GFX 100 re PDAF banding - now fixed.

[Rand47]

According to Jim Kasson's most recent tests:

GFX 100 PDAF banding is fixed - the last word

When I tested the GFX 100, I found an effect called PDAF banding, in which Fuji, in a misguided attempt to mitigate PDAF striping -- caused by reflections from mettalic bits in on-sensor PDAF pixels -- created a far worse problem: dark bands in boosted deep shadows. When I tested the GFX 100s...

blog.kasson.com

Rand

 

[Paul2660]

I still find this the most over blown problem with a camera I can remember. Problem, banding when you push a file +5 stops. OK, Not much else is very good when you push a file 5+ stops at least from what I have seen over the years.

Edit:

I wanted to add, I have used the GFX100 since announcement, and have loved it. Worked hard to find the banding at first then simply quit as I just never saw it, in my work. I also realized that the GFX100, can't give an image a 5 shot push without excessive noise, (note 5 stops is basically pure black). I started bracketing all my shots with the GFX100, and still do. What I do find concerning even 2 years later (or close to it) are:

1. The GFX100 files are not as malleable as I had expected, and noise can and will creep into some of them even at base ISO if pushed much past 2.5 stops.
2. The GFX100, still can't reliably lock AF in low light or especially low contrast conditions, ever with the PDAF capability of the sensor. For for sunrise and sunset shots or foggy morning work, this leads to hunting around at times, and many times not acquiring the best focus.
3. Fuji has the same sensor as the IQ4 and a "partnership" with Phase One at least in regards to Capture One, Fuji could learn a lot from either a Frame averaging or Dual exposure solution. It's possible the camera can't process this due to internal limitations but the 100s should have been upgraded to accommodate this.
4. The GFX100 even with the fastest SD card will buffer out pretty quickly when using 16 bit file and shooting bracketing. Forcing a move to the 14 bit files, which so far have not had the same problem.

Still the unit is hand holdable even with a 250mm lens something that is impossible for me with Phase.


Paul C

 

[SrMphoto]

I still find this the most over blown problem with a camera I can remember. Problem, banding when you push a file +5 stops. OK, Not much else is very good when you push a file 5+ stops at least from what I have seen over the years.

Edit:

I wanted to add, I have used the GFX100 since announcement, and have loved it. Worked hard to find the banding at first then simply quit as I just never saw it, in my work. I also realized that the GFX100, can't give an image a 5 shot push without excessive noise, (note 5 stops is basically pure black). I started bracketing all my shots with the GFX100, and still do. What I do find concerning even 2 years later (or close to it) are:

1. The GFX100 files are not as malleable as I had expected, and noise can and will creep into some of them even at base ISO if pushed much past 2.5 stops.
2. The GFX100, still can't reliably lock AF in low light or especially low contrast conditions, ever with the PDAF capability of the sensor. For for sunrise and sunset shots or foggy morning work, this leads to hunting around at times, and many times not acquiring the best focus.
3. Fuji has the same sensor as the IQ4 and a "partnership" with Phase One at least in regards to Capture One, Fuji could learn a lot from either a Frame averaging or Dual exposure solution. It's possible the camera can't process this due to internal limitations but the 100s should have been upgraded to accommodate this.
4. The GFX100 even with the fastest SD card will buffer out pretty quickly when using 16 bit file and shooting bracketing. Forcing a move to the 14 bit files, which so far have not had the same problem.

Still the unit is hand holdable even with a 250mm lens something that is impossible for me with Phase.


Paul C

Adding frame averaging would really help. Somehow it is not on the marketing guys' radar . Olympus has implemented frame averaging in its latest cameras. I can produce files that equal GFX100 in noise characteristics. However, they are marketing it only as an ND-filter replacement.
Phase One and Capture One are now two different companies (new CEO since 12/2020). I do not think that Fuji has access to Phase One engineering.
I am not sure whether to shoot GFX100S in 16 or 14-bit mode. The camera is more sluggish with 16-bit (longer delay after taking a shot) and I am not seeing any benefits at the moment.

 

[Paul2660]

I have followed the Phase One and Capture One separate direction etc. I still feel behind the curtains it’s all still P1 since Capture One still doesn’t support Hasselblad. I was glad to see the marketing agreement between Fuji and Capture One/Phase one.

