GM1 Electronic Shutter lower DR (GH4 too?)

Discussion in 'Panasonic Cameras' started by kwalsh, Mar 13, 2014.

  1. kwalsh

    kwalsh Mu-43 Top Veteran

    775
    Mar 3, 2012
    Baltimore, MD
    I wrote a post over on DPR about the GM1 electronic shutter resulting in lower dynamic range, it is only outputting 10 bits RAW instead of 12 bits.

    Thought I'd mention it over here, if you need the most DR at base ISO don't use the electronic shutter if possible on the GM1 (e.g. pushing shadows a lot). Likely not a major issue in normal day to day photography.

    It is expected the GH4 may have the same behavior as its sensor datasheet indicates 10 bit only when doing the fast electronic shutter and they are advertising the camera as using the fast mode like the GM1 does. Sample pictures and longer description in the DPR post:

    http://www.dpreview.com/forums/post/53300241

    Please read that post before asking questions here.
     
    • Like Like x 4
  2. val

    val Mu-43 Top Veteran

    548
    Dec 19, 2013
    Australia
    William
    Definitely an eye opener, I use electronic shutter a lot for low-light scenes on my GX7 yet haven't noticed a difference =/ hmm time to pixel peep
     
  3. kwalsh

    kwalsh Mu-43 Top Veteran

    775
    Mar 3, 2012
    Baltimore, MD
    Let us know what you find with the GX7 if you give it a test. There were two posts from other GX7 users at DPR that seemed to show it probably isn't affected the same way as the GM1. Someone measured the scan/read time and it is slower than the GM1 and another person was vague but I think did the same push the shadows test and didn't see a difference between the electronic and mechanical shutters on the GX7.
     
  4. FlyPenFly

    FlyPenFly Mu-43 Veteran

    448
    Feb 15, 2011
    So any shutter speed above 500 gets this hit?
     
  5. dhazeghi

    dhazeghi Mu-43 Hall of Famer

    Aug 6, 2010
    San Jose, CA
    Dara
    Thanks for posting this. Interesting to note. I did a casual test on the GX7 and didn't see a visible difference in shadow DR, but I should probably try something more controlled. Will RawDigger give a good histogram of the values?
     
  6. meyerweb

    meyerweb Mu-43 Hall of Famer

    Sep 5, 2011
    On the other hand, if you just like to take pictures and make photographs, rather than shoot brick walls and do the most extreme pushing possible, then look for flaws at 100%, you'll probably never notice the difference.

    Why don't you post the full image, in addition to the 100% crops, so we can see what real-life results look like?
     
  7. kwalsh

    kwalsh Mu-43 Top Veteran

    775
    Mar 3, 2012
    Baltimore, MD
    Well as I noted in the original DPR post this only matters in the rare case you push shadows a lot in base ISO. Not a common situation, but people do in fact do it. Landscape photographers especially (think sunset/sunrise with pushed foreground). This is in fact one of the reasons people were so excited about the E-M5 sensor and to a lesser degree the GH2 sensor before it - they were the first sensors that had very good base ISO DR. Now with the GM1 (and quite likely the GH4) you'll throw that improvement away if you use the electronic shutter. If you don't need to push the shadows then you don't care, if you do then best to know.

    I can tell you one landscape photographer on another forum who has been using m43 cameras for a few years found this a revelation - he had in fact noticed the loss of base ISO DR with his GM1 compared to his other m43 cameras and didn't know why. After my post he realized he had defaulted to the E-shutter and that was the difference. So yes he noticed it in his real photography from the start.

    Why, well not to put too fine a point on it but because I was fairly certain anyone who actually cared about this would understand exactly what was at play and how it would or wouldn't affect them. And if they were concerned they would take the few minutes to test it themselves to see if it was an issue for them. I'm not at your beckon call to provide anything, be glad I posted anything making people aware of something that wasn't known before.

    It is tiresome when posters swoop into a discussion and throw around the standard "not important to me so not important to anyone", or "not a problem if you expose correctly", or "prove it to me because I'm too lazy to follow up on my own" and so forth.
     
