igor_s: Looks like a perfect camera in its price range. But I noticed one thing in its AF test video: the shots were taken after a small delay during which the model remained steady. How does the AF system cope with truly moving objects?
Rishi, I am not sure about what you are speaking. At the Page 8 there is an AF test with the bearded man as model. I could not find anything below it.And again, what about the unification of the AF tests?Hope also you (DPR) will do the IS tests for all cameras equipped with the IBIS.
May I suggest the DPReview team to do some unified AF tests. Standard target (1 or 2 if you want face/eye detection), standard directions (2), standard distance range (1) and standadrd speeds (until the camera can not cope). That would be not too much work in fact. Why should the readers rely on the tests that vary every time?
How well the AF system can keep on the desired object among other objects is another thing.
Looks like a perfect camera in its price range. But I noticed one thing in its AF test video: the shots were taken after a small delay during which the model remained steady. How does the AF system cope with truly moving objects?
lacikuss: Okay, so according to the DXO marks the G7x Sensor outshines the SONY RX100III sensor in everything: Color depth, DR and High ISO performance.
Quote: "Meanwhile, the details of the sensor strongly suggest it uses Sony's IMX183CQJ sensor, so could well be able to match the latest RX100 series in terms of image quality"
My question is simple, aren't these two cameras supposed to have the same sensor? What is the meaning of "Strongly suggest"? Isn't this another bias preview? or Is it DXO marks with the bias? Or maybe this is due to simple sample variations of the same product?
When DPreview, DXO marks, CameraLabs or any other site base their product reviews on sample size = 1, What is the Statistical significance of any of these reviews?
I think it is just the sensor quality variation. Before testing the ISO/noise, I believe that DxOMark test the lens and knows its actual F-stop values. When testing the sensor they probably use its central part where CAs are negligible. As to the DR, how could it be affected by CAs?
Canon admitted that they use some Sony smaller than APS-C sensors.
@lacikuss: 6.3%/3=2.1%, not 6 (however I am not sure why divide the average difference by 3).Anyway I think that even 6% is not a statistically valid difference for 1 sample. Moreover, DxOMark say that they saw the difference (albeit slight) only in DR whereas your calculations show the biggest difference in ISO....
From DxOMark's review of the GX7:
"The third-generation RX100 III has an almost identical performance as the RX100 II. The differences in Color Depth, DR and ISO are negligible (which accounts for the same overall DxO Mark score) and so it’s safe to assume the two Sony models adopt the same sensor. As the RX100 III is the latest model from the firm, the result tends to support the notion that the sensor is the same, only Canon scores a very slightly better dynamic range."
One thing left uncovered in the article - about the "reach". Many people think that a smaller sensor automatically gives a longer reach with the same lens focal length ("through equivalent focal length"). In reality, there matters not the sensor size but the pixel pitch. Sure, usually the smaller the sensor the smaller the pixels are, but this relation is not straightforward.
RFC1925: I would summarise the article regarding the difference between MFT and FF like this:
If you want:1) Shallow DOF = FF is better2) Deep DOF = MFT and FF are tied (FF can compensate with better ISO performance)3) Portability = MFT is better
All the above have exceptions of course since there are such a great variety of cameras and lenses in both formats.
2) FF can not always "compensate" because on any format the resolution drops from about F/11 due to diffraction. In comparatively good light you can shoot at F/11 on FF as well as on MFT where the DOF will be deeper.
igor_s: Simulation of higher ISOs by brightening in the post decreases the SNR, therefore, the comparison is not fair. However, the 7S has lower level of shadow noise, and therefore should do better at extra high ISOs (where the sensor is poorly illuminated). Perhaps natively by about 1 eV if in "boosted" ISO tests it wins by 2 eV.Unfortunately the above applies only to ISOs above 51200 (if you ever need it and satisfied by the quality). At lower and moderate ISOs the A7S loses to the a7R even in shadows. A specific ultra-low light camera.
Rishi, to compare the normalized Full SNR curves for the 5DIII and A7S, you may simply rise the first one by 3 dB (24 Mp vs 12 Mp). In that case, it will cross the 400K curve for the 7S near the X axis. So, the a7S does have a 2 stops advantage over the 5DIII, but only where the SNR is close to zero(g).
mosc: A lot of comments read like: Yes A7S makes sense at 6400+ ISO but who shoots that? The other cameras beat it soundly at base ISO.
