Pages: 1 [2] 3 4 5   Go Down

Author Topic: GX 100 returned as it could not match my GFX 50S  (Read 3941 times)

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2281
    • The Last Word
Re: GX 100 returned as it could not match my GFX 50S
« Reply #20 on: January 03, 2020, 06:34:52 pm »

"Still also have my fingers crossed that Fuji will come out with some form of pixel shift as that has shown on Pentax K1 to not only provide an increase in resolution but also DMAX. "

Paul, if you are ever up this way shout and let's get together.  I have an Airstream BaseCamp X and love to camp in that critter with my Lady Qing!  So give me an excuse!

As to the Dmax, I have found some interesting results using Aurora HDR 2019 version.  I am shooting with careful exposure for the highlights in a scene and not spending much bandwidth on the shadows.  The I take the RAW file from the 50S into Aurora as a single file and let it process.  Once you get a handle on the uses of their different sliders, I am amazed at the Dmax I'm pulling out of the RAW file without loosing my highlights!  Far more than LR or PS can give me and without the phony HDR look.

Takes a bit of practice, but then I've been in the world of HDR to get Dmax since I first played with Photomatrix Pro years ago - I still keep an updated version of it on my Mac Pro.

Jack

You are using Dmax in a manner with which I am unfamiliar. In a print, I know it as the log base 10 of unity over the minimum reflectivity. In a transparency, it's the log base 10 of unity over the minimum transmittance.

Can you give me your definition?

Jim

Lust4Life

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 705
    • http://www.shadowsdancing.com
Re: GX 100 returned as it could not match my GFX 50S
« Reply #21 on: January 03, 2020, 07:01:39 pm »

Jim, I'm ancient in the world of digital so my age is showing!  LOL!
Guess I need to update my language skills to how Dmax is expressed today.

I used to develop software on the super computers Cray/SGI platform for 12 years before I dropped out at 46 to continue my own studies.
Put together a small team of kids out of MIT and GA Tech as a SGI Power Series VAR and developer.

That took me into working in many areas of digital, initially 3D realms, but lead me into a relationship with Eikonix when it was embryo and building it's first digital camera.  We expressed Dmax as the maximum spread of visual that could be distinguished from the Eikonix array. 

I then worked with "Leaf" in developing drivers for the Leaf 45 scanner, and later to the Howtek HiResolve 8K drum scanner, Nikon and Hasselblad's (very short relationship with Hassie as they suffered from NIH syndrome - not invented here) first digital cameras.

We always expressed how many stops of dynamic range we could "pull out of an array" as "Dmax".  It's just the number of stops of range from maximum distinguishable white to black.  When we passed a Dmax of 7 (7stops) I got rid of all my film gear and went totally digital with a P45+ mounted on a 500cm and 100 T* lens on the nose.  It took the P45+ back at 39MP to beat anything I could pull out of film and get a decent scan on my Howtek 8K scanner - I had my own Wing Linch processor and developed routines to really pull the film to give me what I wanted. 

Have never gone back to film.

So, what term should I update my "little gray cells" to use to represent what I call Dmax?

PS:  Eikonix was purchased by Kodak in an attempt to keep film alive and kill digital.  (At the time of the buy out, Eikonix was headed by a very bright chap named George Helms that I respected.)  It worked for a while but ..........
Logged
Jack

Lust4Life

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 705
    • http://www.shadowsdancing.com
Re: GX 100 returned as it could not match my GFX 50S
« Reply #22 on: January 03, 2020, 07:32:51 pm »

I just did icuts of the screen for each of the camera for anyone that wants to see what I experience between the 50S and the 100.

Sharpening was stopped at the point going any further only degraded the respective images.

Capture One v. 20 used for both.  Both zoomed in 300% in C1 before icut was done.

Jack
« Last Edit: January 03, 2020, 07:43:15 pm by Lust4Life »
Logged
Jack

Martin Kristiansen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1392
    • Martin Kristiansen
Re: GX 100 returned as it could not match my GFX 50S
« Reply #23 on: January 04, 2020, 03:20:27 am »

Jim, I'm ancient in the world of digital so my age is showing!  LOL!
Guess I need to update my language skills to how Dmax is expressed today.

