1. Reminder: Please use our affiliate links to get to your favorite stores for holiday shopping!

OV3 vs RT

Discussion in 'Image Processing' started by Brian Beezley, Jan 17, 2015.

  1. Brian Beezley

    Brian Beezley Mu-43 All-Pro

    Here are two comparisons between Olympus Viewer 3 version 1.4 and RawTherapee version 4.2.48 as RAW converters. I generated JPGs from each program using their highest quality settings.

    OV3 settings: default including/except natural picture mode, gradation normal, sharpness -2 with fine adjustment -2, saturation -1, noise filter off, and false color suppression auto.

    RT settings: Natural 1 profile including/except Amaze demosaicing, chromatic aberration auto-correction, and contrast and lightness adjustments to match OV3.

    Both programs used camera white balance. (OV3 reported it as 5819K while RT said 5890K.) Neither used the final sharpening step that I would normally employ.

    o12.

    r1.

    I took this backlit shot of a green heron using my E-PL2 and 40-150mm f/4-5.6 lens at 150mm, f/5.6, 1/500 sec, and ISO 200. The first image is from OV3 and the second from RT, both 100% crops. They are best compared by instantly switching between them when aligned on your screen. I use IrfanView to do this offline. You can do it online with your browser by right-clicking on each image and putting them in separate tabs. Some of the differences described below may be visible only in a dark room.

    The RT image is noisier. You can see this in the bird's head. Note the chroma noise in RT in the white lines of the feather pattern.

    The OV3 image is sharper. You can see this in the dirt texture and in the bird's feet. OV3 still sharpens the image even when you set sharpening to minimum. (I confirmed this behavior with Olympus.) I consider this a disadvantage. I prefer to leave all sharpening to the very last processing step to minimize the propagation/enhancement of sharpening artifacts. You really can't do this if you use OV3 as a RAW converter. But the sharpening is minimal and I have yet to identify an artifact from it.

    A curious difference between the images is that they have a slightly different scale. I cropped the images as identically as possible using the upper right corner as a reference. But the OV3 image has a narrower field of view even though both are 1200 x 900 pixels. I disabled distortion and lens correction in both programs, but the images suggest that one of them is still doing something.

    Even though I enabled automatic CA correction in RT, some remains. You can see some purple in the twigs to the right of the bird. I did not try manual correction.

    o21.

    r2.

    I took this shot at ISO 3200. Normally I never go above 800 with my E-PL2, but I wanted to better illustrate the difference in chroma noise. Here I also decreased RT color saturation to better match OV3 color. I switched my gaze between the screen and the orange-colored pink grapefruit a few feet away to compare color. The OV3 image color was realistic on my Samsung 204T monitor, which I've manually calibrated with a driver utility. The monitor is specified as 100% sRGB coverage. I didn't adjust the RT hue to attempt to match OV3. I've tried in the past without success. I set WB to 5300K for both images. Like sharpening, I think OV3 is doing some sort of chroma noise reduction even when you disable it. I thought it might be due to auto false color suppression, but there was very little difference when I disabled it.

    Normally I use OV3 to convert RAW files because its colors seem more natural and because it yields lower and far less colorful noise in shadows.

    Brian
     
    • Like Like x 4
  2. wjiang

    wjiang Mu-43 Hall of Famer

    OV3, ACR and LR will faithfully obey the micro four thirds specifications about digital lens correction, which you cannot switch off. The option to turn it off in OV3 appears to be for an extra step that is done after the lens specified correction (e.g. some lenses have barrel distortion even after the specified correction). All micro four thirds bodies will crop away a border around the edges of the frame that is reserved for correction, and the above converters do the same. Other RAW converters will just give you the uncorrected frame, which will be larger, and show all of a lens' nasty uncorrected edges.
     
    • Like Like x 2
  3. Brian Beezley

    Brian Beezley Mu-43 All-Pro

    Thanks for reminding me about the automatic lens correction. That explains the difference I observed in image scale between OV3 and RT.

    About the only time I use RT to convert a RAW image is when I want to recover something in the border area that OV3 suppresses. Below is an example where I used RT to recover the whole football. I used a 200mm legacy lens so there is no visible distortion in the border area. Too bad OV3 doesn't provide an option to recover this part of the image when it is usable.

    [​IMG]

    Brian

    Edit - In fact, it is possible to recover the border pixels in OV3 if you modify some parameter values in the .ORF EXIF. See https://www.mu-43.com/showthread.php?t=72715.
     
    Last edited: Mar 5, 2015
    • Like Like x 1
  4. Brian Beezley

    Brian Beezley Mu-43 All-Pro

    P0012618.
    P0012618.
    Here's another comparison between Olympus Viewer 3 and RawTherapee. The top image is from OV3 and the bottom from RT.

    I used default settings for my E-PL2 except for 5300K white balance, -0.3 EV exposure, and -2 sharpening. I left all settings in OV3 at default values and exported a resized image to RT for sharpening. For the bottom image I used the Neutral RT profile, and exported a resized image back to RT to sharpen at the output resolution.

    I liked the top image, but there was something washed out about the flowers. As usual, OV3 automatically adjusted the brightness to fill the display dynamic range. The OV3 clipping indicator, which I set to 255, shows colored dots scattered throughout the image but no blobs of color. Dropping brightness to -1 eliminates all dots. This is normal behavior. The RT histogram shows a few scattered spots at clipping level.

    The flowers have much more detail in RT and they aren't washed out. But despite its more linear and accurate intensity presentation, I prefer the OV3 image. In person the flowers were a dazzling carpet of yellow. That doesn't come through as well in the RT image.

    Since both images reach the brightness limit, with OV3 hitting it more often, I suspect the fundamental limitation is the brightness of my display. Were it ten times brighter, I might prefer the RT image with its greater intensity detail.

    I may play around with intensity nonlinearity in RT to see if I can get the best of both images.

    Brian
     
  5. Brian Beezley

    Brian Beezley Mu-43 All-Pro

    P0012618a.
    Here I used Contrast by Detail Levels in RT. This function dynamically adjusts contrast. It is kind of like Auto Gradation in OV3, but with much more control. I increased local contrast by three steps. You can adjust it individually for each of six detail levels, but I used the combined setting that yielded a boost of 1.9 at the finest level and 1.15 at the coarsest.

    The yellow became more intense without wiping out a lot of intensity detail. I like this image best.

    Brian
     
  6. mcasan

    mcasan Mu-43 All-Pro

    Feb 26, 2014
    Atlanta
    If you don't like Mac OS or Adobe raw converters, try DxO Optics. You can use DxO as a plugin to Lightroom and use either the Adobe or DxO raw converter and basic adjustments. DxO returns a DNG to LR. Then you can use the other plugging you have setup in Lightroom.