  8. kwalsh

    kwalsh Mu-43 Top Veteran

    775
    Mar 3, 2012
    Baltimore, MD
    On the GM1 yes, above 1/500 only electronic shutter. Also there seem to be some other modes that default to E shutter - I think a manual lens (no electrical contacts so adapted or body cap) also use the E shutter unless you use the flash (no idea why, just the way it seems to act).

    Again, it isn't a particularly awful hit - DR probably still better than the old 12MP sensors we shot with - but it is worse than say the E-M5/GX7 and so forth.
     
  9. kwalsh

    kwalsh Mu-43 Top Veteran

    775
    Mar 3, 2012
    Baltimore, MD
    I haven't played with RawDigger and so I don't know how well it would show the "holes" in the histogram from 10-bit mode. And because of the way RAW encoding occurs (it is actually very slightly lossy in Panasonic RAW files) sometimes trying to look at individual bins in the RAW histogram is confusing. I'd just shoot a underexposed shadow at base ISO using both shutters, push and compare. It is what the processed photo actually looks like that matters after all. And you'll know how much you might ever realistically push something in your own shooting to judge if any difference you detect is actually at all relevant to you.
     
  10. T N Args

    T N Args Agent Photocateur

    Dec 3, 2013
    Adelaide, Australia
    call me Arg
    How can you possibly tell from your test that the number of bits is 10?
     
  11. T N Args

    T N Args Agent Photocateur

    Dec 3, 2013
    Adelaide, Australia
    call me Arg
    Also, how much did you boost the gain in PP to get the image appearances posted on dpr?

    You didn't provide enough information for anyone to try and replicate your result.
     
  12. kwalsh

    kwalsh Mu-43 Top Veteran

    775
    Mar 3, 2012
    Baltimore, MD
    From the visual test (the only one that really matters) you can't tell that at all. All you can tell is that it is noisier.

    We'd suspect it was ten bits because that is what the GH4 data sheet says, but again that proves nothing.

    However, from looking at the RAW histograms (see the first DPR post) you can see in the electronic shutter case only every fourth bin is actually populated. That means two bits are "missing" and since we know from other tests and data the mechanical shutter read out uses 12 bits we can figure out 12-2 = 10.

    In the end though the number of bits is sort of irrelevant and just a curiosity, all that really matters is that it looks more noisy (again only if you push the shadows quite a bit, a case that doesn't apply to very many uses).
     
    • Like Like x 1
  13. kwalsh

    kwalsh Mu-43 Top Veteran

    775
    Mar 3, 2012
    Baltimore, MD
    Read the original post, it has full exposure information and says shadows pushed five stops. Anyone can reproduce this is the take five minutes to do it instead of posting here. Or you'll notice Fredrik Glickner also ran a similar test and posted that to his blog, see link in his post on the DPR thread.

    Seriously people, please read the original post in full before posting critical sounding posts about things already clearly explained in the OP on DPR. Why should I bother posting answers if someone isn't going to bother to read the original post in full, eh? :)

    Sorry - bad day for other reasons over here... As I said in the OP 1/500, ISO 200, deep shadow pushed five stops to make the difference very obvious. Naturally you would usually push less. Though some people prefer to shoot base ISO with exp. comp. instead of raising ISO when shooting RAW. You probably don't want to do that with the E shutter. Those are the kinds of people who care about details like this. Not applicable to most folks.
     
  14. dhazeghi

    dhazeghi Mu-43 Hall of Famer

    Aug 6, 2010
    San Jose, CA
    Dara
    I like to think that we're a friendlier more thoughtful forum than others (e.g. 1041 on DPR), but posts like this make me wonder :frown:

    Well I for one really appreciate you testing and reporting this. I have a buddy who shoots with the GM1 always with the electronic shutter (to avoid shutter-shake) and this is going to make a big difference in how he uses that camera.
     
