Pages: [1]   Go Down

Author Topic: DPReview reshot the GFX 100 images with amazing results...  (Read 1798 times)

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
DPReview reshot the GFX 100 images with amazing results...
« on: July 09, 2019, 03:36:39 am »

Hi,

DPReview has reshot the GFX 100 images with amazing results. Check links below:

https://www.dpreview.com/forums/post/62863644

https://www.dpreview.com/forums/post/62861703

If you think GFX colors are bad:

https://www.dpreview.com/forums/post/62863074

Concern for Phase One? Sure! That said, the GFX 100 is not a digital back, although it may be usable on some technical cameras.

Best regards
Erik
Logged
Erik Kaffehr
 

narikin

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1371
Re: DPReview reshot the GFX 100 images with amazing results...
« Reply #1 on: July 09, 2019, 04:43:50 am »

Yes, I noticed this update too Erik.

It's an impressive result, but:

(a) shame on DPreview for releasing out of focus test shots originally.
(b) it does not compare like with like, until they shoot IQ4-150 against it. BSI vs BSI, same generation Sony sensors.
(c) yes colors are far better on the Phase back in C1, You didn't think Phase would allow Fuji to give better color outputs in Capture One, did you? ;-)
(d) Have you considered refining and releasing your Lumariver GFX-100 profile for download?

I'm traveling now, but wonder if anyone has tried processing GFX-100 Raw files using IQ4-150 color profile in C1? Same generation Sony BSI sensor might give interesting results, though likely the CFA will have different spectral qualities, so not super accurate.

 
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: DPReview reshot the GFX 100 images with amazing results...
« Reply #2 on: July 09, 2019, 05:19:04 am »

Hi Narikin,

Your posting is most appreciated!

Comments, see below.

Best regards
Erik

Yes, I noticed this update too Erik.

It's an impressive result, but:

(a) shame on DPreview for releasing out of focus test shots originally.
Yes, I would agree, they added some writing on the images needing a reshot. I was slightly involved in that.

(b) it does not compare like with like, until they shoot IQ4-150 against it. BSI vs BSI, same generation Sony sensors.
Yes, of course. Actually, I think that BSI really matters until use it on a technical camera. BSI essentially reduces vignetting and cross talk.

(c) yes colors are far better on the Phase back in C1, You didn't think Phase would allow Fuji to give better color outputs in Capture One, did you? ;-)
The raw image in DPReview test shot was processed in ACR, I think.

(d) Have you considered refining and releasing your Lumariver GFX-100 profile for download?
Hi, I may do that. It is OK as is. Problem is that we really need a dual illuminant profile and for that I need two shots essentially Standard Illuminant A and Standard Illuminant D50 (or D65).

I'm traveling now, but wonder if anyone has tried processing GFX-100 Raw files using IQ4-150 color profile in C1? Same generation Sony BSI sensor might give interesting results, though likely the CFA will have different spectral qualities, so not super accurate.

The Lumariver GFX-100 profile will be more accurate than C1, that used be the case any time I tested, but the C1 profile may be more pleasant. Also, C1 uses a more aggressive tone curve than ACR. Lumariver can generate DCP profiles with either ACR or C1 tone curve.
Logged
Erik Kaffehr
 

Paul2660

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4066
    • Photos of Arkansas
Re: DPReview reshot the GFX 100 images with amazing results...
« Reply #3 on: July 09, 2019, 07:50:04 am »

If you have to shoot at F 5.6 to get the best results with the GFX100 it kinda limits your shooting. I never had any issues with the GFX50s at F11 and even F14. If diffraction was visible post production sharpening allowed for a good image.

Paul C
Logged
Paul Caldwell
Little Rock, Arkansas U.S.
www.photosofarkansas.com

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: DPReview reshot the GFX 100 images with amazing results...
« Reply #4 on: July 09, 2019, 09:55:34 am »

Hi Paul,

This plot from Jim Kasson's test of the GF 110/2 on the GFX 50S shows how things work:



So, you can see that lens reaches peak performance, on axis, at f/2.8. 3500 cy/PH is extremely high. Stopped down f/8 it reaches 2300. It loses a lot, but 2300 cy/PH is still an extremely good value. Putting it a bit in perspective, I could reach about 1950 cy/PH at f/5.6 on my Sonnar 180/4 with extremely careful focusing.

