p.45 #1 · Pre-orders open! Sony A7R III and FE 24-105mm f/4 OSS lens!
bjornthun wrote:
Will you also do meassurements for the new pixel shift mode in the A7R3?
I certain hope to.
Have you seen this recent work regarding Olympus High Res Shot Mode?
p.45 #2 · Pre-orders open! Sony A7R III and FE 24-105mm f/4 OSS lens!
alundeb wrote:
It is reflected in the DxOMark curve. There is a bump at ISO400, but since they don't measure at intermediate ISO, we can't see that it is lower at ISO320. The next result is at 200, that has higher DR again than 400, but not a full stop better, just as your data.
To my eye the DxOMark curve is too smooth. There really should be two "parallel" sections.
FWIW, Read Noise figures derived from DxOMark data (as pioneered by sensorgen) also show no bump.
Contrast
www.photonstophotos.net/Charts/ReadNoise_ADU.htm#Nikon%20D850_14
(The open circles are the DxOMark figures)
with
www.photonstophotos.net/Charts/RN_ADU.htm#Nikon%20D850_14
Oct 29, 2017 at 11:25 AM
Steve Spencer Offline • • • • • • • Upload & Sell: On
p.45 #3 · Pre-orders open! Sony A7R III and FE 24-105mm f/4 OSS lens!
bclaff_too wrote:
I wanted to add separately that if you look closely at the DxOMark graphs there is some reason to doubt their measurements for the D850 (or at least to suspect they have been smoothed for reporting).
Specifically look at the Landscape Print curve and compare it with PhotonsToPhotos; see
www.photonstophotos.net/Charts/PDR.htm#Hasselblad%20X1D-50c,Nikon%20D850
Note that PDR goes up at ISO 400 because this is where the Nikon D850 invokes high conversion gain.
This correct behavior is not reflected in the DxOMark curve.
Bill, while you are hanging around I hope I can ask you a question. How are the PDR scores affected by manufacturers cheating on how they label ISO values, for example saying an ISO is 200 when it is actually 100?
p.45 #4 · Pre-orders open! Sony A7R III and FE 24-105mm f/4 OSS lens!
Steve Spencer wrote:
Bill, while you are hanging around I hope I can ask you a question. How are the PDR scores affected by manufacturers cheating on how they label ISO values, for example saying an ISO is 200 when it is actually 100?
That is a deceptively complex topic.
I don't know of any manufacturer that "cheats" on their ISO ratings.
ISO ratings are based on JPEG output and of the two different ISO standards one is "objective" (SOS) and the other "subjective" (REI).
REI is a very loose definition.
However, even the SOS rating does not relate back to raw data in a meaningful way.
So, when DxOMark measures ISO by yet another standard (SSAT on the raw data) this causes a lot of confusion.
DxOMark Measured ISO is useful data but I disagree on using it as the x-axis in presenting data.
So yes; DxOMark uses Measured ISO as their x-axis whereas PhotonsToPhotos used the ISO setting.
Some people are quite passionate about which approach is best.
In addition to the concerns outlines above there are other reasons I use my approach.
Oct 29, 2017 at 11:39 AM
Steve Spencer Offline • • • • • • • Upload & Sell: On
p.45 #5 · Pre-orders open! Sony A7R III and FE 24-105mm f/4 OSS lens!
bclaff_too wrote:
That is a deceptively complex topic.
I don't know of any manufacturer that "cheats" on their ISO ratings.
ISO ratings are based on JPEG output and of the two different ISO standards one is "objective" (SOS) and the other "subjective" (REI).
REI is a very loose definition.
However, even the SOS rating does not relate back to raw data in a meaningful way.
So, when DxOMark measures ISO by yet another standard (SSAT on the raw data) this causes a lot of confusion.
DxOMark Measured ISO is useful data but I disagree on using it as the x-axis in presenting data.
So yes; DxOMark uses Measured ISO as their x-axis whereas PhotonsToPhotos used the ISO setting.
Some people are quite passionate about which approach is best.
In addition to the concerns outlines above there are other reasons I use my approach. ...Show more →
Very interesting, so DXO uses SSAT on the raw data and then they use that measurement for the X-axis in their graphs. I think that explains a clear discrepancy between your PDR scores and DXO in rating the OMD EM-1 MkI and MKI. DXO has the EM-1 Mk II having labelled ISO very different from their SSAT measure. For example where Olympus labels ISO 800, their SSAT measure suggest ISO 352, so over a stop difference. The EM-Mk I shows much less of a discrepancy between labelled ISO and DXO's SSAT measure, so where Olympus labels ISO 800 DXO measures 487. Then when DXO plots their SSAT on the x-axis the two cameras having almost identical dynamic range. This is actually quite similar to the PDR graphs that show the MK-II and having higher PDR at almost every labelled ISO. If the DXO graph used labelled ISO on the x-axis their data would looks quite similar to the photons-for-photos graph. Do you have any comments on which in this particular case which you think better reflects the capability of the camera? Do you think Olympus was able to markedly improve the dynamic range of the camera from MK-1 to Mk-II or could this apparent increase be at least partially relabelling of the ISO values?
p.45 #6 · Pre-orders open! Sony A7R III and FE 24-105mm f/4 OSS lens!
