The Lumix GH3, for example, has an electronic shutter, but it takes 1/10s to read out all the rows, making it less useful. Read more about the rolling shutter artifacts associated with the electronic shutter here.
An interesting Panasonic sensor data sheet has popped up. We don't know for sure, but it is widely believed that this sheet describes the sensor sitting in the Lumix GX7 and GM1. Some also speculate that it will be in the upcoming Lumix GH4.
The data sheet says that there are two options for electronic shutter readout: 12 bits in 1/15s and 10 bits in 1/22.5s. It has been demonstrated that the Lumix GX7 has a readout speed of 1/15s, and I have measured the speed of the electronic shutter in the GM1 to around 1/25s. This indicates that you lose two bits of dynamic range if you use the electronic shutter mode of the GM1. Is this a problem? I'll try to find out with a comparison.
A common way to measure the dynamic range, is to see how dark shadow details you can extract, without getting an excessive amount of noise. So the test will be to photograph a picture with a dark area, and see what the noise characteristics are in the dark areas. Here are some pictures taken using both the electronic and mechanical shutter, at various ISOs:
Mechanical shutter
|
Electronic shutter
| |
| ISO 200 |  |  |
| ISO 1600 |  |  |
As you can see, the JPEG output files are very similar. You'd be hard pressed to pinpoint any difference at all.
To better see the difference, I've increased the brightness by three stops, and enlarged the centre dark area. I did the raw processing in Panasonic's own SilkyPix software. Here are the results:
As you can see, there is a more noise in the shadows when using the electronic shutter, both at low and high ISO. So using the electronic shutter has an impact on the image quality. If you increase the brightness by three stops in post processing, anyway.
And that is the point: Most of the time, you don't lift that shadows that much, and you are not likely to note much image quality difference.
Conclusion
When designing the Lumix GM1, it appears that Panasonic made a compromise: To avoid rolling shutter artifacts from the electronic shutter, they increased the sensor readout speed.
At the same time, this loses some dynamic range, and you will get more noise from the shadow areas. However, most people will never notice any difference, since you have to dig deep into the shadows to see any. So I think Panasonic struck a sensible balance here.
I haven't use the Lumix GX7. However, as it has the slower 1/15s electronic shutter mode, one could guess that it uses the high quality 12 bit sensor output. Hence, you should expect the same quality across the two shutter options.
Camera
|
Electronic shutter readout speed
|
Electronic shutter, number of bits
|
1/10s
| 12 | |
1/15s
| 12 | |
1/22.5s
| 10 | |
1/30s
| 10 |
The table above summarizes the speed of the electronic shutter readout. A slow speed means a higher risk of rolling shutter artefacts.
The electronic shutter is almost always on on my GX7, except when I use flash. Like you said, the effect will not matter in most circumstances, but it is good to be aware of this. Thank you.
ReplyDeleteCould you please explain the difference between color depth and noise?
ReplyDeleteThe color balance between the 2 seems quite different and also the left one looks overall softer which makes it hard to compare.
When there is less bit depth (10 bits in E-shutter mode), the shadows will contain less information. This results in the camera processing, or even the RAW processor, not being able to interpret the colours in the shadows correctly.
DeleteFrom the big images, you can see that the colour balance is pretty much identical between the shutter methods. Only when lifting the shadows by three stops, the poor colour rendition appears in the E-shutter images. The colour tends towards green, probably because the sensor has the most green image sensors.
I agree that the difference is smaller at ISO 1600. This makes sense: At ISO 1600, the dynamic range of the sensor is smaller anyway, which means that less bit depth is needed to represent the image.
But then this aint noise but loss of color depth causing a dithering effect in color transitions. But this in consequence might very well be a result of the raw demosaicing process oft the converter you used. Others might be able to interpolate Vetter with this color depth.
DeleteOh and also the ISO 1600 looks actually less noisy with electronic shutter above.
ReplyDeleteI wonder if the electronic shutter mode reduces the ability to pull back blown highlights in raw too. I love this mode for concerts, but the lighting can obviously be quite high contrast...
ReplyDeleteTheoretically, fewer bits would affect the dark areas the more. I haven't tested this, though. Most realistically, I would say that the difference is probably very minor. I wouldn't worry about it.
Delete(I'm using a GH4)
ReplyDeleteSuper cool article thank's for all this precious information. Up until very recently i never even thought about the issue since i was using a FF DSLR which only has a mechanical shutter, but having purchased the Lumix LX100 recently and discovering the mirrorless world as i go along the question arose yesterday and now it is cristal clear to me. Thank's for sharing :-)
ReplyDeleteRay
Montreal, Canada
Be careful with artificial light. E-shutter unaligned with the 50/60 hz will result in banding more easily than normal shutter. My gx7 with 50hz light can do 50, 25, 100th of a sec with no problem, at 200 and above I need the normal shutter. Otherwise I never noticed a difference.
ReplyDelete