  15. kwalsh

    kwalsh Mu-43 Top Veteran

    775
    Mar 3, 2012
    Baltimore, MD
    You are welcome of course. The nice thing with the GM1 is that if you are concerned by the DR loss of the E shutter and also worried about shutter shake I think shooting with the mechanical is going to be fine. I'd be very surprised if anyone finds shutter shake from the GM1. First of all it is electronic first curtain and on NEX and Canon this appears to fix shutter shake almost completely on its own. Besides that it is also very low mass and comparatively very slow. Finally it is stepper motor driven instead of spring driven and I suspect that means even less mass moving around and less jolting.

    At this point I plan to leave the shutter control on Auto (mechanical below 1/500, electronic above) and in the rare case I'd be shooting something where I cared about extreme DR I'll know to keep the shutter speed below 1/500 if possible. Like I said in my original DPR post I think Panasonic made the right compromises here for the vast majority of users.

    Longer term if the GH4 behaves the same way (we don't know for sure yet) that might rub a few people the wrong way. In many cases the same people that would obsess about shutter shake are the same who would obsess about base ISO DR. There are of course ways around both, and I've been using them when necessary since shooting both with older Canons and the original G1. Naturally no one even had an electronic shutter option until fairly recently, but once you give it to them they want it to be perfect :smile:
     
  16. T N Args

    T N Args Agent Photocateur

    Dec 3, 2013
    Adelaide, Australia
    call me Arg
    My questions were perfectly friendly and genuine. Methinks you jumped on the bandwagon too enthusiastically in your rush to criticize me. At least kwalsh has apologized and amended his post -- which somehow you missed was also in need of your critique.

    I did read the dpr post. I read the bit "A shadow region was then strongly pushed in post processing", so I asked how much, but I missed the repetition in the following paragraph, with 5 stops mentioned.

    The raw plot/bin explanation is pretty convincing to me, after reading post #12. I didn't know how to interpret from the lesser number of points in the raw plot that it implies a different number of bins filled (extreme techie), and didn't see anything in the post that explains it.

    I actually did want to try the kwalsh test with my GX7, it was genuine interest, I was trying to figure out how much the brickwork area was underexposed. Not necessarily 5 stops.....

    I was also going to shoot raw + jpeg, thought it might be interesting to see how bad the area in concern looks at 8 bit with processing....

    I also was interested to see if kwalsh had a GH3 and if it is 12-bit, since it takes 1/10th sec to read. If I read the dpr post correctly, kwalsh is connecting the loss of bits to the faster read speed. I read the Glickner article (might have missed a word though) and he looked at the GH3 vs GM1, but I don't see anything that answers that question.

    Hoping for some more friendliness and less personal critique.
     
  17. kwalsh

    kwalsh Mu-43 Top Veteran

    775
    Mar 3, 2012
    Baltimore, MD
    I only have the GM1 and E-M5 these days. From reading others posts and previous comparisons it looks like for the GH3 and G6 the electronic shutter and mechanical shutter give the same DR. Someone made a short post at DPR saying they tried the same test on their GX7 and say no problem and two other people posted to confirm the GX7 read out is slower than the GM1 though slightly faster than the GH3/G6 and that would also probably indicate no problem on the GX7.

    To try out your GX7 just underexpose a shot by three or four stops and push it back in post (or push a deep shadow in a normally exposed shot like I did). Fredrik's blog post now shows a good example of pushing a shadow with the whole image shown. If you have the time to try it let us know what you find out! I'm guessing you'll find no problem and the electronic and mechanical shutters have the same DR on the GX7.

    Also I bet with JPG on the GM1 we'd never see the difference. You really have to push the RAW quite a bit.
     
    • Like Like x 1
  18. meyerweb

    meyerweb Mu-43 Hall of Famer

    Sep 5, 2011
    If you want us to read and post on DPR, why bother posting here at all? By posting this here you've effectively invited comment and discussion. If you don't want people to react to your post, don't post.
     
  19. kwalsh

    kwalsh Mu-43 Top Veteran

    775
    Mar 3, 2012
    Baltimore, MD
    The point was your first post, and your second didn't contribute anything useful to the discussion.
     
  20. orfeo

    orfeo Mu-43 Top Veteran

    673
    Sep 27, 2013
    FR
    Man you don't have a gm1... Keep your Oly and peace out...