically screams "take me off my tripod and see what I can do handheld" - but it has close to the resolution of 8x10" film... 100 MP backs from Phase One and Hasselblad are mostly studio beasts that are used much like 8x10" cameras always were - they're big, heavy, monstrously expensive and have mirror slap that vibrates the whole camera in your hand. We're still just beginning to learn what 8x10" level resolution in a more flexible package might look like. Both technique and technology will evolve, but it'll take time.
I'm glad they confirmed what I suspected, that they mis-focused.
However, this does not alarm me in any way, nor does it trigger any concerns over the Fuji lenses or getting sharp images out of the GFX 100 sensor. I have seen occasional mis-focuses from DP Review tests for years, even on little micro 4/3 sensor 16mp images. So this has nothing to do with increasingly difficult tolerances, lots of pixels, etc.
Over the years, most of the images they produce in that scene have been reasonably, expectedly sharp, just there is the occasional miss. I see this as nothing more than that and DPReview's response confirms that.
Steve Hendrix/CI
-
Hi,
I have found that Imaging Resource test shots suck. Absolutely useless for sharpness evaluation. The main reason is that slanted edge targets are not flat. Also, the targets used are probably not proper size for sharpness evaluation. For 24x36 mm you would need a target like 100x150 cm, absolutely well aligned. I think IR falls short on that.
Lens gear tests may be better, but sample images (on IR) are simply useless!
Best regards
Erik
Imaging-resource also have their samples up.
-
Hi,
I have had some private communication with Rishi Sanyal. He says that they have not achieved optimal focus. DPReviews will reshoot the image, once they have final firmware from Fuji.
I think that this is a difficult time for testers. Access to test samples are limited while potential buyers are much interested in early results, but early results are error prone.
Best regards
Erik
Hi,
The Fuji GFX 100 holds a lot of promise, offering a modern Sony CMOS sensor at a reasonable 10k$US price point. Therefore I eagerly awaited DPRs studio test image. It seems that there is some serious shortcoming of that test image.
(http://echophoto.dnsalias.net/ekr/Articles/TMP/GFX100_MTF.png)
Measuring and plotting MTF indicates that sharpness is worse than on the GFX 50S.
So, what is the problem?
- It could be a bad sample of the lens. I would consider this to be a possible explanation.
- Correct focus has not been achieved.
- Camera vibration?
- IBS causing unsharpness?
From the EXIF, it seems that electronic shutter was used.
I have mailed Rishi Sanyal at DPReview a few days ago, no response yet.
Best regards
Erik
-
There has also been at least on post on a Fuji rumor site that all the lenses will need a firmware update to accommodate the focusing technology in the 100.
That may have contributed to the problem with the test.
-
Hi,
That may be a good point. When I test lenses I would use an automatic focusing rail, like Stackshot and take images at 2.5 mm intervals. That is useful for finding out things about the lens, but it is not practical in the field.
Finding out maximum performance, fine stepped focus bracketing makes some sense.
Best regards
Erik
There has also been at least on post on a Fuji rumor site that all the lenses will need a firmware update to accommodate the focusing technology in the 100.
That may have contributed to the problem with the test.
-
Hi,
I have had some private communication with Rishi Sanyal. He says that they have not achieved optimal focus. DPReviews will reshoot the image, once they have final firmware from Fuji.
I think that this is a difficult time for testers. Access to test samples are limited while potential buyers are much interested in early results, but early results are error prone.
Best regards
Erik
Thank you, Erik, for detecting and reporting the issue with DPR's studio scene.
-
Hi,
What Rishi mailed me, they revised their protocols to assure optimal focus. But he said that the final images had bad focus.
I am not doing lens testing a lot but I learned to use an automated focusing rail and shoot say 40 images 2.5 mm apart to get optimal focus.
With focus by wire focusing, firmware is involved in maintaining focus, especially with linear motors.
Best regards
Erik
I'm glad they confirmed what I suspected, that they mis-focused.
However, this does not alarm me in any way, nor does it trigger any concerns over the Fuji lenses or getting sharp images out of the GFX 100 sensor. I have seen occasional mis-focuses from DP Review tests for years, even on little micro 4/3 sensor 16mp images. So this has nothing to do with increasingly difficult tolerances, lots of pixels, etc.
Over the years, most of the images they produce in that scene have been reasonably, expectedly sharp, just there is the occasional miss. I see this as nothing more than that and DPReview's response confirms that.
Steve Hendrix/CI