Pages: [1] 2   Go Down

Author Topic: Olympus E-M5 firmware 2.0: the new 'ISO LOW' is a real ISO160  (Read 19431 times)

Guillermo Luijk

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1765
    • http://www.guillermoluijk.com
Olympus E-M5 firmware 2.0: the new 'ISO LOW' is a real ISO160
« on: January 30, 2014, 06:50:21 PM »

It seems Olympus is now learning from other brands such as Fujifilm, offering its customers really useful firmware updates. In the last days a new software release seems to provide, among other interesting improvements, an 'ISO LOW' setting. But is that a real ISO setting? or just an ISO200 capture holding metadata so that RAW converters can apply clandestine underexposure to make the camera JPEG display as lower ISO?.

I have done the calculations over RAW data, and concluded the new 'ISO LOW' setting corresponds to a real ISO160 in the camera's ISO scale, providing 1/3EV of extra highlight headroom to RAW shooters with expected consequences in SNR and dynamic range:


RAW histograms in log scale


You can read the whole demonstration in the article (English): OLYMPUS E-M5 NEW FIRMWARE. IS 'ISO LOW' REAL?.

Regards

Johnny_Johnson

  • Guest
Re: Olympus E-M5 firmware 2.0: the new 'ISO LOW' is a real ISO160
« Reply #1 on: January 30, 2014, 08:12:10 PM »

I thought that Olympus was already overstating the ISO in that camera by a factor of two.

Later,
Johnny
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6297
Re: Olympus E-M5 firmware 2.0: the new 'ISO LOW' is a real ISO160
« Reply #2 on: January 30, 2014, 09:00:32 PM »

I thought that Olympus was already overstating the ISO in that camera by a factor of two.

Later,
Johnny
No, that is belief is based on misinterpretation of the fact that EM5 raw files have more than the _minimum_ highlight headroom specified by the ISO standard for minimum or base exposure index, so that the _base ISO speed_ is less than the lowest ISO exposure index setting. It is nothing to do with the camera giving more exposure than specified by the ISO exposure index setting. People who study those DXO measurements really need learn the differences between the various different measurements of sensor characteristics defined by ISO standard 12232, and avoid referring to them all as simply "ISO".

And that extra raw highlight headroom is why offering a new lower minimum ISO exposure setting makes sense: the ISO 200 exposure index setting has more than the minimum required headroom, and DXO's measurements show that the saturation based minimum sensitivity is well below ISO 200, so there is room for this new, lower exposure index setting: it simply reduces the (abundant) raw headroom, but still leaving more than the minimum highlight headroom required by the ISO standard.
« Last Edit: January 30, 2014, 09:04:11 PM by BJL »
Logged

TriadX1

  • Newbie
  • *
  • Offline Offline
  • Posts: 1
Re: Olympus E-M5 firmware 2.0: the new 'ISO LOW' is a real ISO160
« Reply #3 on: January 30, 2014, 09:26:14 PM »

I wonder why you cannot choose that as the default in the ISO-Auto Set?  Looks like the lowest you can set is still 200...
Logged

Telecaster

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3569
Re: Olympus E-M5 firmware 2.0: the new 'ISO LOW' is a real ISO160
« Reply #4 on: January 31, 2014, 03:55:04 PM »

I'd guess Olympus is assuming folks who want to use ISO LOW with the E-M5 will understand the need to be more careful with exposure. It's what you choose when you want to maximize image quality. When I use Auto-ISO I'm not aiming for ultimate IQ but rather optimal versatility.

I noticed early on with the E-M5 that it had plenty of RAW highlight headroom. Most of my early tripod-mounted pics with that camera were exposed ETTR + 1/3rd stop at ISO 200. I stopped doing that when I switched to RAW + JPEG shooting, for quick & easy iPad processing & sharing. But I will use the new ISO LOW if/when desired.

