p.7 #1 · Sony A7R sensor gets high scores from DxOMark
theSuede wrote:
The word "accuracy" is the most simple and easy way to define color. IMO all other ways are perceptual, and we all have different perceptions. Often what we perceive as "good" is quite inaccurate.
You've been reading perceptual psychology, haven't you?
You should talk with some of my colleagues at TNO Human Factors, NL (or whatever they've been recently renamed to).
p.7 #2 · Sony A7R sensor gets high scores from DxOMark
It sure will be interesting to see just how far the love goes for the 7 and 7R. Certainly their sensor performance won't be their undoing. For a traveling landscape enthusiast it could be the ultimate too.
p.7 #3 · Sony A7R sensor gets high scores from DxOMark
theSuede wrote:
They're identical cyans.
+1
0,255,197
That's why I almost always check rgb numbers @ known/desired neutrals rather than trust my own perception alone. Our eyes/brain can fool us (accommodation), imo most particularly @ cast detection.
p.7 #7 · Sony A7R sensor gets high scores from DxOMark
Sony's blurb:
'They feature Sony’s new BIONZ X image processing engine with Triluminos 14-bit RAW image for the most accurate and lifelike color ever.'
Brian Smith:
'The A7R features the most lifelike colors I’ve ever seen from a digital camera...The dynamic range is really incredible holding detail from highlights to the deepest shadows. Sony calls this as ‘Detail Reproduction Technology’ and it features a more subtle and sophisticated sharpening system with less apparent emphasis on edges, giving a more convincing representation of fine detail.'
I have a feeling Sony figured the sensor was 'good enough' in general performance measures (it is) and so went after these less tangible aspects of image production. For the RX1 they say they went to some trouble to have shadows and highlights look more natural to the human eye.
p.7 #8 · Sony A7R sensor gets high scores from DxOMark
Worldinlens wrote:
Never seen.. Seems here some effect like to rgb pixels on the display. Cyan mixed with yellow and purple. When we
greatly increase picture we see all colors separately.
Well, it isn't... The entire effect is in your eyes (and brain) unless your monitor (laptop?) is set to some interpolation resolution. When set at the native resolution, all pixel values are perfectly identical, there's no color difference between the two stripe areas at all. They just differ in local surround, at a spatial frequency where your vision system is most sensitive to contrast (if you're looking at it on a "normal" screen at a "normal" viewing distance)
RustyBug wrote:
+1
0,255,197
That's why I almost always check rgb numbers @ known/desired neutrals rather than trust my own perception alone. Our eyes/brain can fool us (accommodation), imo most particularly @ cast detection.
Yes, very good. All too many people "just trust" their eyes - which is perfectly fine if you're working on a controlled environment setup - monitor in a low-normal lit room to D50, no strong colors except the monitor within the peripheral view when you sit in front of the monitor.
And even then you get fooled sometimes, especially if you zoom in on an image so that it covers the entire screen surface. You adapt to local gray balance very quickly, so if there's strong colors in the image in the zoomed in area, your neutral reference gets kicked way off.
I could probably do a portrait session PP on a grayscale monochrome monitor, just working by numbers. But I realize that's kind of a not-normal mindset, coming from working with software development in visual perception for to long...
p.7 #9 · Sony A7R sensor gets high scores from DxOMark
freaklikeme wrote:
Possible.
Tariq, did you take the test on a four year old Macbook Pro with anti-glare screen that you haven't calibrated in a year or so?
It's actually much more accurate (and fair to your eyes...) to do that test with the monitor profile/calibration turned off. It's not about color accuracy, it's about hue resolution - which isn't the same thing at all.
On the old MBP's I wouldn't be surprised at all if it's the monitor profile that's disrupting the green hue resolution for some people, by affecting the monitor curves.
When there's only 8 (6+2 or 7+1) bits of tone resolution possible on the hardware (actual LCD panel control), every single minimal correction the profile has to make removes at least one bit-step from the 255 original bit-steps. A profile working hard to correct something can leave some 230 effective bit-steps per primary, taking away almost 10% of tone resolution per channel.
p.7 #10 · Sony A7R sensor gets high scores from DxOMark
When it comes to noise at higher ISO speeds, it is actually visible already at 400, and comparatively clear of 800. Compared to eg Nikon D800 so the noise is more noticeable, but on the other hand, the images at high ISO much sharper with A7R, suggesting that D800 reduces the noise harder.
