Discussion in 'Panasonic Cameras' started by stingOM, Mar 17, 2011.
Panasonic Lumix GH2 noise tests
Thanks for that.......good reading.
That's an interesting link.
I have posted this in a number of other places before, so I hope that those of you who have seen it before don't mind, but for those who haven't, this is an example of a candid, hand held picture snapped at a birthday party... I was in the interior of the room, no windows behind me, but windows behind the subject. GH2 at ISO 2500, 14-140 lens fully extended to 140mm (280mm equiv.), 1/15th of a second (excellent OIS on), again hand-held:
No noise reduction, simply processed from RAW in LR3:
With a small amount of NR:
I draw two conclusions from this:
1) if you shoot the GH2 in RAW and don't underexpose, the noise levels at higher ISOs (2500 in this case) can be surprisingly acceptable.
2) the OIS of the 14-140 is amazing!
I agree. One practical issue pertaining to GH2 noise performance is that read noise is highest at the low ISO values. This is in contrast to the D7000, which seems to have similar read noise at all ISO values. The consequences of this are that:
Low ISO GH2 file which has been underexposed and then "pushed" during RAW processing will have more shadow noise than a file which was shot at higher ISO without need to subsequently push
Low ISO D7000 file which has been underexposed and then "pushed" during RAW processing will have equal shadow noise than a file which was shot at higher ISO without need to subsequently push
In other words, with the GH2 (and most other digital cameras) it is usually best to use whatever ISO is necessary to get an apparent exposure "to the right".
Here's an ISO 3200, 45mm, f/2.8, 1/50s GH2 file processed in LR3 with minimal curves adjustments and no change to default NR/sharpening:
The horizontal lines shown in the referenced article are interesting. I have no idea what causes them and don't see those in my day to day photography with the GH2, though I did see them with the GH1.
Amin.... wow, that is exceptionally clean for ISO 3200. When you say "no change to default NR/sharpening" I take that to mean that some NR was applied. Do you mean the default NR that was applied in camera by the JPG processing, or some "default" NR was applied in LR? If in LR, do you know the specific amount that was applied?
Don, it was the default NR setting in LR3. I don't remember exactly what that is set to, but I'll check later and report back.
Good write up. I think there's one small error or at least misleading sentence in that though where he/she says, "The maximum amount of photons one sensor pixel can handle is 4096. This number is not exactly equal to the real amount of photons, but the ratio can be compared. Lets call them virtual photons. v-photons."
Unless I'm misunderstanding what's being said, bit depth of the image, or number of gray levels, though it can relate to dynamic range, is not the same as full well capacity of each pixel. Also, the number of photons which can be successfully converted to photo-electrons is quantum efficiency. Expressing things in terms of image bit depth seems, if not wrong, at least a very round about way of trying to describe these two different aspects.
Separate names with a comma.