-Dave-
Logged

Vladimirovich

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1311
Re: Olympus E-M5 firmware 2.0: the new 'ISO LOW' is a real ISO160
« Reply #5 on: January 31, 2014, 06:27:37 PM »

It seems Olympus is now learning from other brands such as Fujifilm, offering its customers really useful firmware updates.
I am with Thom - Fuji's upgrades are because their cameras were released not fully executed... come on, how many firmware updates for AF speed  ;D  = http://www.dslrbodies.com/newsviews/how-myths-get-started.html  ...
Logged

Vladimirovich

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1311
Re: Olympus E-M5 firmware 2.0: the new 'ISO LOW' is a real ISO160
« Reply #6 on: January 31, 2014, 06:29:40 PM »

I thought that Olympus was already overstating the ISO in that camera by a factor of two.

you can see DxOMark graphs how nominal ISO compares with Ssat ISO that DxO likes... however neither Olympus nor others do not break any standards, they are perfectly in line w/ those that allow OOC JPG brightness to be used... so word "overstating" shall not be used
Logged

Vladimirovich

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1311
Re: Olympus E-M5 firmware 2.0: the new 'ISO LOW' is a real ISO160
« Reply #7 on: January 31, 2014, 06:32:23 PM »

I'd guess Olympus is assuming folks who want to use ISO LOW with the E-M5 will understand the need to be more careful with exposure.
this setting is for JPG shooters really, raw shooters can use whatever existing gain is the best for the needed exposure (aperture and exposure time) and raw processing in mind.
Logged

Vladimirovich

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1311
Re: Olympus E-M5 firmware 2.0: the new 'ISO LOW' is a real ISO160
« Reply #8 on: January 31, 2014, 06:44:45 PM »

Finally I wonder why camera makers never like to provide really low ISO settings in digital cameras that would be very useful to a good amount of users. I guess there is some technical reason behind and not just a marketing decision.


take Sony 16mp m43 sensor in GH3 and EM1 - there is a theory that to provide "LOW" ISO Panasonic does allow ADC not to clip well into non linear area near well saturation and Olympus does not (always clip well below well saturation) ... hence DxO sees Panasonic "LOW" ISO as real one and Olympus "LOW" ISO as just firmware trick with exposure/metering... so Panasonic wants to squeeze something extra for itself @ the expense of potential ill effects near saturation (that does not mean that GH3 gets better DR than EM1).
« Last Edit: January 31, 2014, 06:47:44 PM by Vladimirovich »
Logged

Pete Berry

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 431
Re: Olympus E-M5 firmware 2.0: the new 'ISO LOW' is a real ISO160
« Reply #9 on: January 31, 2014, 07:29:59 PM »

take Sony 16mp m43 sensor in GH3 and EM1 - there is a theory that to provide "LOW" ISO Panasonic does allow ADC not to clip well into non linear area near well saturation and Olympus does not (always clip well below well saturation) ... hence DxO sees Panasonic "LOW" ISO as real one and Olympus "LOW" ISO as just firmware trick with exposure/metering... so Panasonic wants to squeeze something extra for itself @ the expense of potential ill effects near saturation (that does not mean that GH3 gets better DR than EM1).

Having shot for year with the GH3, I can confirm ISO 100 acts perfectly normally, at twice the exposure, with possibly slightly more highlight headroom than at IS0 200 (certainly not less), but in RAW only. The JPG's for some reason have much more easily blown highlights than at 200 - as if Panny wanted to sabotage this "extended ISO" for the more casual user with a straight-line highlight JPG tone curve. The problem here for us RAW shooters is that metering is keyed to JPG's, and learning how much of an ISO 100 histogram highlight spike (or chimped blinkies) can be compensated at 100 is far from ISO 200 takes more bracketing.

PB
« Last Edit: January 31, 2014, 07:37:58 PM by Pete Berry »
Logged

Vladimirovich

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1311
Re: Olympus E-M5 firmware 2.0: the new 'ISO LOW' is a real ISO160
« Reply #10 on: January 31, 2014, 08:57:28 PM »

Having shot for year with the GH3, I can confirm ISO 100 acts perfectly normally, at twice the exposure, with possibly slightly more highlight headroom than at IS0 200 (certainly not less), but in RAW only.

