Olympus E-PM2 Actual ISO vs IS0 set in camera?

I am sorry to start the well beaten subject...

I have two 4/3 cameras: Olympus E-PM2 and Panasonic GX1. On ISO graph by dxomark.com for these two cameras the E-PM2 actual ISO is about half of what it set in camera (for ISO set to1600 the actual ISO is 800), while the actual and set in camera ISO for GX1 are equal. All the above measurements apply to RAW settings as DXO test the sensor.

As I have two cameras one, which perfectly match the set and actual ISO (GX1) and one with the actual ISO only half of the set ISO, I decided to test if there is any difference in the exposure between these two camera.

I have set the cameras to the same ISO, aperture, expose compensation and exposure metering and photographed the xRite White Card filling the viewfinder. The two cameras showed identical shutter speed. I have opened the images in the Lightroom and the RGB readings for both images were the same within few percent points difference.

I do trust DXO, however … my test does not follows the DXO statement.

Leo

sigala1 wrote:

Leo wrote:

I am sorry to start the well beaten subject...

I have two 4/3 cameras: Olympus E-PM2 and Panasonic GX1. On ISO graph by dxomark.com for these two cameras the E-PM2 actual ISO is about half of what it set in camera (for ISO set to1600 the actual ISO is 800), while the actual and set in camera ISO for GX1 are equal. All the above measurements apply to RAW settings as DXO test the sensor.

As I have two cameras one, which perfectly match the set and actual ISO (GX1) and one with the actual ISO only half of the set ISO, I decided to test if there is any difference in the exposure between these two camera.

I have set the cameras to the same ISO, aperture, expose compensation and exposure metering and photographed the xRite White Card filling the viewfinder. The two cameras showed identical shutter speed. I have opened the images in the Lightroom and the RGB readings for both images were the same within few percent points difference.

I do trust DXO, however … my test does not follows the DXO statement.

Leo

All cameras should give you the same JPEG exposure at the same ISO.

However, Olympus ISO 200 is really ISO 100 underexposed by a stop, which Olympus has done intentionally to make sure that the photos don't have blown-out highlights.

Hi sigala1,

Thank you for reply. I was expecting different exposures, however exposures were the same. I have run the test as now thinking to purchase Olympus M1.

I photograph only in RAW format. My all three cameras, the two mentioned in the post and Nikon D7000, are permanently set to RAW. I use D7000 mostly for photographing art, like paintings, for artists because D7000 allows 14 bits A/D conversion. This camera profiled for different ISO and lighting conditions. My experiment in the post performed with cameras set to RAW. Imaged were opened in Lightroom.

Leo

sigala1 wrote:

Leo wrote:

I am sorry to start the well beaten subject...

I have two 4/3 cameras: Olympus E-PM2 and Panasonic GX1. On ISO graph by dxomark.com for these two cameras the E-PM2 actual ISO is about half of what it set in camera (for ISO set to1600 the actual ISO is 800), while the actual and set in camera ISO for GX1 are equal. All the above measurements apply to RAW settings as DXO test the sensor.

As I have two cameras one, which perfectly match the set and actual ISO (GX1) and one with the actual ISO only half of the set ISO, I decided to test if there is any difference in the exposure between these two camera.

I have set the cameras to the same ISO, aperture, expose compensation and exposure metering and photographed the xRite White Card filling the viewfinder. The two cameras showed identical shutter speed. I have opened the images in the Lightroom and the RGB readings for both images were the same within few percent points difference.

I do trust DXO, however … my test does not follows the DXO statement.

Leo

All cameras should give you the same JPEG exposure at the same ISO.

However, Olympus ISO 200 is really ISO 100 underexposed by a stop, which Olympus has done intentionally to make sure that the photos don't have blown-out highlights.

Hi sigala1,

I am so glad you have replied for this post also, as now i you may see my replays for sure and help to understand the problem. If Olympus cooks the RAW file then probably this adjustment can be considered as a part of the sensor activity as from the outside that is all how I can interpret it. I have checked the noise level and resolution (just evaluative) at ISO 3200 and found that PM2 is about the same on the noise and resolution matters as my D7000 at D/A set to 12 bits. Also D7000 with the lenses is too heavy and big. I am on the older side  Copy of my 4/3 reply is below.

