Olympus E-M5 Mark II - love and hate at first sight

[fuzzynormal]

If IBIS becomes the next big thing in ILCs with the BUYING public and IBIS happens to kill video quality, then kiss video quality goodbye in ILCs.

​Sure, that could happen.  What you propose is a hypothetical possibility.  I don't, however, think it's a realistic possibility.

As for the Ex Tele on the GX7, I use vintage glass at maximum aperture for a lot of what I do.  As such, I just stay away from the digital zoom with video.  Yes, for some, it's perfectly useable, especially with new lenses stopped down a touch.

[Portlandia]

Lens auto focused during movie shooting --while in Manual focus mode, and this was with an Oly lens.  Strange.

​Had the same issue several times an couldn't figure it out.  Finally traced it back the function button on the front of the camera.  When in filming If you're using an autofocus lens in manual focus mode and accidently press this button it will autofocus the lens.  Definitely an operational bug, because even after I unassigned any functions from this button it still occurred.  This same function button (default DOF) works properly when taking stills.

[mat33]

I've heard this before, but do we actually know that IBIS creates sensor heating issues which affect video quality?  I always thought sensor heating was more of any issue for long exposures, hence the additional cooling seen on many astrophotography cameras.   For video, I thought most heat was created from the image processing/compression which occurred off sensor, and that line skipping etc was used to minimise the data needed to process (while also speeding sensor readout to minimise rolling shutter).  Given that the E-M5ii is olympus's first camera where video is a main feature it would seem likely that any issues with video quality have more to do less experience with processing/compression etc than IBIS, look at how long it took Nikon to produce D750 quality 1080p.  

 

 

[Bob Goldberg]

​Sure, that could happen.  What you propose is a hypothetical possibility.  I don't, however, think it's a realistic possibility.

As for the Ex Tele on the GX7, I use vintage glass at maximum aperture for a lot of what I do.  As such, I just stay away from the digital zoom with video.  Yes, for some, it's perfectly useable, especially with new lenses stopped down a touch.

​That's because I'm not sure you hang out in the same corners I do. Olympus seems to have much more of a bigger (and more rabid) cult following than Panasonic and Samsung (and possibly Sony), although for the life of me, I can't figure out why. I liken them to a pack of wolves waiting for the slightest thing to pounce on and proclaim Olympus the king of video. I've honestly never seen anything like it before. For instance, I've been reading the DPReview forums for a long time (especially micro four thirds), and I've never seen them care about video in the slightest.

Even when Panasonic was breaking totally new ground with the GH2 and GH4, not a peep out of that forum. Then, all of a sudden, Olympus so much as markets their camera for video (without actually delivering anything), and these rabid wolves come out from nowhere with thread after thread about how the E-M5 II is better for video than the GH4 and its 4K.  There were a few threads inquiring about getting a GX7 or GH3 vs an E-M5 II for video, and the consensus seemed to be that the E-M5 II video was on par with the GH3 but the added IBIS gave it an edge. Well, this is utter nonsense, because you'd have to be blind to believe the E-M5 II was anywhere near the GH3. This type of stuff is really what prompted me to get involved.

Never mind the fact that these people don't shoot a lick of video and don't care in the slightest about video and don't know a thing about video. They just want to promote their brand. And it simply isn't one or two people doing this but a large subset of the forum. When I started some of my threads about how the E-M5 II was actually poor for video (contrary to the press releases and John Brawley), the anger was palpable. And you'd have to wonder why, since, again, the vast majority of these people don't shoot video and don't care about it.

I simply don't understand this mentality because I need something that actually works, and I don't care which brand it comes from. And I believe these type of people are honestly why technology can regress. There's no logic. There's no reason. There's only some kind of misplaced emotion.

Regarding the GX7, I'm not convinced that the ETC mode on that is going to be as good as the GH series. It may be decent but the GH cameras are an entirely different beast. But I don't really have any of the more modern non-GH cameras, such as the GX7, to compare. So, I could be wrong and possibly the GX7 is right up there.

[Bob Goldberg]

I've heard this before, but do we actually know that IBIS creates sensor heating issues which affect video quality?  I always thought sensor heating was more of any issue for long exposures, hence the additional cooling seen on many astrophotography cameras.   For video, I thought most heat was created from the image processing/compression which occurred off sensor, and that line skipping etc was used to minimise the data needed to process (while also speeding sensor readout to minimise rolling shutter).  Given that the E-M5ii is olympus's first camera where video is a main feature it would seem likely that any issues with video quality have more to do less experience with processing/compression etc than IBIS, look at how long it took Nikon to produce D750 quality 1080p.  