I did test GH3 using software from William Claff ( http://home.comcast.net/~NikonD70/ )

here is what it is (averaged 3 series of shots) :



PS: sawtooth @ high ISO is because Panasonic also uses ISO-by-tag approach there (where certain gains are not just analog and/or digital but also just tagged in raw file instructing raw converter to add hidden expocorrection)

« Last Edit: January 31, 2014, 08:59:38 PM by Vladimirovich »
Logged

OldRoy

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 470
    • http://
Focus box resizing? Mine isn't.
« Reply #11 on: February 01, 2014, 09:04:08 AM »

Installed this update successfully. "Low" ISO setting now available but the focus box size (and the grey graticule it inhabits) remain at the same size as previously. Now I can use one of the Fn buttons to get a smaller box - as I always could, at the expense of having the histogram vanish (the sort of oddity that I simply detest about this firmware) - however as the current seetings have been retained nothing seems to have changed. Can someone enlighten me as to why?

Incidentally I noticed that the updater's dialog contains the phrase "Unable to install updates when more than one lens is attached...". I there something else we should be told?
Roy
Got it. SCP/info/ arrow keys.
« Last Edit: February 01, 2014, 09:08:28 AM by OldRoy »
Logged

bcooter

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1505
  • me
    • working
Re: Olympus E-M5 firmware 2.0: the new 'ISO LOW' is a real ISO160
« Reply #12 on: February 01, 2014, 03:37:13 PM »



I noticed early on with the E-M5 that it had plenty of RAW highlight headroom.

I had a shot a month ago, where the subject was in front of a double glass door.  Shot it with the em-5 and hit the end of card and they handed me my 1dx. 

In the computer the base exposures looked very close, except the 1dx wouldn't hold any detail in the window.  (I shot a background plate just in case).

It was dead white.

When I got the the em-5 image it also was white, but in lightroom moved the light slider over and it pulled up outside detail, so I went back to the 1dx file to see if it would and nope, still dead white.

Really don't get it but that crazy little em-5 just shoots so well and it was like lightroom was made for it.



IMO

BC

Guillermo Luijk

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1765
    • http://www.guillermoluijk.com
Re: Olympus E-M5 firmware 2.0: the new 'ISO LOW' is a real ISO160
« Reply #13 on: February 01, 2014, 03:39:10 PM »

AMAZING!!!

Doing some SNR comparisions between the ISOLOW and the ISO200, I have realized I made a mistake when analyzing the RAW files to determine the relative exposure: the gap of 1/3EV is correct but... ISOLOW is the file with higher RAW values!. I thought I made a mistake when renaming the files and ignored that ISOLOW had more exposure because I thought that was not possible.

I.e. when you shoot using the same aperture and shutter at ISOLOW, you get RAW data 1/3EV more exposed (confirmed) than at ISO200. That means that if you use camera's metering at ISOLOW (that seems to work as an ISO100 for metering), RAW files will blow up to 1,33EV of highlight information vs correct exposure using any other ISO. This matches a reported video on other forum with more blown highlights when using ISOLOW than when using any other ISO value.

If that were tue, the new ISOLOW would be an effective ISO250 (250=200*1.25), not ISO160 nor ISO100.

Could this be a mistake by Olympus or I am missing something?

I am requesting RAW files with blown highlights at ISOLOW and ISO200 to confirm, keeping the same aperture/shutter.
If anyone wants to do some analusis, these 2 RAW files were shot using the same aperture/shutter but different ISO setting:

isolow.ORF
iso200.ORF

Unfortunately they have no blown highlights.

Regards
« Last Edit: February 01, 2014, 04:01:50 PM by Guillermo Luijk »
Logged

Guillermo Luijk

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1765
    • http://www.guillermoluijk.com
Re: Olympus E-M5 firmware 2.0: the new 'ISO LOW' is a real ISO160
« Reply #14 on: February 01, 2014, 04:13:34 PM »

I think I now Know what's going on here: the LOWISO is the opposite concept of Canon HTP to enhance dynamic range. HTP means capturing at a lower ISO than the user setting and metering, to preserve extra highlight information. Then the camera engine or RAW developer would push exposure in the shadows while preserving the highlights. The cost is some extra noise in the shadows.