I have started only using 14 bit raw due to the same sluggish behavior in writes.

Paul C

 

[SrMphoto]

In V21, Capture One also added support for Leica S (Typ 007) and S3. I still hope that they will add support for Hasselblads as well.

 

[budfox]

Maybe when HB discontinue their larger format sensors that might happen, leaving P1 as the only real operator in that space. Until then, I think the competition between P1 and HB at the high end (despite being 'different' companies) means C1 won't process HB.

 

[drevil]

"now fixed" not that it actually ever was a real problem

 

[Christopher]

I don’t think it was a big problem. However, it still was there, which it shouldn’t have been as it was created by Fuji firmware.

 

[Rand47]

Just a quick comment for those experiencing “buffering out’ or sluggish write time in 16 bit mode. My GFX 100 has been in 16 bit mode, uncompressed since day one. I use Sony G Tough 128 SD cards. I shoot raw + jpeg. I’ve yet to fill the buffer or have to wait for the camera. I routinely shoot focus brackets of “a lot” of exposures, with zero time between shots. In my experience with Fujifilm cameras, they are very fussy about what cards are used. The specs of the cards don’t seem to be the “only thing” that matters. E.g. I completely quit using Lexar cards with any of my Fuji cameras. They just didn’t “like” them. No clue why.

Also the “PDAF banding issue“ was never an issue for me, but I’m glad that Fuji “fixed’ it since it was fixable.

Rand

 

[SrMphoto]

Just a quick comment for those experiencing “buffering out’ or sluggish write time in 16 bit mode. My GFX 100 has been in 16 bit mode, uncompressed since day one. I use Sony G Tough 128 SD cards. I shoot raw + jpeg. I’ve yet to fill the buffer or have to wait for the camera. I routinely shoot focus brackets of “a lot” of exposures, with zero time between shots. In my experience with Fujifilm cameras, they are very fussy about what cards are used. The specs of the cards don’t seem to be the “only thing” that matters. E.g. I completely quit using Lexar cards with any of my Fuji cameras. They just didn’t “like” them. No clue why.

Also the “PDAF banding issue“ was never an issue for me, but I’m glad that Fuji “fixed’ it since it was fixable.

Rand

Could you elaborate on why you shoot uncompressed and why 16-bit? Thanks.

 

[Rand47]

Could you elaborate on why you shoot uncompressed and why 16-bit? Thanks.

Several reasons, none of them based on scientific testing, but only my sense of things (i.e. not worth much).
1. Storage is cheap.
2. I see zero down side to doing so.
3. Even if there is only limited benefit in “some image types” how would I know when to switch over? (Especially if there’s no down side to shooting this way as default [other than file size].)
4. I’m a fiend about smooth tonal transitions.
5. I don’t think Fuji just did it for advertising specsmanship.

Rand

 

[rdeloe]

Several reasons, none of them based on scientific testing, but only my sense of things (i.e. not worth much).
1. Storage is cheap.
2. I see zero down side to doing so.
3. Even if there is only limited benefit in “some image types” how would I know when to switch over? (Especially if there’s no down side to shooting this way as default [other than file size].)
4. I’m a fiend about smooth tonal transitions.
5. I don’t think Fuji just did it for advertising specsmanship.

Rand

For what it's worth Rand, Jim Kasson's recent testing of the 100 seems to demonstrate that 16-bit now does offer a bump in some situations.

On point 3, the larger issue for me would be remembering to turn it on and off based on situation!

 

[SrMphoto]

Several reasons, none of them based on scientific testing, but only my sense of things (i.e. not worth much).
1. Storage is cheap.
2. I see zero down side to doing so.
3. Even if there is only limited benefit in “some image types” how would I know when to switch over? (Especially if there’s no down side to shooting this way as default [other than file size].)
4. I’m a fiend about smooth tonal transitions.
5. I don’t think Fuji just did it for advertising specsmanship.