I used to develop software on the super computers Cray/SGI platform for 12 years before I dropped out at 46 to continue my own studies.
Put together a small team of kids out of MIT and GA Tech as a SGI Power Series VAR and developer.

That took me into working in many areas of digital, initially 3D realms, but lead me into a relationship with Eikonix when it was embryo and building it's first digital camera.  We expressed Dmax as the maximum spread of visual that could be distinguished from the Eikonix array. 

I then worked with "Leaf" in developing drivers for the Leaf 45 scanner, and later to the Howtek HiResolve 8K drum scanner, Nikon and Hasselblad's (very short relationship with Hassie as they suffered from NIH syndrome - not invented here) first digital cameras.

We always expressed how many stops of dynamic range we could "pull out of an array" as "Dmax".  It's just the number of stops of range from maximum distinguishable white to black.  When we passed a Dmax of 7 (7stops) I got rid of all my film gear and went totally digital with a P45+ mounted on a 500cm and 100 T* lens on the nose.  It took the P45+ back at 39MP to beat anything I could pull out of film and get a decent scan on my Howtek 8K scanner - I had my own Wing Linch processor and developed routines to really pull the film to give me what I wanted. 

Have never gone back to film.

So, what term should I update my "little gray cells" to use to represent what I call Dmax?

PS:  Eikonix was purchased by Kodak in an attempt to keep film alive and kill digital.  (At the time of the buy out, Eikonix was headed by a very bright chap named George Helms that I respected.)  It worked for a while but ..........

Well that’s a relief. I had no idea what you meant when when speaking of Dmax. My background was photographic sensitometry and Dmax meant a whole other thing. Digitally at a sensor level I suppose Dmax to me would mean a sensor receiving no exposure so all that would impact it would be noise. I simply couldn’t understand the fuss. Thanks for clearing that up. You have had an interesting career.
Logged
Commercial photography is 10% inspiration and 90% moving furniture around.

nazdravanul

  • Full Member
  • ***
  • Offline Offline
  • Posts: 108
    • Stefan Iacob - visual artist
Re: GX 100 returned as it could not match my GFX 50S
« Reply #24 on: January 04, 2020, 04:27:36 am »

I just did icuts of the screen for each of the camera for anyone that wants to see what I experience between the 50S and the 100.

Sharpening was stopped at the point going any further only degraded the respective images.

Capture One v. 20 used for both.  Both zoomed in 300% in C1 before icut was done.

Jack

Well, if 300% zoomed in equivalent sharpness was your benchmark, you did good to send the GFX100 back.
But for me that comparison makes no sense. Because we don’t use “equivalent zoomed in at x%” images, when printing - we use the same print size (for web display the whole resolution discussion is meaningless, due to the reduced image sizes, color depth and pixel density). So, at the SAME PRINT SIZE, your GFX 100 file should be zoomed in at 200% vs 300% for the GFX 50, to make the comparison meaningful for real world photographic output - zooming in without any final image dimensions output reference is just empty pixel peeping. But even for pixel peeping, the “pixels per duck”  reference should be the meaningful reference - the details per home or per flag, looking at your image. Also, part of the perceived sharpness of the 50 is simply fake detail, digital artefacts - look at the ugly aliasing on the left balcony. Again, if the question is “Which pixel is sharper ?” your approach is fair (in fact we’ve seen this with each new increase in MPs, you don’t have to ever buy and test a higher pixel density camera to see softer zoomed in pixels, that will always be the case) But if the question is “Which camera gives me more real detail per subject ?’” you might want to go back to those files and process / review them differently. Also a critical question is f-stop. To me the 100 seems dulled down from diffraction - going over f8 will simply degrade the detail and negate any resolution advantage (an f11 or f16 image on the GFX 100 will not give you anything over the 50 MP reference). What f-stop did you use ? (also was IBIS on or off  ?  as that can really mess up my GFX 100 images).
Logged

Lust4Life

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 705
    • http://www.shadowsdancing.com
Re: GX 100 returned as it could not match my GFX 50S
« Reply #25 on: January 04, 2020, 06:33:51 am »