But similarly, who needs all the detail of a FF camera at base ISO? You put it on a monitor, even a 4k monitor, that's a lot of crop. You print it out, that's a lot of wall space. The A7S is fractionally less wall space and fractionally less cropping. That's not much of a sacrifice for the added sensitivity. If I have all the light in the world and reach for an A7S instead of an A7R, how much practical difference does it make? I say little to none. In bad light, it makes a big difference. You guys have it backwards.
Not simply at ISO above 6400 but ONLY in shadows. A device for detecting a black cat in a dark room.
@Rishi, thank you for the pointing out the specific behaviour of Canon DSLRs. As to the DxOMark data, either I misunderstand something or the data is presented incorrectly. For the 5DIII on the SNR 18% graph at ISO 100K SNR=15.2 dB. However, on the Full SNR graph for ISO 100K at 18.166% gray SNR=10.8 dB only, which makes about 1.5 stops difference. Which one should we believe/consider?In any case, I agree that in dark tones the 7S performs considerably better than the competition. But it seems that that is not supported by the similar performance in mid- and bright tones. So, with the A7s you can see more details specifically in deep shadows AND at very highest ISOs. Is it enough to speak loudly about the advantage of the 7S? (I mean in general, not your article).
Also, you can check the DxOMark data. Up to ISO 25600, the normalized SNR for the 7S and 7R are equal and the lines are parallel. I do not see any particular reason why at ISO 400K there would appear such a huge difference as 2 eV.With 5DIII, there is less than 1 eV lead by the 7S at ISO 100K.
Rishi, I won't go into the theoretical dephths as to why brightening an underexposed image results in a lower SNR compared to increasing the ISO when shooting. You can do a simple experiment yourself. Shoot a normally exposed image, say, at ISO 1600, then an underexposed image at ISO 400, brighten it and compare. I did it with my 600D, and the difference is huge. As to the digital in-camera boost at high ISO settings that TrojMacReady pointed out, I am not sure. Theoretically, it could cancel my reservations for that ISO range if it is true.
Simulation of higher ISOs by brightening in the post decreases the SNR, therefore, the comparison is not fair. However, the 7S has lower level of shadow noise, and therefore should do better at extra high ISOs (where the sensor is poorly illuminated). Perhaps natively by about 1 eV if in "boosted" ISO tests it wins by 2 eV.Unfortunately the above applies only to ISOs above 51200 (if you ever need it and satisfied by the quality). At lower and moderate ISOs the A7S loses to the a7R even in shadows. A specific ultra-low light camera.
So far the discussion has been purely qualitative. We know only that the field curvature will depend on the lens design, FL and aperture. So, is it possible to create a not-too-expensive family of lenses (including zooms) with roughly constant FC? If that is not feasible and the FC in the family would range, say, from 1 to 9 arbitrary units, how much improvement we would see with the optimized (5 units) sensor compared to the flat one?
Low contrast at f/1.4 (not uncommon, though).
igor_s: Is the A77II's IBIS configurable for work with non-chipped lenses of different FL?
What adapter model did you use? I mean, it should be programmable for FL.
In Pentax SLRs you can set the lens FL via the menu. I assume the camera will apply a different amount of sensor shift depending on your setting. You can play with it to achieve the best results with every lens. In any case that is better than nothing. The A77 will lock at default 50 mm which makes its IBIS almost useless already for a 100 mm lens.
Is the A77II's IBIS configurable for work with non-chipped lenses of different FL?
igor_s: I can not figure out why Sony refuses to add the on-chip PDAF feature to its A-mount models. For now, it seems the only way to make the AF ultimately correct while enough fast (like on the A6000). Now I see again that cr----- AF microadjust which means that the user won't be able to shoot reliably with fast lenses. If one needs 12 fps with tracking he can use the SLT system. But in a different situatuion he may be interested in the ultimate reliability of the AF system. Will see how fast the A77II's conventional CDAF system is.
For me, the only question is, whether the on-chip PDAF would be faster than the CDAF (btw, does the A77II really have at least CDAF?)
If it would, Sony should have added it. I do not care about the SLT in this respect.