Thank you for reply. I was expecting different exposures, however exposures were the same. I have run the test as now thinking to purchase Olympus M1.

I photograph only in RAW format. My all three cameras, the two mentioned in the post and Nikon D7000, are permanently set to RAW. I use D7000 mostly for photographing art, like paintings, for artists because D7000 allows 14 bits A/D conversion. This camera profiled for different ISO and lighting conditions. My experiment in the post performed with cameras set to RAW. Imaged were opened in Lightroom.

Leo

Leo wrote:

I am sorry to start the well beaten subject...

I have two 4/3 cameras: Olympus E-PM2 and Panasonic GX1. On ISO graph by dxomark.com for these two cameras the E-PM2 actual ISO is about half of what it set in camera (for ISO set to1600 the actual ISO is 800), while the actual and set in camera ISO for GX1 are equal. All the above measurements apply to RAW settings as DXO test the sensor.

As I have two cameras one, which perfectly match the set and actual ISO (GX1) and one with the actual ISO only half of the set ISO, I decided to test if there is any difference in the exposure between these two camera.

I have set the cameras to the same ISO, aperture, expose compensation and exposure metering and photographed the xRite White Card filling the viewfinder. The two cameras showed identical shutter speed. I have opened the images in the Lightroom and the RGB readings for both images were the same within few percent points difference.

I do trust DXO, however … my test does not follows the DXO statement.

See my reply to you in this other thread.

http://www.dpreview.com/forums/post/54220550

The in-camera ISO and the DxO "measured ISO" are simply two entirely different things.

The in-camera ISO is a measure of how much light it takes to achieve a certain brightness in the OOC jpegs. It also regulates the metering. As a rule, this ISO should be the same for all cameras since it corresponds to a norm issued by ISO that the manufacturers have agreed to follow. How well a camera conforms to this norm is tested by DPR and reported in the "noise and noise reduction" setting of each review. Very few cameras deviate significantly from the norm. One exception, however, are the Fuji X cameras, where the OOC jpegs come out about half a stop too dark, and which are therefore unduly favored (by given more exposure) in the DPR studio scene samples.

The DxO "measured ISO" is simply a measure of how much light it takes to reach the clipping point in the RAW file at a certain camera ISO. The more the camera ISO exceeds the "measured ISO", the more highlight headroom the camera reserves. The manufacturers are free to set the clipping point in the RAW file as they see fit and there is no cheating involved in setting it this rather than that way.

The ISO values correspond to levels of brightness but the standard does not dictate how to achieve these values in terms of analog gain vs. digital gain.

Panasonic cameras rely mostly on analog gain while Olympus cameras rely on both.

In your case, the GX-1 reaches the brightness of ISO 200 (for example) mostly (if not solely) by analog gain. The E-PM2 reaches the brightness of ISO 200 by less analog gain (50% of the GX-1) and then some digital gain that brightens the image. So you end up with the same brightness, but by two different approaches underlyingly.

You may find this interesting.

In a digital system, ISO is a nearly useless kludge, a way to adapt a concept of film chemistry where it really doesn't belong. In film chemistry, ISO and ASA actually referred to how much light the photosensitive emulsion had to be exposed to in order to achieve a given density. But in an electronic system, there is no sensitivity other than the sensor's base sensitivity, and there is no intrinsic brightness to the data. There is no density in the way a film negative can have a measurable density. The signal isn't anything until a computer, translating the recorded signal into something human-readable, assigns a pixel saturation level an RGB number. This is done in one of five methods as specified in ISO 12232:2006, and as employed in cameras, only applies to resultant JPEGs in the sRGB color space (so for those of us who shoot RAW and edit in ProPhotoRGB, the concept is already failing us).

The only real reason we still have ISO in cameras is because we got used to it over a century and a half of chemistry-based photography, and we still haven't gotten over ourselves despite the fact that everything has changed. Electronic signal capture has almost nothing in common with recording of a latent image on a chemical medium.

ISO's only real point of being handy is the way CMOS sensors generally have reduced read noise at higher ISO settings, and thus slightly better signal-to-noise ratio at those higher ISO settings.

As others have mentioned, the method used by DxO has nothing at all to do with the standards laid out by the International Organization for Standardization.