Since there is a similar 36Mp sensor in A7R as the Nikon D800/D800E, so we were a little curious to see if there was any difference in exposure, colors, etc. We stowed therefore also a D800 and took identical pictures at the same time, and as expected, they gave slightly different results. What differed most was how they perceived the colors and shades, ie white balance. Exposure was not exactly the same every time, A7R was not to be fooled as often as D800 of dark or light element in the scene. However, on the whole relatively small differences. But then, it was after all this with clarity and detail reproduction, and which looked better in all the pictures taken with A7R. The biggest difference was noticeable on small aperture around f/16, probably thanks to Sony's built-in reduction of diffraction. But the lens was not the same either, and obviously have some impact.
p.7 #12 · Sony A7R sensor gets high scores from DxOMark
freaklikeme wrote:
Tariq, did you take the test on a four year old Macbook Pro with anti-glare screen that you haven't calibrated in a year or so?
Ha Ha. I have not taken the test yet but if I do, it would be on my wide-gamut, 10-bit calibrated studio monitor and not on my recently calibrated Macbook Pro Retina screen (which still only has an SRGB gamut).
I have taken these sort of tests before and do pretty well but my wife always does better. I believe women in general are more sensitive to color than men. Anytime I'm doing color critical reproduction work for clients I always get my wife's opinion.
Edit. I thought I had read that women were more sensitive to color:
"Guys' eyes are more sensitive to small details and moving objects, while women are more perceptive to color changes, according to a new vision study that suggests men and women actually do see things differently."
"In one part of the study, the researchers asked the volunteers to describe different colors shown to them. They found that the guys required a slightly longer wavelength of a color to experience the same shade as women and the men were less able to tell the difference between hues."
p.7 #13 · Sony A7R sensor gets high scores from DxOMark
All makes good sense, headroom, and backs up what pre-release users are saying.
Noise is almost a misnomer in the case of recent Sony sensors, the quality of it even affects bokeh in unexpected (and pleasant) ways. It often pains me to see high ISO C/N images, a diametrically opposed approach: 'see it - smooth it'.
Look forward to hearing more about diffraction reduction too - some kind of deconvolution maybe? Interesting developments for the near future.
These are the kind of results that are not going to be reflected in DxO data, perhaps, but which the end user will like and see as a benefit.
p.7 #14 · Sony A7R sensor gets high scores from DxOMark
Tariq Gibran wrote:
Ha Ha. I have not taken the test yet but if I do, it would be on my wide-gamut, 10-bit calibrated studio monitor and not on my recently calibrated Macbook Pro Retina screen (which still only has an SRGB gamut).
Oh, oops. That should've been directed at Jordan. Sorry, Tariq.
p.7 #15 · Sony A7R sensor gets high scores from DxOMark
Having better color discrimination is a statistic deviation, not something you can say is a sliding range for "a woman" or "a man".
Most women have the exact same color sensitivity as most men. Quite a few women have better color discrimination than most men. Very few men have better color discrimination than most women...
-And also, studies that are based on quantitative research (letting a lot of people "take a test") fail to discriminate for the quite high percentage of men that have some sort of color blindness - integrating them into research results will say that "women have slightly better color discrimination" - which is an erroneous conclusion *IMO.
-And AFAIK, only women can have functional tetrachromacy, though some research seems to point to a low percentage of men having it too. The most common tetrachromacy is a split L group, increasing hue discrimination all the way from green to deep red. How many (in % population) this group contains is quite unclear.
p.7 #16 · Sony A7R sensor gets high scores from DxOMark
theSuede wrote:
Having better color discrimination is a statistic deviation, not something you can say is a sliding range for "a woman" or "a man".
Most women have the exact same color sensitivity as most men. Quite a few women have better color discrimination than most men. Very few men have better color discrimination than most women...
-And also, studies that are based on quantitative research (letting a lot of people "take a test") fail to discriminate for the quite high percentage of men that have some sort of color blindness - integrating them into research results will say that "women have slightly better color discrimination" - which is an erroneous conclusion *IMO.