Very interesting. Panasonic are going to release a pixel shift mode as well on their next G9 camera too (see my post screen in the 4/3 page on Ken Duncan leaked info).
bclaff_too wrote:
I certain hope to.
Have you seen this recent work regarding Olympus High Res Shot Mode?
p.45 #7 · Pre-orders open! Sony A7R III and FE 24-105mm f/4 OSS lens!
Luvwine wrote:
Guess that depends on whether pixel shift is a good feature ...
Based on posted images from the Pentax K-1, I consider it a good feature but quite specialized. Anything that moves between frames will cause a serious chromatic cockup. For studio tabletop and product (my main activity), the feature will eliminate color aliasing larger than one pixel and the aliasing will be color-neutral when the image is scaled down. This is a big deal with any detailed subject, and not otherwise correctable in post. Moire can creep into a shot at any time, but is worst with fabric. My current solution is shooting beyond f/16, which is generally fine for the DOF but taxing for the strobes. Pixel shifting really needs hardware at the sensor plane—no other way to fake it.
I have skipped the "R" series so far, but this one looks good for pixel shift and big battery alone.
Very noticeable improvement Bill. Thanks for the link.
Would you say that any RAW file could benefit from a simple image averaging (mean stack)?
I see a noticeable improvement even with only 2 images. I'm curious about the difference between pixel averaged and pixel shifted in PDR. (4 images each)
Thanks for pointing me to it. I had not seen, so this is interesting. Olympus gets an almost two stop increase in PDR. I had mostly thought of the high res mode as a way of gaining higher resolution and avoiding Bayer artifacts. It will be interesting to see the reults for Sony.
p.45 #10 · Pre-orders open! Sony A7R III and FE 24-105mm f/4 OSS lens!
bclaff_too wrote:
To my eye the DxOMark curve is too smooth. There really should be two "parallel" sections.
The DxO curve sections are no less "parallell" than yours, if you extend more than one ISO stop. Your values jump a lot. I don't think this means anything outside margins of error and X axis resolution.
p.45 #11 · Pre-orders open! Sony A7R III and FE 24-105mm f/4 OSS lens!
Steve Spencer wrote:
Very interesting, so DXO uses SSAT on the raw data and then they use that measurement for the X-axis in their graphs. I think that explains a clear discrepancy between your PDR scores and DXO in rating the OMD EM-1 MkI and MKI. DXO has the EM-1 Mk II having labelled ISO very different from their SSAT measure. For example where Olympus labels ISO 800, their SSAT measure suggest ISO 352, so over a stop difference. The EM-Mk I shows much less of a discrepancy between labelled ISO and DXO's SSAT measure, so where Olympus labels ISO 800 DXO measures 487. Then when DXO plots their SSAT on the x-axis the two cameras having almost identical dynamic range. This is actually quite similar to the PDR graphs that show the MK-II and having higher PDR at almost every labelled ISO. If the DXO graph used labelled ISO on the x-axis their data would looks quite similar to the photons-for-photos graph. Do you have any comments on which in this particular case which you think better reflects the capability of the camera? Do you think Olympus was able to markedly improve the dynamic range of the camera from MK-1 to Mk-II or could this apparent increase be at least partially relabelling of the ISO values?...Show more →
Sorry for the long quote .. hard to know what to clip
I understand the appeal of using Measured ISO on the x-axis when comparing cameras, particularly from different manufacturers using different ISO rating methodologies.
However, this approach is fatally flawed !
I have found a very wide variation in metering between cameras and, more importantly, within the same camera.
Metering modes, metering constants, overall scene brightness, etc. have a huge effect on the exposure chosen by any camera.
This swamps the effect of Measured versus Stated ISO (yes even a stop difference).
So, the best we can assume in comparing cameras is that we will "nail" the exposure of each camera at the chosen ISO setting, even if that means a different exposure between cameras
There are even more reasons that I use ISO setting rather than Measured ISO; but I've already presented two good ones.
p.45 #12 · Pre-orders open! Sony A7R III and FE 24-105mm f/4 OSS lens!
Fred Miranda wrote:
Very noticeable improvement Bill. Thanks for the link.
Would you say that any RAW file could benefit from a simple image averaging (mean stack)?
I see a noticeable improvement even with only 2 images. I'm curious about the difference between pixel averaged and pixel shifted in PDR. (4 images each)
Sure, stacking is always helpful because it improves the Signal to Noise Ratio (SNR).
But .. the devil is in the details, isn't it?
To get the benefit of stacking you have to (effectively) lengthen the exposure time.
Assuming you're already shooting at an exposure time necessitated to freeze motion you can't do that.
(Shooting faster and stacking to the same total exposure is not a net gain.)
This is probably why widespread use of stacking is mostly confined to astrophotography.
