Why isn't more being made of sensitivity differences in cameras?

Discussion in 'Open Discussion' started by RichA, Mar 28, 2012.

  1. RichA

    RichA Mu-43 Regular

    Mar 28, 2012
    I've got a G1. I had a GH2. I've ordered an OM-D. I owned a Nikon D7000 for a couple weeks. I've owned 7 other Olympus DSLRs. I used to have a D300, and I owned a D40. I also owned K10D and K20D Pentax's.

    What I've found is that the sensitivity, the true ISO delivery of each sensor was different. Noticeably so. The Pentax sensors were about 0.7 stops slower at a given ISO than the Olympus cameras. The GH2 is at least 1/2 a stop slower than my G1 is. The OM-D is apparently even less sensitive.

    You can equalize illumination levels easily enough in post-processing, but reported noise figures and example images on websites mean little if the exposures and apertures used are different. An expensive f2.8 lens becomes (in effect) a much cheaper f4.0 lens on a camera with a less sensitive sensor. I can't understand why something with such an impact isn't being talked about more.
    • Like Like x 1
  2. pjohngren

    pjohngren Mu-43 Top Veteran

    Oct 15, 2010
    I have a G1, GF1, and G3 and haven't noticed any difference in the ISO's whatsoever. Of course, I am now using in-camera processing exclusively, because in my opinion it works better than converters made by other companies - certainly better than Adobe Camera RAW - way less noise, and I have all the in-camera noise reduction options set as low as possible or turned off.

    Is it possible that the differences you notice are really differences in your RAW converters? After all, each RAW converter is different for each sensor, and they are all created purely by guess work, since the sensors are proprietary and the people making the RAW converters don't really have a clue what is really needed - they don't have the proprietary specs on the sensor on which to base their converter. The in-camera processing/RAW converting, on the other hand, is created by the same people who made all other aspects of the camera, so very likely it is spot-on accurate.
  3. 0dBm

    0dBm Mu-43 Top Veteran

    Jun 30, 2011
    Western United States
    Welcome aboard.
    Please begin the dialogue.
  4. RichA

    RichA Mu-43 Regular

    Mar 28, 2012
    I use Photoshop, but I've used other converters as well. What I plan to do with the OM-D arrives it to run images through a few converters (I did this with my Olympus E-1 long ago and did find major differences in colour and detail rendering) and see what comes up. But if they bear out what I've seen with PS, SilkyPix and Nikon's converters, it will be interesting. What I'm most concerned with is the effect on noise levels once the presumed differences in the sensors is compensated for. Claims of 2-stop improvements in noise control might drop to one stop, that sort of thing.
  5. Hikari

    Hikari Mu-43 All-Pro

    Nov 26, 2010
    How did you measure this and how did you factor lens transmission?
    • Like Like x 1
  6. foto2021

    foto2021 Mu-43 Veteran

    Nov 5, 2011
    SE England

    It's easy to obsess about small details. A difference of 0.5 to 0.7 stops should not come as a surprise; different manufacturers have different ideas about whether their cameras should be optimised for retaining details in shadows or highlights.

    It has never been any different. In film days, the sensor (the film itself) also often varied in sensitivity from what was printed on the box. Many positive (slide) films were best underexposed by 0.3 stops and many negative films gave their best results when overexposed by 0.3 or 0.7 stops. The important thing was to find what worked best for you and then stick to it.

    The same advice is still valid today.

    It probably doesn't help that you have used so many different cameras in such a short time. Stick to one camera, learn how to get the best out of it and don't obsess about slight differences with other cameras. You will probably be a lot happier if you adopt that approach.
  7. Amin Sabet

    Amin Sabet Administrator

    Apr 10, 2009
    Boston, MA (USA)
    It's been discussed very often in some places. So much so that when it gets discussed, you end up with people making replies like this one.

    It depends how you define ISO. If you compare manufacturer in-camera JPEGs or processed RAWs in any of the major RAW processing apps which have profiles to support the camera, you'll find that for a given exposure (dependent only on ambient light, f-number, and exposure time), the ISO determines the final brightness (apparent exposure) of the image. There is no standard definition of ISO which correlates with sensor sensitivity per se, although DxOmark uses ISO to mean this.

    Let's say we have two cameras:
    Camera A has a sensor which is half as sensitive as Camera B, and Camera A "pushes" the RAW file by an extra stop relative to Camera B when making the in-camera JPEG. Lightroom and Aperture know to apply this extra push by default. You end up with equally bright images in the end for any given ISO, which meets the usual definition of ISO. However Camera A is giving you more highlight headroom in the RAW file, while Camera B is giving you more shadow latitude in the RAW file. Neither one is "cheating".

    DPReview's approach is to first make sure that under circumstances of equal exposure, cameras are giving equaly bright images for a given nominal ISO. The vast majority of cameras are within 1/2 stop of one another in this regard. Then, for sake of convenience, they all the lighting conditions to vary, while the shutter speed and aperture such that the exposure is the same (eg, if the light source is half as bright, they may give twice the exposure time). They introduce more variables in doing it this way, and the tradeoff is that it's more convenient for them.

    Here are a couple posts which shed light (no pun intended) on the DPR method:

    Re: Olympus OM-D (EM-5) comparison samples are now.. (continued): Micro Four Thirds Talk Forum: Digital Photography Review

    Sorry, Bob, but you've made a critical mistake: Micro Four Thirds Talk Forum: Digital Photography Review
    • Like Like x 2
  8. RichA

    RichA Mu-43 Regular

    Mar 28, 2012
    I used the same manual lenses on the cameras and eye-matched the illumination levels.
  9. RichA

    RichA Mu-43 Regular

    Mar 28, 2012
    If you at the edge of options, that 1/2 stop can make or break as shot

    Scenario: I have two cameras and want to shoot a street scene at night. One will give me the noise and illumination levels I need at 1/30th at f2.0. The other by virtue of its less sensitive sensor requires either 1/15th at f2.0 or 1/30th at f1.4 and I don't have f1.4 available on the lens I'm using. With the 1/30th sec at f2.0 camera, I am right on the edge of what is possible for the image without compromising it. The other camera, despite claiming the same ISO cannot make the shot with the settings of aperture and speed I need. I'm forced to shoot at 1/15th sec at f2.0 and compromise the speed too much, or, 1/30th at f2.0 and compromise the exposure. That is where the difference in sensitivity effects the outcome of the shot and why this matters.
  10. Djarum

    Djarum Super Moderator

    Dec 15, 2009
    Huntsville, AL, USA
    Real Name:
    With a camera that underexposes its RAW files, you'd still be shooting at the same shutter. The RAW converter would apply a different tone curve to get the image to the same exposure level. In low light shooting, this typically means more noise. In daytime shooting, this means preserving highlight detail.

    Most cameras are going to display the image on the LCD as if it were a jpeg anyways. So, if you were using two different cameras with slightly different sensitivities, the same f stop, and the same shutter speed, both images in the LCD will look pretty similar. You wouldn't notice any exposure difference unless looking at RAW, but you wouldn't do this until you PP anyways.