It would help if ISO themselves put some input into the issue, such that correct exposure is predictable, given an aperture and shutter speed, from one digital camera to tbe next.  Sensitivity has been abused too long.

As for what Olymous is doing, I'm happier with it since they moved to the Sony sensor.  With tbe 12mp sensor, I was able to get good (if less predictable) results with ETTR and experimenting with the raw files.

tinternaut wrote:

It would help if ISO themselves put some input into the issue, such that correct exposure is predictable, given an aperture and shutter speed, from one digital camera to tbe next. Sensitivity has been abused too long.

Well, correct exposure is in general predictable based on camera ISO as long as you define correct exposure in the traditional way. If you meter for middle gray, middle gray will come out right.

However, in the digital age, for those of us who shoot RAW and aim for minimal noise but with those highlights preserved that we care to preserve, metering for middle gray is inferior to metering for the highlights. An efficient way to do that with Oly cameras is by means of the live view blinkies. An alternative available on any modern MFT camera is the live view histogram. A third way is to use the spot meter, meter on the highlights, and then dial in appropriate EC based on that highlight reading. What is appropriate here depends on the DxO "measured ISO". On my old G1, whose "measured ISO" is about 1/3 EV higher than the camera ISO, I would dial in about two stops. On my E-M5, whose "measured ISO" is about one EV lower than the camera ISO, I would dial in 3.3 stops.

As for what Olymous is doing, I'm happier with it since they moved to the Sony sensor. With tbe 12mp sensor, I was able to get good (if less predictable) results with ETTR and experimenting with the raw files.

Thank you to all for helping. The other benefit: after reading all replies I have check my recently taken photo and YES I am underexposing most of them. I am ETTR using the displayed histogram, however the histogram is related to JPG. The histograms for the most of my photos are approximately from -0.5EV to -1EV below 255. I have found several images that before correction were overexposed and that there are enough room to adjust them down -1EV to -1.5EV without loosing the highlits. No wonder that some of my images are noisy in the low light areas . However, intentionally overexposing may cause overexposure beyond the dynamic range reserve. How you manage this problem?

Leo

Leo wrote:

Thank you to all for helping. The other benefit: after reading all replies I have check my recently taken photo and YES I am underexposing most of them. I am ETTR using the displayed histogram, however the histogram is related to JPG. The histograms for the most of my photos are approximately from -0.5EV to -1EV below 255. I have found several images that before correction were overexposed and that there are enough room to adjust them down -1EV to -1.5EV without loosing the highlits. No wonder that some of my images are noisy in the low light areas . However, intentionally overexposing may cause overexposure beyond the dynamic range reserve. How you manage this problem?

Yes, the live-view as well as post-capture histogram is based on the OOC jpeg. However, the clipping point in the OOC jpeg is as a rule very close to the clipping point in the RAW file so that shouldn't make much of a difference. To avoid exposing beyond the clipping point of the RAW file, just use the histogram or, on the E-PM2, the live-view blinkies (highlight warnings) which have the advantage over the histogram of showing not only that, but also where, the highligts are clipping.

sigala1 wrote:

Leo wrote:

Thank you to all for helping. The other benefit: after reading all replies I have check my recently taken photo and YES I am underexposing most of them. I am ETTR using the displayed histogram, however the histogram is related to JPG. The histograms for the most of my photos are approximately from -0.5EV to -1EV below 255. I have found several images that before correction were overexposed and that there are enough room to adjust them down -1EV to -1.5EV without loosing the highlits. No wonder that some of my images are noisy in the low light areas . However, intentionally overexposing may cause overexposure beyond the dynamic range reserve. How you manage this problem?

Leo

At the exposure when you first see an orange blinkie, you can increase the exposure by at least another 2/3 EV without actually clipping the blinkie in the raw file.

No. If you test the matter by means of RawDigger (as I have), you will find that you can't go more than 1/3 EV above what the live-view blinkies indicate without significant clipping in the RAW in at least one channel (usually green). So there's very little difference between what the blinkies tell you and the actual clipping point in RAW. This is with the warning level set to 255 which for obvious reasons is the value that makes most sense.

The existence of the new ISO 100 mode should have put us on alert for the truth of the above.

The above holds for ISO 100 too.

