Real World Tests
So what’s the high and low ISO performance actually like?
To give you a meaningful example I shot the same subject at both base ISO160 and ISO1600; the point where I found detail loss begins to show. The two higher sensitivities of 3200 and 6400 are still very useable in the right circumstances though and not ‘for emergency use only’ as used to be the case; see the third example shot below.
Example 4. Base ISO160 – Natural light plus off-camera flash, 1/60th @ f/5.0, Aperture priority, 45-200mm lens @ 124mm(Note: I deliberately went for a dark background on this shot as such would emphasise any noise present at the G3’s base ISO. It’s possible though that reducing the image for display on this site will actually introduce artefacts not present in the original file. Check out the full res’ version on Flickr -link at end of review- for accuracy)
As you can see from the second shot the level of noise control/removal is remarkably good in G3 JPEGs. Viewed in Silkypix, with all sharpening turned off, the RAW of the ISO1600 image also demonstrates much less noise than I was expecting while the ISO160 RAW is effectively noise free. This is perhaps proof that Panasonic’s research into lower noise versions of the Four-Thirds sensor is really delivering results.
Example 6. ISO6400 – Dim tungsten light plus on-camera TTL flash, 1/80th @ f/5.6, Aperture priority, 45-200mm lens @ 200mm, Exposure Compensation -2EV(Accidentally set!- hence this shot is a little dark)
The lighting conditions for the above shot were pretty grim and yet the G3 managed to deliver a usable JPEG at its maximum sensitivity. It was during this shoot that I experienced several of the ‘handling issues’ referred to earlier, principally because I was working quickly, eye to viewfinder, trying to catch interesting moments. My thanks to Acoustic Chairs for being willing test subjects.
In summary, in-camera produced JPEGs are effectively noise free up to ISO800, show a slight detail loss (but very little noise) at 1600 and only begin to really suffer at 3200 – though again are greatly improved over past models. At ISO6400, the top sensitivity on offer, obvious artefacts are present, particularly in darker areas, but it has to be said that this is a remarkable performance by the standards of any size of sensor. Untreated RAW files (viewed for now in the copy of Silkypix 3.1.6 supplied with the G3) look a little soft and do start to show noise in darker pictures around the ISO400 mark. However, the nature of that noise, right up to 6400, is surprisingly ‘smooth grained’ and ideal for effective removal by processing in one of the more advanced RAW development engines – something I’ll report on in detail once Adobe catch up with the arrival of the G3 as I’m not convinced the current edition of Silkypix is quite up to the task of optimally dematrixing RAWs from the new sensor. I came to this conclusion after loading some of my ‘best’ G1 RAWs into Silkypix for comparison, where they suffered in a similar fashion when compared to their appearance in Lightroom/ACR.
The noise performance of the G3 is another major advance. Whether or not it’s a ‘killer’ feature will depend or your own shooting style, but it’s hard to deny that it will enable us all to take better pictures in difficult circumstances.
But does the G3 produce better shots than the G1/G2 at base ISO?
I know that’s a question that many members will be asking. The quicker answer is ‘yes’ as far as I’m concerned. However, you really do need to compare two identical shots at full 1:1 resolution to see the improvement as; fine detail resolution aside, the difference can be fairly subtle. Here’s an example using two almost identical shots at the base ISO for each model.
Example 7. G1. ISO100 – Natural light, 1/8th @ f/10.0, Aperture priority, 45-200mm lens @ 144mm
Example 8. G3. ISO160 – Natural light, 1/8th @ f/10.0, Aperture priority, 45-200mm lens @ 147mm
These shots were pattern-metered and I think it’s just a coincidence that both cameras chose the same exposure settings in the varying natural light. What you may notice even at this reduced resolution is that the G1 shot looks more saturated than that from the G3. Comparing the histograms of the two JPEGs in Lightroom shows them to be quite different with the G3 having a more even distribution of colour levels that the G1; which very nearly clips the red channel in this shot. This is all open to interpretation but my thoughts so far are that the G3 is even more neutral at its default settings with ‘hot’ subjects like this than the G1. I’d call that a welcome advance, especially for those of us who shoot RAW.
Note for newbies: Saturated reds have always given digital cameras problems; often clipping (losing all detail) well before other colours in the scene.
Everything I took over the week with the 16MP G3 indicated that it is capable of taking better pictures than previous G’s using the original G1 type 12MP sensor. You won’t see a staggering difference in every shot, but it wins every time with highly detailed and textured subjects.
Has the Dynamic Range Improved?
I don’t have a lab full of test equipment, so to test the G3’s dynamic range I looked for shots that would be difficult for the camera to handle.
Example 9. ISO160 – Natural light, 1/6th @ f/10.0, Aperture priority, 45-200mm lens @ 175mm, EC +1EV
This shot of a white porcelain cat on a black silk embroidered cloth was a deliberate set-up intended to push the G3’s dynamic capability to the limit. After pattern metering, plus 1EV exposure compensation was required to bring out the black embroidery – no surprise there; you’d need to manually adjust a shot like this with any camera. What was gratifying is that the shot’s histogram only shows clipping where you’d expect it to, in the deepest shadows and the small highlights in the cat’s highly reflective glaze.
Studio set-ups like this are one thing, but I also wanted to find some real-world examples.
Example 10. ISO160 – Bright day light, 1/100th @ f/8.0, Aperture priority, 45-200mm lens @ 200mm, iR & iD on
This shot isn’t perfect (my fault in this case) but it does show pretty good handling of a naturally wide dynamic range subject. The G3 has managed to maintain most of the feather detail in both the extremely white and mostly dark ducks. From the many shots I’ve taken at this location over the years I’d say that the G3 did very well on this shoot (battery performance aside). So, I think the G3 demonstrates some improvements in dynamic range handling over previous models. Don’t expect miracles though, as I found out on the last shooting day it’s still pretty easy to blow bright skies in difficult conditions.