![]() Here is the transition curve applied (shown below). To close this section, let me show you a desperate attempt to salvage some detail from the RGB image by applying a luminosity adjustment curve to stretch the rightmost histogram area: after all, the peak at the right has some width, so maybe we can get something? This is not exposure compensation, just moving the white point down, like printing an overexposed picture on a darker paper. ![]() Sample is just generally darker than (more exactly, it has the white point lower), but it does not reveal any detail in the white, burned-out area. If you are viewing this on an LCD monitor, then you will probably think that I mixed up the samples this one looks identical to the one above. In-camera JPEG (processed by the in-camera image engine) In particular, there is a loss of highlight detail in wall siding above the door. ![]() Traditionally, Olympus has one of the best JPEG engines in the industry, so this is perhaps as good as you get, except that I should have added 1/3 EV more of negative autoexposure compensation. The first sample shows the JPEG generated in the camera, by the imaging pipeline in the firmware. The raw file, if you would like to play with it, is here. (If you do not have one, the free FastStone can be recommended.) Smallest differences are then instantly visible, much better than in side-by-side viewing. The best way to compare between various results is to copy all XGA folder to one folder, and view them full-screen in an image browser. While the image fragments shown here are just 400×400 pixels in size, larger ones, cropped to 1024×768 pixels, are provided via the links in captions. (All these can be disregarded by the raw-to-RGB conversion software, or used as that software pleases.) Other parameters: white balance at 5300°K (sunny), Normal Picture Mode with contrast and saturation at -1, noise filtering at Low. Taking advantage of that, I decided to post this small article, showing examples.Īll pictures below are from one shot, taken with an Olympus E-30 camera with a 12-60 mm F/2.8-4.0 ZD lens at 19 mm aperture priority with -0.3 EV exposure compensation: 1/320 s at F/8, ISO 100. ![]() Some of my respondents were kind enough to send me the results of applying their favorite raw development application to my image (which was saved in both formats: ORF and JPEG). Interestingly, nobody said "my RAW pictures look better than JPEGs", which means that people who write to me are more technically literate than some magazine writers, who say similar things not realizing that you simply cannot view raw image information it has first to be translated into RGB pixels, either in the camera or in a separate software application. That's not true "shooting raw" allows for some latitude in fixing overexposure, visibly improving the final result as compared with in-camera JPEGs. That's right: if you don't get the exposure right, you cannot fix it in postprocessing, and people using the raw format are wasting their time to attain at most a marginal improvement The voices could be, roughly, divided into two groups: ![]() In two weeks that followed, I've received quite a number of emails on the subject from the Readers. I wasn't expecting the Olympus imaging pipeline to waste any significant fraction of information available. If this is possible in a given camera, it only means that the built-in JPEG engine does not efficiently use the whole tonal range recorded in the raw signal. I also expressed my skepticism about a possibility of recovering overexposed highlights from the information recorded in the raw image file, even with the overexposure being as slight as it was. In my E-30 image samples I've posted a picture with the highlights blown out by a slight overexposure: perhaps 1/3, maybe 1/2 EV. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |