Micro 4/3 New Sensors Talk

[agentlossing]

I thought it might be interesting to separate out a discussion on the merits of the sensors in the OM-1 and GH6, since anytime Micro Four Thirds introduces a step up in sensor resolution or IQ, it's a momentous occasion. As time goes on hopefully we'll have several firsthand accounts to add to the discussion, but as it stands, for the moment, I am mainly looking at samples from different websites. The GH6 uses some wizardry to combine a lower-gain and a higher-gain amplification at ISO 800 and above to reduce noise, and obviously a bump in resolution (past the typical past/current gen 24mp sensors in many APS-C cameras, but just behind the newest 26mp ones - not a bad place for M4/3 resolution to be, at all). The OM-1 retains a 20mp output, but combines data from 80mp of photodiodes and aims for better high-ISO performance via its stacked BSI design. Different approaches, actually the most different that flagship M4/3 sensors have been, and the first time since the introduction of PDAF in Olympus cameras that they've really branched apart from each other.

I've seen some good initial samples from both. Interestingly, DPR's static test shots for the GH6 look considerably worse than the OM-1, but they appear to have different exposure levels given the information they provide, and I feel the GH6 samples are underexposed. Certainly there are better examples of what it can do in the low-light real-world shots they've included.

Anyway, what do you think of these two companies' interpretations of the way to move forward with their sensors? It's an interesting time, if nothing else, after another substantial drought.

 

[bartjeej]

Panasonic made it pretty clear their sensor isn't provided by Sony. Sony have been leading the worldwide sensor race for about a decade now, but apparently this mysterious other party could offer Panasonic some things that Sony couldn't - presumably some specialist features on the video front (the dual simultaneous gain mode sounds pretty cool for dynamic range, but I don't know if that's the sensor's raison d'etre). The readout speeds on the GH6 aren't ridiculously fast, but the video bitrates are high agross the board for all settings.

I noticed the GH6 costs the same in the US as the OM-1, despite the GH6 arguably being fuller featured, (even) more professional use oriented, with features like built in fan, dual screen movement, full HDMI out, rec light and countless other high end video specific features. So maybe one benefit of buying from a party that isn't Sony is also a cheaper sensor?

 

[agentlossing]

Panasonic made it pretty clear their sensor isn't provided by Sony. Sony have been leading the worldwide sensor race for about a decade now, but apparently this mysterious other party could offer Panasonic some things that Sony couldn't - presumably some specialist features on the video front (the dual simultaneous gain mode sounds pretty cool for dynamic range, but I don't know if that's the sensor's raison d'etre). The readout speeds on the GH6 aren't ridiculously fast, but the video bitrates are high agross the board for all settings.

I noticed the GH6 costs the same in the US as the OM-1, despite the GH6 arguably being fuller featured, (even) more professional use oriented, with features like built in fan, dual screen movement, full HDMI out, rec light and countless other high end video specific features. So maybe one benefit of buying from a party that isn't Sony is also a cheaper sensor?

Didn't Panasonic work in the past with Tower? Do we know if their collaboration is ended, or still going on? I do find it really interesting that both cameras have the same US price, as you mentioned. Possibly Panasonic's full frame lineup provides some incentive to keep smaller-sensor prices down, even on the flagship GH camera? At any rate, OM Solutions (and Olympus before that) has no higher-specced product available.

 

[agentlossing]

One thing's for sure doing some more browsing of DPR's sample galleries: the Panasonic is not outputting very good JPEGs for detail and NR. Possibly the non-final firmware is to blame, but we'll have to wait to see how well the RAWs can be processed. The OM-1 seems to put out more pleasing JPEGs even at very high ISO, but it also has a major advantage in that RAWs processed with the new tools in OM Workspace can appear on the internet and get a lot of eyes before the full power of Panasonic's sensor can even be shown off.

 

[davidzvi]

.....
I noticed the GH6 costs the same in the US as the OM-1, despite the GH6 arguably being fuller featured, (even) more professional use oriented, with features like built in fan, dual screen movement, full HDMI out, rec light and countless other high end video specific features. So maybe one benefit of buying from a party that isn't Sony is also a cheaper sensor?

I would STRONGLY disagree with this statement.

The GH6 is the more full featured and professional VIDEO camera. But other wise no. I'm trying to compare a few things for wildlife and sports; it looks like the older G9 and E-M1.3 are more full featured in those areas than the new GH6.

I haven't seen anything on the GH6 sensor yet, but here is a nice compare video of the E-M1.3 vs OM-1.

 

[agentlossing]

DPR put out an explainer for dual-gain sensors, and particularly how the GH6 version differs: What is dual gain and how does it work?

I found the explanation of standard dual-gain sensors a little hard to comprehend and had to reread it a few times to "get" it, but the explanation of the GH6 further down is mercifully easier for tired brains. Basically the GH6 takes two readouts of a single exposure, one with gain applied to capture more shadow information, and one without gain, which preserves more highlights. These are then merged together on saving the data. This might result in more shadow noise, because of the gain applied to dark areas, but also gives an additional one stop of detail to highlights - which could be really nice. Plus, of course it allows you to expose a brighter photo since you'll have more highlight headroom, which is better for the shadows. Essentially, it sounds like while "expose for the shadows" won't be the rule, it's a step in that direction. Real world experience (with editable RAWs) will have to bear this out.

 

[albertk]

I have the Panasonic S5, dual gain. I like the low ISO, but high gets flatter imho. And there are some artifacts . . .
Like the electrolic rolling shutter in electronic shutter mode
The thing is : do we like a camera having an inbuilt 110 year old 'Lartigue-style' (the oblong lines of the Grand Prix of the ACF, Dieppe, by Jacques-Henri Lartigue, 1912) ? ?

This effect by Lartigue is caused by a rolling shutter and panning:

With many new sensors in the electronic shutter mode, the sensor frame is read out in a slow fashion, with a horizontal mode, line for line. You will get the same.
Just to poke some fun I will experiment. In the following I thought I saw the effect in the rolling red moto wheel but now I am not sure.

Join to see EXIF info for this image (if available)

 

[agentlossing]

DPR has published their test images for the OM-1, and states that there isn't much difference between the E-M1 III and the OM-1. But I think a lot of folks are missing the point about usable ISO. I switched the lighting to low-light (since you're mainly bumping ISO in low light, unless you're capturing really fast action) and looked at the results at ISO 6400, 12800 and 25600. I'd say the E-M1 III remains usable at 6400, but falls apart after due to the blotchy color noise that's nearly impossible to correct fully. The OM-1 stays very decent through 25600, and the color noise blotches show up at 51200. For me, that makes a usable ISO 12800 which I'd never have considered with previous M4/3 cameras. Here's an example someone else posted in the comments.