 

 

​The answer is, I don't know for certain. There has certainly been a lot of speculation that IBIS is not compatible with the most high end video and that it might be possible to produce decent or passable video (such as D750 quality), but not GH4 type video. Panasonic execs have weighed in on this issue in interviews with IR:

http://www.imaging-resource.com/news/2012/09/28/qa-with-panasonic-the-story-behind-the-new-gh3-and-compact-system-tech

http://www.imaging-resource.com/news/2014/06/17/panasonic-executive-interview-part-ii-gh4-4k-isnt-just-about-video-and-read

Again, it's entirely possible the speculation is wrong, but, until I actually see a camera with IBIS active in video mode (not just having IBIS like the GX7) and producing 4K video like the GH4, I'm going to continue to assume that it does have a negative impact.

[fuzzynormal]

...out from nowhere with thread after thread about how the E-M5 II is better for video than the GH4 and its 4K.  There were a few threads inquiring about getting a GX7 or GH3 vs an E-M5 II for video, and the consensus seemed to be that the E-M5 II video was on par with the GH3 but the added IBIS gave it an edge...Never mind the fact that these people don't shoot a lick of video and don't care in the slightest about video...There's only some kind of misplaced emotion.

Well, look, I bought one simply because the 5-axis.  It will make some specific work I do a bit more practical and faster.  That's the bottom line for me, but I think I'm a specific case.  Regardless, there is a reason the IBIS is coveted.   Even so, I don't see how it would really move the needle of the entire marketplace to preferring 5-axis over great IQ.  Moreover, it's probably not something worth worrying about.  People that are blindly brand loyal are a weird lot not to be trusted anyway.  My 2cents.

Obviously, we'd all like great video with 5axis, but I'm guessing for right now we can rely on Panasonic and Sony to deliver IQ before stabilization...and that's even assuming that 5-axis directly affects IQ, which may or may not be the engineering reality.

I'll probably buy a Gh4 later this year; heck, right now it's only $100 more than a EM5II...

[John Brawley]

The impact of the sensor working at higher temperatures is that you get more noise and other sensor faults like FPN and dead pixels are more likely to show up.  These are the symptoms of sensor heating.  We talk about overheating and it's not really because the components will somehow fry, it's just that more heat can change the performance of the sensor.

If you were looking for symptoms of overheating (or operating outside of normal temperature range)  that's what they would be...

Some cameras have black balance features that enable you to set the black levels (and thus the noise floor) and to also re-map dead pixels.

Most recommend you do this at the "operating" temperature of the camera.

I have been shooting a lot with the Sony F55 a lot.  We're shooting day nights at the moment and on the very first setup about 20 mins into our day I spotted a dead pixel.  The camera had been on for a while and I hadn't seen it on the two rehearsals we did.  After an APR the dead pixel went away.

The F55 has a black balance / dead pixel re-map feature they call "APR" and I normally have my assistants have the camera on for 30 mins and then do the APR function once it's at operational temperature.

RED have similar issues because their cooling generally isn't as good.  On very long takes when their cooling fans wind down for sound, they can often have the sensor temp go above the range specified during their black balance.  They have a specific warning for this and their suggestion is to pre-heat your sensor to your normal operating condition before doing a black shading / black balance.  The idea is that you're calibrating the sensor for that operating temperature, even if it's higher than normal.

Alexa (and Blackmagic) approach it differently.  Neither camera has a black balance / dead pixel facility available to the user.  Instead they use solid state cooling to maintain the sensor temperature to their ideal operating temperature.  It's like a refrigerator with a thermostat on the back of the sensor.  They maintain a constant thermal performance and that, along with dual gain architecture sensors is why they both have such large DR (the 2.5K and pocket).

JB

 

 

 

 

 

[Bob Goldberg]

The impact of the sensor working at higher temperatures is that you get more noise and other sensor faults like FPN and dead pixels are more likely to show up.  These are the symptoms of sensor heating.  We talk about overheating and it's not really because the components will somehow fry, it's just that more heat can change the performance of the sensor.

If you were looking for symptoms of overheating (or operating outside of normal temperature range)  that's what they would be...

Some cameras have black balance features that enable you to set the black levels (and thus the noise floor) and to also re-map dead pixels.

Most recommend you do this at the "operating" temperature of the camera.

I have been shooting a lot with the Sony F55 a lot.  We're shooting day nights at the moment and on the very first setup about 20 mins into our day I spotted a dead pixel.  The camera had been on for a while and I hadn't seen it on the two rehearsals we did.  After an APR the dead pixel went away.

The F55 has a black balance / dead pixel re-map feature they call "APR" and I normally have my assistants have the camera on for 30 mins and then do the APR function once it's at operational temperature.