I have photographed xRite color checker 24 card with seven exposures stepping by +1/3EV:

0EV, +1/3EV, +2/3EV, +1.0EV. +1 1/3EV, +1 2/3EV, + 2.0EV

Then I have opened the images in LR and with expose slider adjusted the brightest Gray to 94.5 each image. Then I have verified the same color patch for all images (yellow). Its R, G and B reading were about the same up to +1.7EV. At the +2.0EV I had to adjust exposure control more negative. The adjustments were exactly:

0EV, -1/3EV, -2/3EV, -1.0EV. -1 1/3EV, -1 2/3EV.

The yellow patch R,G,B numbers were read as 90.7, 87.7, 37.6

The brightest gray patch showed that it is overexposed (red). At exposure -2.0EV the this patch reading were 97. 97, 97. I had to adjust exposure slider to -2.5EV to make the same reading = 94.5, however the whole image became little darker and the R, G and B for the yellow patch changed to: 87.7, 84.1 31.0. The overexposure red marking disappeared.

My conclusions from this test are:

1) I am underexposing routinely approximately -1.7EV.

2) The underexposure marking (red) is a good indicator and it well correlated with the RAW format.

Any suggestions?

Leo

Leo wrote:

I have photographed xRite color checker 24 card with seven exposures stepping by +1/3EV:

0EV, +1/3EV, +2/3EV, +1.0EV. +1 1/3EV, +1 2/3EV, + 2.0EV

Then I have opened the images in LR and with expose slider adjusted the brightest Gray to 94.5 each image. Then I have verified the same color patch for all images (yellow). Its R, G and B reading were about the same up to +1.7EV. At the +2.0EV I had to adjust exposure control more negative. The adjustments were exactly:

0EV, -1/3EV, -2/3EV, -1.0EV. -1 1/3EV, -1 2/3EV.

The yellow patch R,G,B numbers were read as 90.7, 87.7, 37.6

The brightest gray patch showed that it is overexposed (red). At exposure -2.0EV the this patch reading were 97. 97, 97. I had to adjust exposure slider to -2.5EV to make the same reading = 94.5, however the whole image became little darker and the R, G and B for the yellow patch changed to: 87.7, 84.1 31.0. The overexposure red marking disappeared.

My conclusions from this test are:

1) I am underexposing routinely approximately -1.7EV.

2) The underexposure marking (red) is a good indicator and it well correlated with the RAW format.

Any suggestions?

A better way of verifying how you actually manage with exposure in this or that scenario is to use RawDigger, which immediately lets you see how far you got from the clipping point in each channel.

My practical advice is the same as before. Use the live-view highlight "blinkies" (with warning level set to 255). Dial in more exposure in steps of 1/3 EV until the first blinkies turn on in areas you want to preserve. This will work well in most cases but you may have to be a bit cautious in very cold or warm light or with very cold/warm subjects. Normally, the green channel is the first to clip but in these scenarios it might be red or blue in which case the blinkies might not give you proper warning. Note also that there may be scenes with such high DR that you are forced to give up the general strategy I propose (to expose such that the brightest highlights you care to preserve are at but not beyond the clipping point) since shadow noise becomes unbearable otherwise. In these cases, I usually bracket exposure and then select the best compromise at the PP stage or merge/align multiple exposures.

More about exposure in the post to which I link below, including some information on scenarios where the blinkies might not work as they should and how to handle that. This is for the E-M5 but everything probably applies to the E-PM2 as well since the sensor as well as camera generation is the same.

http://www.dpreview.com/forums/post/51169217

Anders,

I am in the same initial confused state after comparing Oly PM2 with Pana GX1. Based on DXO ISO test and displayed results: the PM2 actual ISO is 50% of the set ISO and the GX1 actual ISO = to the Set ISO.

I have photographed the same subject with both cameras set exactly the same (exposure meter, aperture, compensation, lens) at 0EV compensation and the shutter speed was the same for both cameras. Then have increase compensation for each camera until highlights clipping just started. For both cameras it started at +2EV. I was expecting that GX1 would start clipping 1EV early. I still missing the understanding of actual application of the DXO ISO test.

Pleas help to fill the void

Leo

Hi Henry,

Thank you for the link. However, I have set camera at the lowest ISO (sensor base sensitivity).

Leo

Hi Anders,

I finale got it :-). Thank you for the help and patience.

Leo