GH1-GH3 compared at ISO 200

Just for the fun of it, I compared the whole Panasonic GH-series of cameras at the same shutter speed/aperture and ISO 200. My conclusion from looking at the histograms of the out of camera JPEGs, and the RAW images converted with Lightroom, is that the GH1 is a bit more sensitive at ISO 200, while the GH2 and GH3 are quite similar.

More details and images here:

http://m43photo.blogspot.com/2012/12/gh1-gh2-and-gh3-iso-200.html

Of course, there might be a source of error here that Lightroom treats the RAW files differently, even if I select no exposure compensation.

yes, it does

In order to say something about sensitivity on the RAW level (how much exposure it takes to reach the clipping point in RAW at this or that camera ISO) you need to use a program that lets you see the RAW data levels, e.g., RawDigger.

I did that for the GH3 versus the E-M5 here

http://forums.dpreview.com/forums/thread/3348241

and found, to my surprise, that the saturation level is approximately the same (actually a tad higher for the E-M5) so that on both cameras, a camera ISO of 200 corresponds to a DxO "measured ISO" of approximately 100. On the GH2, by contrast, a camera ISO of 200 corresponds to a DxO "measured ISO" of approximately 200. So for a RAW shooter, the base ISO of the GH3 is significantly lower than that of the GH2.

Ok, I did use RawDigger now. I made an old school HTML table about half way down in this article:

http://m43photo.blogspot.com/2012/12/gh1-gh2-and-gh3-iso-200.html

I took the average values from the wall part of the image exclusively, to avoid the possible bias of different framing.

I think it still looks like the sensitivity of the GH2 and GH3 are quite close. Or am I interpreting the results wrongly?

How did you handle the black-level offset? When you open a GH2 file in the current version of RawDigger it will set the proper-level offset automatically. I would guess it would do that on the GH1 too. On the GH3, by contrast, it won't since the camera is too new for RawDigger to be fully aware of it (unless the file open is a DNG rather than an RW2). So you have to go into Preferences on the File menu, select Data Processing, change Black Level from Auto to Manual and enter the proper value. For ISO 200, that value is 144.

If 144 has in fact to be subtracted from your GH3 RAW levels, that would place it below the GH2 but by a much smaller margin than the one I found (about one EV difference in my case, about 1/3 EV in yours). So things would remain a bit mysterious.

The difference between the GH1 and the GH2 is pretty much in line with expectations. The DxO "measured ISO" of the GH2 is approximately the same as the camera ISO whereas the "measured ISO" of the GH1 is about 1/3 EV higher than the camera ISO (e.g., about 260 at a camera ISO of 200).

No, I did not change the offset.  Setting it to 144, as you say, will deduct 144 from the values in my table.

OK. That's part of the explanation. But it's still strange that your and my results are so different. One further question: Did you happen to have shading compensation turned on for the GH3? I don't really suspect that based on what your images look like but I thought I had better ask anyway. This option is about the only one I can think of that actually changes the RAW data values (at least it does on the GX1 and the E-M5) and since you shot the 20 at 1.7 (where it has quite a bit of vignetting) the effect may be noticeable.

No, I did not use any shadow compensation. I prefer to have as clean as possible in camera processing, and rather do changes to the images later if needed

Figured so. But that means the discrepancy between our results remains a bit mysterious. One additional (although perhaps far-fetched) possibility is that the GH3 plays around with RAW level amplification based on camera settings other than ISO alone. Using ExifTools to inspect the complete EXIF information in the GH3 and E-M5 RAWs I am comparing, I found a setting named GainControl, appearing just before Contrast, Saturation, and Sharpness, which in the GH3 files has the value "Low gain up" but in the E-M5 files has the value "None".

Is this set similarly in your case and do you or anyone else know what this setting is about? Apparently, it is not what Panasonic refers to as "Intelligent dynamic range control". I can see in the EXIF data that the parameter IntelligentD-range is set to off.

the GH1 is the best bang for buck in the m43 line. period.

From some various GH2 images inspected, and from a set of dark-shots at all ISO settings between ISO=160 and ISO=1600: ISO=160 and ISO=200 show "Low Gain Up" in that Exif tag; and ISO=250 (and all settings above up to ISO=1600) show "High Gain Up".

Nothing special or exotic in any of my GH2 on-camera control-settings in any of the above cases.

Looking through a bunch of GH3 images downloaded from the French site "Hybridcams", ISO=125 and ISO=200 show "Low Gain Up", and the higher valued ISO settings (including 400,800,1600,3200,6400,12800, and 25600) all show "High Gain Up".

"None" is indicated in the field of the same Exif tag labelled "Gain Control" (EM5 @ ISO=200).