You have a good point, it is an open question how much acutance/resolution that make sense, as diffraction and the need of DoF will always be the ultimate limitation.

The other side is, that until diffraction reduces acutance to zero at pixel level resolution, we can restore much of 'sharpness' using sharpening methods.

Best regards
Erik



If you have to shoot at F 5.6 to get the best results with the GFX100 it kinda limits your shooting. I never had any issues with the GFX50s at F11 and even F14. If diffraction was visible post production sharpening allowed for a good image.

Paul C
Logged
Erik Kaffehr
 

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Re: DPReview reshot the GFX 100 images with amazing results...
« Reply #5 on: July 09, 2019, 10:15:16 am »

If you have to shoot at F 5.6 to get the best results with the GFX100 it kinda limits your shooting. I never had any issues with the GFX50s at F11 and even F14. If diffraction was visible post production sharpening allowed for a good image.

Paul C
That limit to largish apertures and lowish DOF is an inherent trade-off for getting “100MP worth of detail”, regardless of camera model or format size. Yes, at 100MP and up, such trade-offs do significantly limit use cases, at least for single-shot photography (no focus stacking). The good news is than when it is better to stop down for more DOF, the sharpness is never less than with a lower resolution sensor; it is just that the resolution gain from the higher MP sensor is partially lost to diffraction.

I’m guessing that by 150 to 200MP, the extra res. will mostly be achieved with almost flat or very distant subjects: paintings, aerial mapping/surveillance and so on. Or very stationary subjects, allowing focus stacking.
Logged

Paul2660

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4066
    • Photos of Arkansas
Re: DPReview reshot the GFX 100 images with amazing results...
« Reply #6 on: July 09, 2019, 11:18:43 am »

Just not the case for me, with either 100Mp or 150MP.  Plenty of details left at F 8 and ever F11, with C1 on the Phase files, huge amounts of detail.   XF or tech lenses, however tech lenses get slightly better results.

F 5.6 for me is just not an option for MF, such a limited DOF. 

But I realize everyone's use of such systems is different. 

Paul C
Logged
Paul Caldwell
Little Rock, Arkansas U.S.
www.photosofarkansas.com

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: DPReview reshot the GFX 100 images with amazing results...
« Reply #7 on: July 09, 2019, 01:11:07 pm »

Hi,

I don't think it is an either or situation. I would recall that Jim Kasson calculated that contrast extinction appears at 2.9 micron pixels for f/11. So, what happens when we stop down beyond optimum is that we need to sharpen more. Sharpening may cause artifacts, though.

I would think it may be reasonable to choose resolution according to print size. Say that I would print at maximum 30"x40" and that I would need 180 PPI for an excellent print. So, I would need 39 million good pixels. In my case I have a 39 MP Phase One P45+ and a 42 MP Sony A7rII. Both would just fill the bill. But, rendition may be a bit coarse and it may need pretty aggressive sharpening to get there. Having more pixels may result in a smoother image, needing less aggressive sharpening.

Shooting in DoF limited settings, there would obviously be a balance, stopping down reduces the blur caused by defocusing but increases the blur because diffraction. Optimum is probably when CoC from defocus equals the full width half maximum of the diffraction blur, or something like that.

Nothing of this is new, it was well known in the eighties. But in that time we needed a microscope to see full detail in slides. Just to say, I had the opportunity to check out slides on a professional microscope at 100-400X magnification and that was an interesting experience. Now days we can just open an image in Lightroom and zoom to 1:1 view.

Best regards
Erik

That limit to largish apertures and lowish DOF is an inherent trade-off for getting “100MP worth of detail”, regardless of camera model or format size. Yes, at 100MP and up, such trade-offs do significantly limit use cases, at least for single-shot photography (no focus stacking). The good news is than when it is better to stop down for more DOF, the sharpness is never less than with a lower resolution sensor; it is just that the resolution gain from the higher MP sensor is partially lost to diffraction.