Well, if 300% zoomed in equivalent sharpness was your benchmark, you did good to send the GFX100 back.
But for me that comparison makes no sense. Because we don’t use “equivalent zoomed in at x%” images, when printing - we use the same print size (for web display the whole resolution discussion is meaningless, due to the reduced image sizes, color depth and pixel density). So, at the SAME PRINT SIZE, your GFX 100 file should be zoomed in at 200% vs 300% for the GFX 50, to make the comparison meaningful for real world photographic output - zooming in without any final image dimensions output reference is just empty pixel peeping. But even for pixel peeping, the “pixels per duck”  reference should be the meaningful reference - the details per home or per flag, looking at your image. Also, part of the perceived sharpness of the 50 is simply fake detail, digital artefacts - look at the ugly aliasing on the left balcony. Again, if the question is “Which pixel is sharper ?” your approach is fair (in fact we’ve seen this with each new increase in MPs, you don’t have to ever buy and test a higher pixel density camera to see softer zoomed in pixels, that will always be the case) But if the question is “Which camera gives me more real detail per subject ?’” you might want to go back to those files and process / review them differently. Also a critical question is f-stop. To me the 100 seems dulled down from diffraction - going over f8 will simply degrade the detail and negate any resolution advantage (an f11 or f16 image on the GFX 100 will not give you anything over the 50 MP reference). What f-stop did you use ? (also was IBIS on or off  ?  as that can really mess up my GFX 100 images).

1.  I agree completely with your comment about 300% has no relative value in light of the print.  I used it as it is what Jim, who's work I have respected for many years, referenced for alias and it answered two questions at once. 

2.  My decision to send it back was based on far more than one factor, but that was the primary one.  I evaluate the net worth of an array by what I get at 100%.  I use in my work and evaluation 100% as that is the base one could "up-res" from if you had to go to a larger print than what the native data will give you.

3.  Yes, a huge flaw in my eye's view of the files is the GFX 100 image is truly "flat".  I shot at f4 without IBS in both examples - of course not the optimal sweet spot of the 23, but what I have found to be a good setting for evaluation of landscapes - next is f11 in my mind on GFX lenses.  I was not concerned with the flag, rather the joints in the buildings being "flat" and not crisp on the detail area of the indents between exterior wall panels.

In short, I think we are in agreement, though I do not entirely agree with your comment about there being a threshold of detail that is lost in any increase of pixels.  My work says you are correct IF we pack more pixels into the same size silicon, but not if we keep the size of the wells the same and increase the footprint silicon accordingly.  Thus, I'll take 60MP on a substantially larger size silicon base than 100MP on the space of the GFX 100 chip.  Keys are of course that the DAC, lens, etc are also improved to compensate for the extra load of capturing and processing the data gathered. 

In short, all I'm asking for is sharpness at 100%, controlled aliasing pollution and Dmax at 15 stops for $10K or under.   
Today, that is too much to ask for, but it will come but in a larger format than what Fuji is calling MF.  I call it "MMF"/MINI Medium Format.
The axiom is no different from film at 4x5 compared to 8x10, etc.

Attached you'll find the same files at 100% and I think it is far more revealing what selling your used GFX 50S for $3k and then throwing in an extra $7K will get you.  Not with my money!!

A tangent thought not relative to this thread directly, but rather a reflection of the passing of time in my life, I'm now 73:
I have a very distinct memory that goes back to when the Leaf 45+ was first sent to me to evaluate.  I shot a 4x5 scene with the best optics of the time and the same scene with the P45+.  Processed the sheet in my "best soup", scanned it on my Howtek drum scanner an compared it to the P45+.  This moment in my mind was quite consequential, and in a negative way to "Fine Art Photography".

I purchased the final production P45+, mated it to my 500cm and my 100 T* lens and started shooting, then scanning at 8K.  I have an image over my fireplace that is quite large which is one of the very first I did with that set up.  It's resourced from two rows of 5 shots per row and three shots deep (-, 0, + 1 stop) for each of the 10 frames.  The file is HUGE and the work to generate it was extensive BUT  man is it sharp and everything but flat!  I've wasted a lot of time and money by not just sticking with the setup.