Rand

Thank you for the answers.
What benefit do you expect from shooting uncompressed vs. lossless compressed?
AFAIK, the jury is still out if 16-bit brings any benefits in Fuji cameras, but I do not think that it helps with tonal transitions. I believe it is mostly about DR. I am undecided if I should shoot 14 or 16-bit. I probably will start using the 16-bit mode, just in case ;-).
IMO, the only relevant downside of 16-bit mode is the slower readout (electronic shutter) and a bit more sluggish camera behavior.

Raw bit depth is about dynamic range, not the number of colors you get to capture

 

[Rand47]

Thank you for the answers.
What benefit do you expect from shooting uncompressed vs. lossless compressed?
AFAIK, the jury is still out if 16-bit brings any benefits in Fuji cameras, but I do not think that it helps with tonal transitions. I believe it is mostly about DR. I am undecided if I should shoot 14 or 16-bit. I probably will start using the 16-bit mode, just in case ;-).
IMO, the only relevant downside of 16-bit mode is the slower readout (electronic shutter) and a bit more sluggish camera behavior.

Raw bit depth is about dynamic range, not the number of colors you get to capture

Here’s my probably faulty logic. It isn’t about benefit, exactly, it is about what I can do. Here’s my logic (faulty or not). If 14 bit, lossless compression was completely equal to 16 bit uncompressed - truly - then why bother to have it as an option on the camera? Silly thinking perhaps, but since it costs me nothing to shoot “unbridled” so to speak, why not? As to camera sluggishness, I’ve experienced exactly ZERO of that with my GFX 100. I’ve never had to wait for the camera, ever, except when doing long exposures and waiting for the long exposure noise reduction dark frame to process. Again, the camera has performance mode, and that’s my default as well. It eats more batteries they say. I say who cares? I’ve yet to go through both batteries in any single day’s shooting. I’ve seen posts in the past complaining about jumpy EVF etc. and come to find out the camera is in econo mode. For me that’s like buying a Ferrari and keeping the RPM’s under 2k so that I get good gas milage.

As far as slower readout speed with ES, since ES is only used (for me) in very static situations, the read-speed is irrelevant. If the difference in read speed ever became evident, that is to say “problematic,” it would be a simple thing to engage 14 bit. My practical sense is, that any image distortion problems caused by inappropriate ES use would not be “solved” by speeding it up nominally by using 14 bit mode.

Thanks for the link to the article! Very informative. So, even if 16 bit only provides a slight increase in DR, why not?

Rand

 

[SrMphoto]

Here’s my probably faulty logic. It isn’t about benefit, exactly, it is about what I can do. Here’s my logic (faulty or not). If 14 bit, lossless compression was completely equal to 16 bit uncompressed - truly - then why bother to have it as an option on the camera? Silly thinking perhaps, but since it costs me nothing to shoot “unbridled” so to speak, why not? As to camera sluggishness, I’ve experienced exactly ZERO of that with my GFX 100. I’ve never had to wait for the camera, ever, except when doing long exposures and waiting for the long exposure noise reduction dark frame to process. Again, the camera has performance mode, and that’s my default as well. It eats more batteries they say. I say who cares? I’ve yet to go through both batteries in any single day’s shooting. I’ve seen posts in the past complaining about jumpy EVF etc. and come to find out the camera is in econo mode. For me that’s like buying a Ferrari and keeping the RPM’s under 2k so that I get good gas milage.

As far as slower readout speed with ES, since ES is only used (for me) in very static situations, the read-speed is irrelevant. If the difference in read speed ever became evident, that is to say “problematic,” it would be a simple thing to engage 14 bit. My practical sense is, that any image distortion problems caused by inappropriate ES use would not be “solved” by speeding it up nominally by using 14 bit mode.

Thanks for the link to the article! Very informative. So, even if 16 bit only provides a slight increase in DR, why not?

Rand

Thank you for the thoughts.

I also wondered why there is an uncompressed mode. Jim Kasson wrote:
Uncompressed files show reduced impact of bit errors in the file that can occur over time, compared to losslessly compressed files. In a compressed file, an error can affect multiple pixels.

That is the only advantage of uncompressed files, AFAIK. I will stay in uncompressed mode.

On GFX100S, 16-bit mode, there is a slightly longer delay after taking an image. I do not remember how it was on GFX100. I will alternate between 14 and 16-bit modes.