Was kind of surprised by the RAW performance of the NEX-7 compared to the GX1 at a relatively mild ISO these days of ISO 1600. Thought it would be the other way around, given the large and modern sensor of the Sony.
Resampling manipulation does not allow you to see how a sensor actually renders image details at its native resolution, something I'm more interested in comparing.
Never use Dpreviews samples to demonstrate anything.
There are wild variations across the frame due to lighting, lens issues and most frequently focus differences and depth of field. Even with the same format cameras at F8. On of the side effects of using a studio macro range test subject.
One look at the 'Queen' playing card across different images will show you that.
You can use their samples to 'prove' anything you like but it in fact won't prove anything.
The NEX7 seems to have issues with its own native lenses (especially the zoom) when you approach the corners, which is why it looks smeared (NEX5N shows an appreciable difference in corner quality with the same zoom lens) . Pulling crops from near the center shows its actual resolving capabilities...
Of course this quirk it has with its own lenses should be considered when making the purchase but when talking about the sensor and what its capable of I believe using a crop that's not limited due to the reason I brought up would be best.
There isn't much difference in the mid-tones at these ISOs. Shot noise limited in this region, only a small advantage to the NEX sensors. Look at thread spools in the shadow box instead, there you'll see more of an advantage to the NEX sensors with their very low read noise.
You MIGHT want to do a reality-check on those exposures (DPR)...
The Pana often gives a WHOLE stop longer shutter speeds at the same "reported ISO value". This means that when the Pana says "ISO1600" it actually means the same thing as when the Sony says "ISO800". In a real situation, you will have to use a stop higher ISO in the GX to do the same thing.
Kind of like what RED did with their video camera backs, where they published doctored stock fotage shot at what in video terms equals "ISO20,000" that made all the videophiles go "ooohhhh!" and "aaahhh!", and spend a hundred thousand dollars worth of money on RED periphernalia - only to find that in reality what RED meant by "ISO20,000" was more like "ISO6400" in other cameras.
This kind of fooling around with numbers makes it easy to cheat, and fool many people rich in 'internet wisdom' that the FT system is about twice as good as it would be in a real situation.
Always look at the shutter speed and aperture - IGNORE the ISO values in comparisons like this.
This is the NEX7 and the GX1 shot with the same constant light, both at F8 and 1/320s shutter. Guess which is which.
You can't do that, though...you have no idea what the lighting variation might be...its not hard for the lighting to have fluctuations and such. They probably use the same exposure compensation and shoot in Av
If you look at DxO, they measure actual ISO vs nominal ISO, and they actually show that the GX1 is MORE sensitive than the NEX7 at the same ISO.
Click on measurements, then ISO sensitivity. The SNR graphs echo what you see in the DPReview samples...namely that per pixel noise is very similar between the two, with the NEX 7 having a small edge in print noise due to its higher resolution. The NEX's advantages are much more evident in dynamic range and color depth...but noise is really well controlled for a small sensor on the GX1.
Of course DxO doesn't tell you about what the noise looks like at all. To me, they look really close at the pixel level for most tones, with the NEX having superior shadow noise and detail at the same ISO. Still, the GX1 is a great little camera, and I love mine. I have no problem using up to 3200 if needed, and even a well exposed 6400 shot will make good smaller prints (8x10)...even 12,800 can make good 5x7 prints and web shots.
p.1 #12 · High ISO NEX-7 vs Panasonic GX1 & others
Jman13 wrote:
You can't do that, though...you have no idea what the lighting variation might be...its not hard for the lighting to have fluctuations and such. They probably use the same exposure compensation and shoot in Av
If you look at DxO, they measure actual ISO vs nominal ISO, and they actually show that the GX1 is MORE sensitive than the NEX7 at the same ISO.
Click on measurements, then ISO sensitivity. The SNR graphs echo what you see in the DPReview samples...namely that per pixel noise is very similar between the two, with the NEX 7 having a small edge in print noise due to its higher resolution. The NEX's advantages are much more evident in dynamic range and color depth...but noise is really well controlled for a small sensor on the GX1.
Of course DxO doesn't tell you about what the noise looks like at all. To me, they look really close at the pixel level for most tones, with the NEX having superior shadow noise and detail at the same ISO. Still, the GX1 is a great little camera, and I love mine. I have no problem using up to 3200 if needed, and even a well exposed 6400 shot will make good smaller prints (8x10)...even 12,800 can make good 5x7 prints and web shots....Show more →
DxO measures ISO sensitivity on the absolute scale, givning no consideration what so ever to what the manufacturers sets the "base exposure" at. They measure the amount of light needed to overflow the raw file at a given ISO. The manufacturer can set the base exposure (mid gray) at anything between 9% and 18% raw ADU.
Two cameras can have the EXACT same "real ISO", according to DxO, still give a 1Ev shutter speed difference, and still give the same end image brightness. The one with the lower base exposure will then give more noise.
Imaging resource typically check their illumination to be within 1/6Ev. This goes for all shots newer than the old lighting set, which I think they changed into Solux some time around 2007-8.
p.1 #13 · High ISO NEX-7 vs Panasonic GX1 & others
alundeb wrote:
DxO calibrated ISO sensitivity is fixed to 18% of RAW saturation. The individual cameras may position the RAW headroom differently.
Question: Do we know that the same brightness in converted images represent the same percentage of sensor saturation?
We do not. The amount of correction you have to apply when doing a raw conversion varies as between +/- 0.5Ev from a 12.5% base point, in the average lineups we see in the stores today of different brands. Most raw converters adjusts this difference automatically from knowing what camera made the raw, you never notice that it's happening - except when you check the amount of highlight headroom, of course.
p.1 #14 · High ISO NEX-7 vs Panasonic GX1 & others
FYI:
NEX7 avg raw ADU in white patch in mini-CC24 : 5700 (ISO-corrected according to DxO sensitivities)
GX1 -avg raw ADU in white patch in mini-CC24 : 5400 (ISO-corrected according to DxO sensitivities)
The difference in lightning between the two shots was (if DxO is 100% correct, the lenses have the same T-stop at F8.0 and so on) : 0.07Ev.
EDIT:
sorry, got one ratio the wrong way. The IR GX1 @ ISO1600 and the NEX7 @ ISO800 differs by log10(5700/4950)/0.301