I've been ranting in my personal notes about that moment after several chats with Carol at Photography West Gallery in Carmel.
Think I'll start a thread on that subject, but I expect I'll have to wear a kevlar vest  once I write down my thoughts for public view!  LOL!
« Last Edit: January 04, 2020, 11:22:26 am by Lust4Life »
Logged
Jack

Rand47

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1606
Re: GX 100 returned as it could not match my GFX 50S
« Reply #26 on: January 04, 2020, 10:33:40 am »

Quote
... “Well, if 300% zoomed in equivalent sharpness was your benchmark, you did good to send the GFX100 back.
But for me that comparison makes no sense. Because we don’t use “equivalent zoomed in at x%” images, when printing - we use the same print size. . .  “

Same here.  I’ve always been confused by this approach.  Admittedly, I’m not a technical person.  When I compare A to B I’ll take identical images in terms of angle of view, f/stop and other parameters.  Then, I’ll make identical prints - say 20x30” or so.  Put them up side by side on my 8’ magnetic wall, illuminated by Solux 4700k light, and compare the results.   Final output is all that matters to me. 

In my case, compared in this manner, the GFX 100 blows the 50s in the dirt - though neither camera is a slouch in image quality.

Rand
« Last Edit: January 04, 2020, 10:37:55 am by Rand47 »
Logged
Rand Scott Adams

elliot_n

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 932
Re: GX 100 returned as it could not match my GFX 50S
« Reply #27 on: January 04, 2020, 11:10:34 am »

Attached you'll find the same files at 100% and I think it is far more revealing...

Please update - you've posted the GFX50 file twice.
Logged

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2281
    • The Last Word
Re: GX 100 returned as it could not match my GFX 50S
« Reply #28 on: January 04, 2020, 11:17:40 am »

So, what term should I update my "little gray cells" to use to represent what I call Dmax?

Photographic Dynamic Range? It's usually expressed in Log base 2 not base 10, though.

https://blog.kasson.com/the-last-word/isolessness-comparisons-across-resolutions/

Jim

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2281
    • The Last Word
Re: GX 100 returned as it could not match my GFX 50S
« Reply #29 on: January 04, 2020, 11:22:29 am »

By the way, from the raw files that I've looked at, the pixel-level sharpness of the IQ4 150 and the GFX 100 are virtually identical. That makes the IQ4 150 sqrt(150/100) - 1 =  22% sharper at the picture level.

Jim

Lust4Life

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 705
    • http://www.shadowsdancing.com
Re: GX 100 returned as it could not match my GFX 50S
« Reply #30 on: January 04, 2020, 11:24:02 am »

Please update - you've posted the GFX50 file twice.

Opps, Sorry about that Folks!

Corrected that error,
Jack
Logged
Jack

Lust4Life

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 705
    • http://www.shadowsdancing.com
Re: GX 100 returned as it could not match my GFX 50S
« Reply #31 on: January 04, 2020, 11:25:42 am »

By the way, from the raw files that I've looked at, the pixel-level sharpness of the IQ4 150 and the GFX 100 are virtually identical. That makes the IQ4 150 sqrt(150/100) - 1 =  22% sharper at the picture level.

Jim

Jim, I wish the price point was also level between the IQ and the GFX!!  :-)

Logged
Jack

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2281
    • The Last Word
Re: GX 100 returned as it could not match my GFX 50S
« Reply #32 on: January 04, 2020, 11:27:30 am »

Jim, I wish the price point was also level between the IQ and the GFX!!  :-)

You didn't like the pixel-level sharpness of the GFX 100, so you wouldn't like the pixel-level sharpness of the IQ4 150. I don't think pixel-level sharpness is an appropriate metric, but chacun a son gout.

Jim

alan_b

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 288
    • West Coast Architecture + Interiors Photographer
Re: GX 100 returned as it could not match my GFX 50S
« Reply #33 on: January 04, 2020, 12:05:24 pm »

We always expressed how many stops of dynamic range we could "pull out of an array" as "Dmax".  It's just the number of stops of range from maximum distinguishable white to black.
...
So, what term should I update my "little gray cells" to use to represent what I call Dmax?