RED have similar issues because their cooling generally isn't as good.  On very long takes when their cooling fans wind down for sound, they can often have the sensor temp go above the range specified during their black balance.  They have a specific warning for this and their suggestion is to pre-heat your sensor to your normal operating condition before doing a black shading / black balance.  The idea is that you're calibrating the sensor for that operating temperature, even if it's higher than normal.

Alexa (and Blackmagic) approach it differently.  Neither camera has a black balance / dead pixel facility available to the user.  Instead they use solid state cooling to maintain the sensor temperature to their ideal operating temperature.  It's like a refrigerator with a thermostat on the back of the sensor.  They maintain a constant thermal performance and that, along with dual gain architecture sensors is why they both have such large DR (the 2.5K and pocket).

JB

​What about color balance and color shift issues? On some of my old CCTV cams (composite video out) that had poor heat sinking, the colors would shift dramatically after they were left on for a while. Of course, those were CCD rather than modern CMOS sensors, but I would be surprised if there were no color shift issues with sensor overheating, even now.

And let's say you don't have a professional camera but a consumer level camera like the E-M5 II that has no warnings for overheating, no time limits on recording, and has to maintain a constant quality regardless of settings, environmental conditions, or how long you're recording.

What type of manifestation would overheating have on a cam like this? I'm no expert on the issue, but I would assume the way they would handle noise issues is to increase the NR to the point where the noise would be concealed, even under worst case scenarios. In that way, you wouldn't see any difference in quality whether you're recording for 1 min or 20 min or whether you're recording in 20 F temps or 100 F.

Presumably, you see no difference in quality on the E-M5 II regardless of conditions or record times, correct?

[Cinegain]

​What about color balance and color shift issues? On some of my old CCTV cams (composite video out) that had poor heat sinking, the colors would shift dramatically after they were left on for a while. Of course, those were CCD rather than modern CMOS sensors, but I would be surprised if there were no color shift issues with sensor overheating, even now.

And let's say you don't have a professional camera but a consumer level camera like the E-M5 II that has no warnings for overheating, no time limits on recording, and has to maintain a constant quality regardless of settings, environmental conditions, or how long you're recording.

What type of manifestation would overheating have on a cam like this? I'm no expert on the issue, but I would assume the way they would handle noise issues is to increase the NR to the point where the noise would be concealed, even under worst case scenarios. In that way, you wouldn't see any difference in quality whether you're recording for 1 min or 20 min or whether you're recording in 20 F temps or 100 F.

Presumably, you see no difference in quality on the E-M5 II regardless of conditions or record times, correct?

​I'm no expert either, but... evaluative noise reduction is a thing? Doesn't the processor just have a preset degree of noise reduction it should apply? This should be regardless of the actual amount of noise perceived/recorded. To my knowledge there's no controller that measures noise and then applies a variable degree of noise reduction to suit the needs of that particular recording. You set it to 0? It will reduce noise according that set degree of noise reduction. You go for NR - 2? It will apply noise reduction accordingly, from what I know, there's no response to operating sensor temperature. That's why you actually experience what John describes: 'The impact of the sensor working at higher temperatures is that you get more noise and other sensor faults like FPN and dead pixels are more likely to show up'. I had this with a few cameras. For obvious reasons most notably shooting with the BMPCC. So... IF heat was that drastically effecting performance to the point it could be perceived as worse peformance (but it's 2015, cameras can handle quite a bit), noise reduction won't just apply a more agressive noise reduction, the quality output will degrade as you keep rolling. Never experienced any noticeable color shifts btw.

[Bob Goldberg]

​I'm no expert either, but... evaluative noise reduction is a thing? Doesn't the processor just have a preset degree of noise reduction it should apply? This should be regardless of the actual amount of noise perceived/recorded. To my knowledge there's no controller that measures noise and then applies a variable degree of noise reduction to suit the needs of that particular recording. You set it to 0? It will reduce noise according that set degree of noise reduction. You go for NR - 2? It will apply noise reduction accordingly, from what I know, there's no response to operating sensor temperature. That's why you actually experience what John describes: 'The impact of the sensor working at higher temperatures is that you get more noise and other sensor faults like FPN and dead pixels are more likely to show up'. I had this with a few cameras. For obvious reasons most notably shooting with the BMPCC. So... IF heat was that drastically effecting performance to the point it could be perceived as worse peformance (but it's 2015, cameras can handle quite a bit), noise reduction won't just apply a more agressive noise reduction, the quality output will degrade as you keep rolling. Never experienced any noticeable color shifts btw.

​That's exactly my point, about NR not continuing to increase as the sensor continues to heat up. My point is that the manufacturer will turn up the NR high enough from the outset that the video won't look any different whether the sensor is cool or hot because they won't want you to perceive a difference in IQ depending on the circumstances. Of course, I could be wrong about this, but I do believe this is what they do.