-And AFAIK, only women can have functional tetrachromacy, though some research seems to point to a low percentage of men having it too. The most common tetrachromacy is a split L group, increasing hue discrimination all the way from green to deep red. How many (in % population) this group contains is quite unclear....Show more →
Interesting. Per the article I referenced, the men and women tested were indeed screened for normal color vision and 20/20 sight. So, color blind men did not taint the results.
"Abramov and his team from CUNY's Brooklyn and Hunter Colleges compared the vision of males and females over age 16 who had normal color vision and 20/20 sight — or at least 20/20 vision with glasses or contacts."
p.7 #17 · Sony A7R sensor gets high scores from DxOMark
Jman13 wrote:
I also scored 4. All of my misses were in the green area.
I tried that a couple years ago and got a 0.
I suspect that how your monitor is set up can also come into play, especially if you have it externally calibrated and profile and lots of correction had to be done within 8 bits.
An interesting side tangent is that a probe can record the same colors on two different monitors but they can look like different colors to a person. All consumer color management uses a simple tri-stim system and yet the primaries in different monitors, the largest difference is when you switch between different backlight types, each send out different spectral spikes, only a super finely grained color management system that measured spectrums could adjust things to look the same across monitors.
Every notice how some colors like quite different when shown side by side on a wide gamut CCFL and a regular CCFL LCD? That's a failure of the simple color management system used in all commercial products. Some monitors, fancier ones, have a metamerism toggle to try to make the wide gamut look to the eye like a regular gamut in terms of how the eye sees colors, but there is huge variance person to person in how the spectral spikes in the monitor primaries affect your vision so it doesn't necessarily work out so well. I know the one in my NEC PA241W while it makes some colors, side by side, look closer to a regular gamut monitor, it actually makes others look even more different to me. At least in my case, if I compare a color checker on a typical wide gamut LCD and on a standard gamut LCD and then to a real life chart viewed under as close to D65 conditions as I could get, the real life chart looked more like what I saw on the wide gamut monitor.
p.7 #18 · Sony A7R sensor gets high scores from DxOMark
RustyBug wrote:
Agreed ... except that 30 seconds later as cloud formations change, or 10 minutes later (in the golden hours especially) as the sun position changes or when you move your camera/subject/lighting orientation, it is no longer valid to the same degree of accuracy. If I can get my neutrals, neutral (when that's my goal), the difference between a full blown color checker vs. white/gray/black reference likely falls within the limits of personal preference anyway.
If your lighting and orientation are stable, then a color checker holds more validity as a reference. But, if you are "on the move" ... or your lighting is ... it's accuracy (as a one shot reference) may / may not hold to the degree it technically offers (i.e. micrometer vs. yardstick) as it does in the studio where your lighting is stable. Thus, you'd need to reshoot a new reference more than just the once to ensure its continued accuracy relevance....Show more →
there are other issues to though in that some camers might be able to distinguish between larger numbers of subtle shades
p.7 #19 · Sony A7R sensor gets high scores from DxOMark
Might be a reasonable thread to post some comments on a claim made By Steve Huff:
claim:
'The low light capabilities of this (DF) camera will be about the best you can get in full frame.'
I chose to compare the same sensored D4 with another new camera that has no right be good at high ISO, and about which few will say those words - the a7r.
Comments based on DxO results:
. the a7r posts a higher true ISO and higher signal-to-noise ratio (SNR) at the makers' ISO rating of 25600;
. the a7r has a DR value at this ISO 25600 level within 0.3 EV of the D4 (DF);
. the a7r has a higher tonal range than the D4 (DF) at ISO 25600 than the DF;
. colour sensitivity at true ISO is a difference of ~0.3 bits with the D5 (DF) a nose in front (14.7-14.4).
Yes, it (the DF) is indeed 'about the best you can get'...but so is the a7r, which adds 36Mp vs 16Mp; and at least matches the D4 (DF) at base ISO for all measures bar DR in which it has a full EV advantage.
The D4 (DF) does extend low ISO into the higher reaches - up to 204800 in fact, which looks like a good win, but all measures lie well into the 'red zone' by a 'true' 25600, with a commensurate dramatic loss of image quality.
Seems here some effect like to rgb pixels on the display. Cyan mixed with yellow and purple. When we