You already said it: Dynamic Range (DR)

Dmax is a single point on that range, as is Dmin.  DR = Dmax - Dmin
Logged

Lust4Life

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 705
    • http://www.shadowsdancing.com
Re: GX 100 returned as it could not match my GFX 50S
« Reply #34 on: January 04, 2020, 12:14:17 pm »

Back when I had hair on top of this Beast, we used the term not as an absolute, but a spread - a range; thus Dmax of 10 stops.
But what did we know - just building one of the first cameras!  LOL!

So much has been done since that time!
I wonder if RIT has one of the original Eikonix cameras in their museum?

I used to have one of the original arrays - I guess I should use the modern term of Sensor - but it was lost in a move.

Dynamic Range makes more sense - I'll have to work hard and re-program the gray cells!
Logged
Jack

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2281
    • The Last Word
Re: GX 100 returned as it could not match my GFX 50S
« Reply #35 on: January 04, 2020, 12:16:52 pm »

You already said it: Dynamic Range (DR)

Dmax is a single point on that range, as is Dmin.  DR = Dmax - Dmin

Actually, the dynamic range of a system starts with a definition of the minimal acceptable SNR. The signal level that produced that SNR defines the digital equivalent of Dmax. The concept of Dmin has no utility in sensor evaluation, since full scale is the reference white always.

Jim

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2281
    • The Last Word
Re: GX 100 returned as it could not match my GFX 50S
« Reply #36 on: January 04, 2020, 12:20:03 pm »

Back when I had hair on top of this Beast, we used the term not as an absolute, but a spread - a range; thus Dmax of 10 stops.
But what did we know - just building one of the first cameras!  LOL!

Even with film sensitometry, Dmax was not a range, but the maximum density achievable.

Congratulations on being involved with building one of the first cameras. You must be about 200 years old.

Jim

Lust4Life

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 705
    • http://www.shadowsdancing.com
Re: GX 100 returned as it could not match my GFX 50S
« Reply #37 on: January 04, 2020, 12:22:31 pm »

Actually, the dynamic range of a system starts with a definition of the minimal acceptable SNR. The signal level that produced that SNR defines the digital equivalent of Dmax. The concept of Dmin has no utility in sensor evaluation, since full scale is the reference white always.

Jim

So maybe I'm not just "over the hill"?

We used it to define similar to the Zone system values - how many stops of "white" with definable image detail to how many "blacks" with definable image detail.  Just shadow detail to white detail.  We did not count absolute black and white.

Logged
Jack

Lust4Life

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 705
    • http://www.shadowsdancing.com
Re: GX 100 returned as it could not match my GFX 50S
« Reply #38 on: January 04, 2020, 12:23:50 pm »

Even with film sensitometry, Dmax was not a range, but the maximum density achievable.

Congratulations on being involved with building one of the first cameras. You must be about 200 years old.

Jim

OK Jim - YOU just crossed the line of tolerable behavior!
Just because I look 200 years old does not mean I am!!
:-)
Logged
Jack

Martin Kristiansen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1392
    • Martin Kristiansen
Re: GX 100 returned as it could not match my GFX 50S
« Reply #39 on: January 04, 2020, 01:09:10 pm »

So maybe I'm not just "over the hill"?

We used it to define similar to the Zone system values - how many stops of "white" with definable image detail to how many "blacks" with definable image detail.  Just shadow detail to white detail.  We did not count absolute black and white.

We would determine dynamic range of film by measuring how many stops of exposure we got on the straight line portion of the curve when the film was processed for a gamma of 0,62 for black and white film. This was so long ago I may have some figures wrong but that was pretty much the methodology. Dmax was never of huge interest when looking at film exposed with a sensitometer. Speed point, base fog, gamma were critical. Dynamic range was obviously important. I’m leaving stuff out of course.

It’s interesting to me the search for ever more DR. We have gone from Ciba prints with about 3 stops max to 15 stops or thereabouts now. My biggest issue is finding a way to map all that in a plausible manner to what ever the output might be, print or screen. 

Sorry for the diversion on this thread. I’m finding it fascinating even though I no longer use MFDB.

Logged
Commercial photography is 10% inspiration and 90% moving furniture around.
Pages: 1 [2] 3 4 5   Go Up