Where with proper tracking you have a subject that doesn't move (relative to the camera) over time.
For terrestrial photography, things are tougher.
Pixel shift as opposed to simple stacking has different benefits.
I suspect it's worthwhile to pixel shift (at faster exposures times) to gain benefits other than simply improved SNR (far less likely for moire, better color fidelity, etc.)
p.45 #13 · Pre-orders open! Sony A7R III and FE 24-105mm f/4 OSS lens!
bclaff_too wrote:
Sure, stacking is always helpful because it improves the Signal to Noise Ratio (SNR).
But .. the devil is in the details, isn't it?
To get the benefit of stacking you have to (effectively) lengthen the exposure time.
Assuming you're already shooting at an exposure time necessitated to freeze motion you can't do that.
(Shooting faster and stacking to the same total exposure is not a net gain.)
This is probably why widespread use of stacking is mostly confined to astrophotography.
Where with proper tracking you have a subject that doesn't move (relative to the camera) over time.
For terrestrial photography, things are tougher.
Pixel shift as opposed to simple stacking has different benefits.
I suspect it's worthwhile to pixel shift (at faster exposures times) to gain benefits other than simply improved SNR (far less likely for moire, better color fidelity, etc.)
p.45 #15 · Pre-orders open! Sony A7R III and FE 24-105mm f/4 OSS lens!
AGeoJO wrote:
If you use a native lens, yes, it can not only follow the face but also the eye, which is more critical than the face. You can get the Eye-AF also using Canon lenses with Metabones IV or V or similar but the performance is not as fluid as a native lens. The adapter, Metabones, in this case is the bottleneck in the AF process but they are updating the FW of their adapter on a regular basis.
Taking one step further on the A9, Leica-M lenses, which are 100% manual focus lenses, can be used on an AF adapter and you can even get Eye-AF with that camera. There is at least one adapter that is readily available from TechArt, called TechArt Pro. And Fotodiox has announced a similar adapter but I am not familiar with that. I assume that the A7r III has a similar AF system as the A9 but how it performs remains to be seen.
Remove/control shooting via bluetooth on your phone can be done with through an application offered by Sony Playmemories. It works with all A7 series camera up to this point. The status of the apps offered through Playmemories is questionable at this point, but the way we understand it at this point, the apps are not available for the A7r III and A9. Maybe some kind of replacement is at work. Who knows....Show more →
Thanks so much for the info! Just to be clear on my question - you mention the eye, rather than the face - I'm talking about servo autofocusing in VIDEO. It can actually track the eyes in video?! That would be extraordinary!
You might find it useless outdoors or when camera handheld. It is very interesting but so limited, otherwise Olympus would have got a lot attention.
I once had the Olympus OM-D E-M5 mkII, so I'm well aware of what it can actually be used for. When it works, it's great. The slightest of wind blowing outdoors, and you're out of luck.
p.45 #17 · Pre-orders open! Sony A7R III and FE 24-105mm f/4 OSS lens!
Any projection on tethered shooting speeds with the usb upgrade? The A7R II left a lot to be desired in my limited time with one... (maybe it was just my adjustment from a 5D III file size, or maybe the improved USB will actually speed up the file transfer)
Also, anyone know if tether tools already has a cable that fits, or if we have to wait for them to create one in the next couple months
p.45 #18 · Pre-orders open! Sony A7R III and FE 24-105mm f/4 OSS lens!
avuroski wrote:
Thanks so much for the info! Just to be clear on my question - you mention the eye, rather than the face - I'm talking about servo autofocusing in VIDEO. It can actually track the eyes in video?! That would be extraordinary!
I have never shoot video but sony AF configuration setup should be the same for both stills and video.
p.45 #19 · Pre-orders open! Sony A7R III and FE 24-105mm f/4 OSS lens!
Fred Miranda wrote:
Yes, with the A7RII the mag. is 12.5x but I don't recall the A7R maximum magnification.
I remember when first getting my A7RII that the viewfinder image IQ seemed to be a downgrade form the A7R's. There was a question of standard or high display quality but it really didn't make much of a difference.
Man, I hope the RIII a good step up from the RII. It's one of the most critical interfaces in the camera, and I felt the same way as you, the RII was a downgrade from the OG R.
p.45 #20 · Pre-orders open! Sony A7R III and FE 24-105mm f/4 OSS lens!
avuroski wrote:
Thanks so much for the info! Just to be clear on my question - you mention the eye, rather than the face - I'm talking about servo autofocusing in VIDEO. It can actually track the eyes in video?! That would be extraordinary!
In theory, it should be able to, but in practice, there may be limitations. A review that does a comparison with a Canon camera equipped with their DPAF technology (the current benchmark video servo AF) would certainly be enlightening!
[lots of things could become problematic; for the Sony system, the particular lens used is likely to make a larger difference, as video servo needs both speed and fine adjustment capability out of lens AF; further, while Sony has very good PDAF coverage and great readout speeds, there are still AF coverage gaps]