One of the main reasons we perceive differences in detail between different cameras is precisely because different cameras have different levels of NR depending on various factors. Smaller sensor cameras will generally have more NR so that they might look as clean as large sensor cameras in low light, but the level of detail will be way different. Similarly, BSI sensors can do with less NR in low light, so the only perceptible difference between a BSI sensor and a non-BSI sensor in low light might be a difference in detail (not necessarily noise) because the NR applied to the non-BSI sensor wipes away that detail.

I believe what manufacturers do is apply NR to suit the worst possible conditions of ambient temperatures, record time, etc. so that there will be no perceptible differences depending on conditions.

 

[Bob Goldberg]

"Regarding the GX7, I'm not convinced that the ETC mode on that is going to be as good as the GH series. It may be decent but the GH cameras are an entirely different beast. But I don't really have any of the more modern non-GH cameras, such as the GX7, to compare. So, I could be wrong and possibly the GX7 is right up there"

Hi Bob, I agree with just about all you say in your posts regarding EM5 II, except maybe with the march of tech. IBIS/heat issues may be resolved in the future.

 

Regarding the ETC feature on GX7, it looks not bad compared to GH4 ...

I did a quick & dirty comparison (same lens) of GX7 & GH4 with ETC ON, there's nothing in it really at the similar settings of ...  Avchd, 28MBS, 50fps setting in both cameras.

Please note these are only frame grabs (GH4 with ETC OFF just for reference) and obviously only tell one aspect of video quality.  Still, the GX7 looks good!

 

GH4 1st

GH4 2nd

GX7 3rd

 

 

​

That's interesting. From your test, I can see no obvious difference between the two, but sometimes it takes additional testing to actually find a difference, if there is one. One of the things I like to do to really put the quality of the ETC mode to the test is to compare it to a 1920x1080 crop from a still image taken with the camera under identical conditions and using identical settings.

This is a comparison of two images from my GH2 using identical settings. The first is a 1920x1080 crop from a jpeg. The second is a frame grab using 1080p ETC mode and simply run through temporal noise reduction (no profile building or anything like that).

It's quite clear to me that the ETC mode retains quite a bit more detail than the jpeg still images because you can't do temporal NR on still images. I can get a frame grab from the ETC mode cleaner and sharper.

Really, this is quite amazing work from Panasonic on a 2010 camera. Their video compression simply defies belief, IMO.

 

[Don Kotlos]

Here is another E-M5ii sample that somebody uploaded on Vimeo. 

Somebody uploaded another example from the E-M5ii:

 

 

[DPC]

I'm not getting anything like that sharpness out of my camera.

[Don Kotlos]

I'm not getting anything like that sharpness out of my camera.

​This is similar to the sharpness that I get too, and that is why I was so surprised with Andrew's take on the camera. 

Either is the settings that can really affect the quality or there is a bad batch of cameras. 

[DPC]

What settings are you using? I've been using various variants of the Natural Picture Mode. 

[Don Kotlos]

I am going around the settings that I showed with the towel example: Contrast:-2 Sharpness:0 Saturation:-1 . From the curves a step decrease of the whites, a step increase of the blacks. That is in All-I 24p 1/50 and with the 17mm f1/8. 

I am trying to find out what settings that guy used for that test. Edit: Apparently he used the natural profile: (Contrast, Sharpness, Saturation ALL set to '0'). Thanks for asking him too ;). 

Maybe the lens is also affecting the quality? So could it also be that with Olympus lenses there is some kind of compensation that manual lenses don't get? 

 

 

[DPC]

I asked that. Natural with everything set to 0. What's interesting is that you are both using 24P and I'm using 25P. I wonder if that could make a difference?

[Bob Goldberg]

I'm not getting anything like that sharpness out of my camera.

​This type of video gives the illusion of sharpness by shooting nothing but close-ups. What I've noticed with people claiming this camera produces sharp video is that they're almost invariably shooting at point blank range.

When someone shows me a sharp video with distant landscapes or even trees, shrubs, and other plant life at moderate distances, then that's a start.

But all this extreme close-up nonsense is simply a coverup for the deficiencies of this camera.

[Don Kotlos]

I asked that. Natural with everything set to 0. What's interesting is that you are both using 24P and I'm using 25P. I wonder if that could make a difference?

​Yeah I figured that out a bit later.. It would be really strange for the framerate to affect it but you never know... 

[Inquisitive]

I finally received my camera and the crop appears to be 80% of my GH2.  So my 12mm looks more like 15mm and it's that way ibis on or off.  I would have rather they given me all the field of view I could get ibis off but it is what it is. As far as quality I haven't had it long but it doesn't seem as bad as some have made it out to be. I'm going to run it against my hacked GH2 if it beats that I will be happy enough.