I'm wondering whether the E-M1ii's sensor is something new or a variant of those used in the Pen F or GX8. Anyone heard any info?
They almost always say it's a new sensor. I suspect it's going to be the same generation sensor as the pen-f with similar image characteristics. But it's new with faster read out speeds and pdaf attached. We haven't seen many recent major advances with sensor technology, so I doubt we'll see huge improvements with DR or noise. Claims about improved DR and noise are probably based on improved jpg from the faster processor
Well, M4/3 hasn't, but APS-C and FF sensors have improved considerably. When the E-M5 was released, it was basically on par with the Fuji X-Pro 1 and NEX-6 that it competed against. The new sensor in the A6300 and X-Pro2, as well as Nikon's D7200 have made significant strides.
Yea - that's what's worrying me. The specs of this new camera are amazing, but if the IQ doesn't show at least a reasonable improvement I'll definitely be considering my options long and hard.
It's difficult to know how good new sensor is before proper tests but there are two interesting things in already published reports. First is info about native ISO being from 200 up to 6400. If it's true it could mean that new sensor has ~2× better full well capacity which leads to one stop better dynamic range (so claims about it seems right to me). Second thing is possibility of saving hi-res file as lower resolution image. For majority of conditions such images will be almost noise free. I also tried to calculate possibilities of hand-held hi-res mode and it looks interesting. With 60 fps camera needs at worst 1/8s for 8 images. With famous Oly's IS it means that lens up to 100 - 300 mm will be hand holdable (max. value depends on type of IS mode - in camera or camera + lens IS).
In my opinion, the improvements in noise, DR, and resolution are all pretty noticeable. If you asked me, I would never have said that the the Sony A6000 or X-T1 produced results that were much better in any real way compared to M4/3. But I think the A6300 and X-T2 are a legitimate step up now. Not a huge one, and not enough for me to justify a switch, but at least there's a gap. The A6300 still lags the quality of the competition, and the X-T2 still has wormy artifacts that I find unpleasant at high magnifications. But the D7200 image quality is quite impeccable, and actually compares well against the Sony A7 II. Shame it's a DSLR... Studio shot comparison: Digital Photography Review Studio shot comparison: Digital Photography Review Studio shot comparison: Digital Photography Review
Looks like my work Mac can't decode X-T2 RAW files. 6000 v 6300 here, it's a, "I can see a slight difference zoomed in at 100% at ISO 6400" thing. When I downloaded the RAWs and looked at the entire image, I actually preferred the 6000 (color temp mattered), and definitely couldn't tell a difference in terms of noise. Nothing on the order of camera-upgrade-warranting levels of improvement, not even in the same galaxy as that. If I've learned anything, it's the extent to which DPR's studio shot lighting and even object movement can change
Re DR and Noise improvements... The official statements from Olympus have said about 1 stop more DR, but it may have been mentioned in the context of the TruePic VIII processor improvements. Many commenters in forums & blogs are speculating that (only) means the NR will be improved and thus only JPEGs will benefit. Question: did that seem right, technically? Does better NR actually translate to more DR (in JPEG)?? Without hand-waving?
BTW, I saw another comment on a blog last night (Steve Huff, iirc) saying the native ISO is 200-6400 vs 200-3200 prior. He said that should result in lower noise at higher ISO.
I hope that I'm wrong, but I suspect it'll be similar raw IQ to the Pen F. Maybe this new sensor has much faster readout, but I suspect the charge-well design is probably the same. OTOH, I've been doing a lot of long exposure stuff this week and the noise from the E-M1 sensor is atrocious. I'm talking multi colour salt and pepper noise here - not just a bit of loss of detail. If that's fixed then I might be interested. But I guess I could get that with the Pen F for less money - I'd just have to take a saw to that knob on the front
This. In order for me to buy one, long exposure noise HAS to be addressed, as anything over 30s just isn't usable on the E-M1. I like the rear LCD, the better EVF blackout, 121 cross-type PDAF points, stabilized 4K video, but in order for it to end up in my camera bag, I have to be able to do longer exposures and not have concerns about noise. I've done a 9 minute exposure with my GH4, and didn't have any complaints about noise. The E-M1 isn't in the same league when it comes to this, and is a serious short-coming of an otherwise excellent camera.
Olympus did state that the sensor was optimized for low power consumption which should improve long exposure noise. power consumption =heat, heat=more long exposure noise.
I've been attempting to figure out the 'magic' behind the curtain and have a few things to share. The sensor is indeed slightly different to the PenF as it's base ISO is 125 rather than 200, 4000 is the last native ISO. The camera is using 10 bit readout for electronic shutter below ISO 2000 and switching to 12 bit when a higher ISO is selected in an attempt to increase graduation. The 1/60th readout speed figure is in 10bit mode only and for 12bit the readout speed is halved to 1/30th, mechanical shutter always uses 12 bit mode. The amplifier stage in the ADC has steps which correspond to 125 (base), 250, 500, 1000, 2000, and 4000. I *think* 8000 and higher are extended ISO (pushed) however someone will need to really dig into RAW files in future to know this for sure. The odd base ISO has been done by Panasonic before (160 in consumer cameras is common, 2000 in professional is also common) however afaik it's a first by Olympus. Edit: this is all based on napkin math, I haven't slept yet (it's 7am) and I could have miscalculated something, time will tell. Please refrain from rushing off and making bold claims based on something someone said on an internet forum (or reporting it as a rumor as was previously done for one of my recent posts... it did get a FT4 status which was amusing to me) - until someone can do a deep analysis of the production camera nothing is certain.
This could works for extended ISO where producers are generally more willing to improve results by software but personally I hope that they didn't applied aggressive denoising to RAW files as it looks often unnatural at high ISO. Theoretically yes, practically no. Dynamic range defined as the ratio between the full-well capacity and the readout noise. If there is some algorithm applied on readout noise then dynamic range is seemingly better because readout noise is seemingly lower. Problem is with information at noise level. At noise levels algorithms have problem to recognize what is and what isn't desirable information. So they often effectively destroy real details in image. Another problem is that such RAW does not represent true output of sensor and as such it's misleading to say that sensor has better dynamic range. In case of JPEG there is much more pronounced dependence on processing of the RAWs. More advanced algorithms gives better results but there is always hardware (sensor + AD convertor) which is the most important part. I'm pretty sure that with new TruePic processors JPEG will be better. What I see as biggest problem in current reports about E-M1II is that it has still 12-bit output. This effectively means limited increase of DR at low ISOs, at high ISOs it will be better (if there is really 1EV better DR of sensor). I saw also info about native ISO being 200-6400. That's why I think that full well of sensor will be significantly better than in previous cameras of Olympus. It doesn't mean lower noise at higher ISO per se but it means lower noise at higher ISO if readout noise is similar as in previous cameras. If so, then dynamic range is really better by ~1EV but only at higher ISO settings. This is well known problem of E-M1 and generally all Panasonic sensors. Dark noise of new E-M1II will be much lower, on par with sensors in E-M5 up to PEN-F.
I think that 14-bit only makes sense if there is a really low S/N. u43 sensors probably don't have a sufficiently good S/N to justify the extra bits. All that would be there would be noise.