Thanks DM. That helps at least a little bit. But do you have any idea of whether this parameter is of any importance when it comes to solving the mysterious discrepancy between my comparisons and those of Fredrik? All the E-M5 and GH3 image pairs that I have at my disposal tell the same story. The "measured ISO" of these two cameras is about the same for the same camera ISO (200), if anything a tad lower for the GH3. This in turn should imply a difference of about one EV between the saturation level of the GH2 and the GH3 when these two are given the same exposure at the same ISO. Yet, Fredrik's RawDigger results seem to indicate a difference of no more than about 1/3 EV.

Two things occur to me.

(1) We know that LR 4.3 is attenuating the GH2 RAW image-data by 0.5 EV (which we know is not the case where it comes to the GH3 RAW image-data).

(2) The table that remains presently posted on the web-page:

http://m43photo.blogspot.com/2012/12/gh1-gh2-and-gh3-iso-200.html

... is stated to not reflect a black-level offset of 144 being applied to the GH3 RAW image-data in RawDigger - which will cause the Average statistics for the image-data to decrease significantly.

Yes, and these values indicates that the truth would be somewhere between Fredrik's and my results. On the other hand, as I told you in the other thread, the baseline value of 0 for the GH3 may be too low relative to the +0.5 of the E-M5. When looking at the files in LR, I have to reduce exposure on the E-M5 by half a stop (or increase it on the GH3 by the same amount) to reach roughly the same brightness.

Yes. But even with the correct black-level value subtracted, the GH2 saturation is only about 1/3 EV higher than that of the GH3 based on those figures. That's what puzzles me. The difference expected from my vantage point is about one EV.

I thought that your information was derived from RawDigger analysis only, right ?

Using Lightroom 4.x, right ? (If somehow referenced from sensor saturation levels, and not from equal exposure), that observation would seem to indicate that the GH3 and EM5 have similar "Saturation ISOs", as you have suggested ?

While the OP has talked (in generalaties) on this thread about the resulting RawDigger "Average" vales after applying balck-level subtraction (of 144) to the GH2 RW2 image-data, the table presently published on the OP's blog (and the only numbers that we have to calculate with) are cleary stated in that blog-post to be the results found without any black-level subtraction ...

As I read Anders' reply, he understands that my table is done without changing the black level in the RawDigger preferences, and if I had applied a black level of 144, that would have deducted 144 from my table in the GH3 column.

I may be wrong, though.

Yes, I just tried it with a GH3 RW2 in RawDigger, and you are correct that the per-channel Average statistics shift downward in numerical value by approximately 144.

Interesting technical read this thread, but here's the bottom line - I have had every Panasonic G series cam apart from the G2 and shot extensively with all of them - you can spend as many hours analysing raw digger data as you like, but in the field, let me tell you that the GH3 is hands-down the best Panasonic G camera so far. The high ISO, regardless of what any technical analysis with a myriad raw tools may tell you in terms of numeric scores, is the best of any G series cam and as good as the OM-D. In fact, it is better than my Sony SLT-a57 whcih supposedly has the 16MP Sonty wonder sensor. The low ISO images are crisp as anyone would want with good DR - seems to be quite a weak AA filter on this camera (hence the moire issues in video).

By all means continue this interesting technical analysis but I would urge any G cam owners to try a GH3 if it is within your affordable price range - it's a very good camera indeed.

Great ! If you shoot RAW, RawDigger analysis is a very valuable tool in order to develop and to check any techniques that users may want to use of maximizing recorded RAW channel ADU levels, along with also maximizing sensor-level Exposure (by using minimum F-Numbers and Shutter Speeds). Such voluntary activities in no way detract from the practice of, or the joys of, photography ...

Anders, perhaps I missed some things that you said about your methods (on the other related thread). Fredrik is using Equal Exposure conditions - not equal Saturation conditions. Unless I have missed it, I do not see any indication that sensor Saturation Exposure levels were determined:

http://m43photo.blogspot.com/2012/12/gh1-gh2-and-gh3-iso-200.html

The "Saturation ISO" is a function of a camera's Saturation Exposure value only.

Ssat = 78/H where Hsat is the saturation exposure in lux-s. The saturation speed only shows the saturation exposure as a result.

Similar to the situation when comparing the E-M5 to the GH2, a difference in the "Saturation ISOs" would be a result of a significantly higher Full Well Capacity in the case of the GH3 - which (it seems to me) can only be assessed by determining the ratio of the Saturation Exposures of the GH3 relative to the GH2. How does an Equal Exposure test determine differences in the Saturation Exposures, and thus the "Saturation ISOs", of the two cameras ?

I am probably missing or forgetting something important ? What would that be ? Please advise.