I’m guessing that by 150 to 200MP, the extra res. will mostly be achieved with almost flat or very distant subjects: paintings, aerial mapping/surveillance and so on. Or very stationary subjects, allowing focus stacking.
Logged
Erik Kaffehr
 

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Re: DPReview reshot the GFX 100 images with amazing results...
« Reply #8 on: July 09, 2019, 01:54:27 pm »

Just not the case for me, with either 100Mp or 150MP.  Plenty of details left at F 8 and ever F11, with C1 on the Phase files, huge amounts of detail.   XF or tech lenses, however tech lenses get slightly better results.

F 5.6 for me is just not an option for MF, such a limited DOF. 

But I realize everyone's use of such systems is different. 

Paul C
Paul, I’m not sure where that specific f/5.6 limit for 100MP in 44x33 comes from (some set stricter standards for diffraction avoidance than others!) so I’ll just say that the doubling pixel count at worst moves the acceptable diffraction threshold one stop. So if f/11 to f/14 works for you with a 50MP 44x33 sensor, then f/8 to f/10 will have equally acceptable diffraction effects with the 100MP 44x33 sensor: everything scales proportional to or reciprocal to linear pixel size and such, not pixel count.

Aside: this is one reason why I prefer the video world’s descriptions, using horizontal pixel count: the new 44x33 sensor is about 11,500 pixels wide, or “11.5K”, vs “8.2K” for 50MP. Not so big a resolution jump when you look at it that way.
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: DPReview reshot the GFX 100 images with amazing results...
« Reply #9 on: July 09, 2019, 07:27:08 pm »

Hi BJL,

No, f/5.6 is no magic limit. It simple is the aperture that achieves maximum sharpness at best focus. Stopping down beyond means that the image may need a bit more sharpening.

Best regards
Erik

Paul, I’m not sure where that specific f/5.6 limit for 100MP in 44x33 comes from (some set stricter standards for diffraction avoidance than others!) so I’ll just say that the doubling pixel count at worst moves the acceptable diffraction threshold one stop. So if f/11 to f/14 works for you with a 50MP 44x33 sensor, then f/8 to f/10 will have equally acceptable diffraction effects with the 100MP 44x33 sensor: everything scales proportional to or reciprocal to linear pixel size and such, not pixel count.

Aside: this is one reason why I prefer the video world’s descriptions, using horizontal pixel count: the new 44x33 sensor is about 11,500 pixels wide, or “11.5K”, vs “8.2K” for 50MP. Not so big a resolution jump when you look at it that way.
Logged
Erik Kaffehr
 

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Re: DPReview reshot the GFX 100 images with amazing results...
« Reply #10 on: July 10, 2019, 05:44:04 am »

Hi BJL,

No, f/5.6 is no magic limit. It simple is the aperture that achieves maximum sharpness at best focus. Stopping down beyond means that the image may need a bit more sharpening.

Best regards
Erik
Yes, I’m sure it is a useful guideline by some criteria; my main point is that comparisons between different pixel counts need to be done with consistent criteria and measurement methods, and then I expect the shift to be about one stop per doubling of pixel count (and of print size in area?). So the criteria that give f/5.6 for the new 100MP sensor also give f/8 for the 50MP—and thus are about a stop stricter than however Paul C is evaluating things.
Logged

SrMi

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 298
Re: DPReview reshot the GFX 100 images with amazing results...
« Reply #11 on: July 10, 2019, 02:56:08 pm »

Yes, I’m sure it is a useful guideline by some criteria; my main point is that comparisons between different pixel counts need to be done with consistent criteria and measurement methods, and then I expect the shift to be about one stop per doubling of pixel count (and of print size in area?). So the criteria that give f/5.6 for the new 100MP sensor also give f/8 for the 50MP—and thus are about a stop stricter than however Paul C is evaluating things.

On DPR site, it seems that GFX100 studio raw file was rendered with C1, while GFX50S was rendered with ACR. Does anyone know which sharpening values DPR used for the raw files? C1 is known to be a bit more aggressive with default sharpening, which I believe led Adobe to increase the default sharpening amount in LR as well. I do not think that DPR has an easy job with studio scenes. Best is to download the raw files yourself and compare them (assuming that they have been properly focused).
Logged
Pages: [1]   Go Up