Olympus LOWISO is capturing at a higher ISO (ISO250) than user setting (ISO100), to improve SNR in the shadows. Then the camera engine or RAW developer would bring exposure down and SNR in the shadows will have improved. It is basically the ETTR principle. The cost is more blown highlights. I think this makes sense given that Olympus cameras have the highest highlight headroom of all brands.

So LOWISO would just be a SNR enhancer through ETTR, at the cost of possibly having some more blown highlights.

Regards.
« Last Edit: February 01, 2014, 04:28:04 PM by Guillermo Luijk »
Logged

Vladimirovich

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1311
Re: Olympus E-M5 firmware 2.0: the new 'ISO LOW' is a real ISO160
« Reply #15 on: February 01, 2014, 04:29:59 PM »

E-M1 has the difference between ISO200 and LOWISO within 1/6 EV or less...
Logged

Guillermo Luijk

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1765
    • http://www.guillermoluijk.com
Re: Olympus E-M5 firmware 2.0: the new 'ISO LOW' is a real ISO160
« Reply #16 on: February 01, 2014, 05:06:47 PM »

E-M1 has the difference between ISO200 and LOWISO within 1/6 EV or less...

Yes, the point here is not so much that gap, but the different amount of light. If you meter for ISO100 (ISOLOW) but then capture the RAW data at ISO250, you are doubling the amount of light that you would have obtained metering at ISO200, and that necessarily is going to improve SNR in the shadows vs shooting at ISO200. But this is just a trick for JPEG shooters; the max DR achievable in RAW is lower at LOWISO than at ISO200.
« Last Edit: February 01, 2014, 05:08:26 PM by Guillermo Luijk »
Logged

bcooter

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1505
  • me
    • working
Re: Olympus E-M5 firmware 2.0: the new 'ISO LOW' is a real ISO160
« Reply #17 on: February 01, 2014, 05:10:50 PM »

Now, I'm not trying to prove anything, just show results.

These images are straight out of lightroom, no matting in, nothing but slider processing.

One the em-5 image I will pull the exposure down and put more detail into the window because there is a lot more there, with the 1dx file, I will have to use a separate background plate I shot because the doors are devoid of detail.

I still find these little olympus cameras amazing.



BC
« Last Edit: February 01, 2014, 05:13:03 PM by bcooter »
Logged

Vladimirovich

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1311
Re: Olympus E-M5 firmware 2.0: the new 'ISO LOW' is a real ISO160
« Reply #18 on: February 01, 2014, 05:22:26 PM »

Yes, the point here is not so much that gap
well, 1/3 EV gap for E-M5 is interesting by itself... do they intentionally apply 1/3EV more gain postexposure to improve S/N in deep shadows  ?
Logged

Vladimirovich

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1311
Re: Olympus E-M5 firmware 2.0: the new 'ISO LOW' is a real ISO160
« Reply #19 on: February 01, 2014, 05:24:00 PM »

These images are straight out of lightroom, no matting in, nothing but slider processing.
ACR/LR will apply different hidden expocorrections for ISO LOW vs the rest - at least for E-M1... I bet the code is not yet update for E-M5 (unless they were thinking or informed by Olympus in advance)... conversion of ORF to DNG using ACR/LR/Adobe DNG converter and checking the relevan DNG tag shall tell.

PS: with the current ACR 8.3 (LR 5.3) Adobe does the following with .ORF for E-M1 :

1) for ISO100 (ISO LOW) it applies (hardcoded) -0.84 EV expocorrection (tag "BaselineExposure")
2) above ISO100 (ISO LOW) it applies (hardcoded) +0.32 EV expocorrection (tag "BaselineExposure")
3) current .dcp profile (Adobe Standard) has no expocorrection set (tag "BaselineExposureOffset" = 0.00) to be applied in addition to "BaselineExposure"
« Last Edit: February 01, 2014, 05:26:12 PM by Vladimirovich »
Logged
Pages: [1] 2   Go Up