More about RAW conversion
DxO Labs' exclusive demosaicing technology enables you to make the most of the best possible optical, lighting and noise corrections on RAW images.
Not only does it offer a seamless workflow, but it also ensures optimum output image quality, as the entire image processing chain is optimized (all the DxO Correction Engines interact with each other).
DxO Raw Engine also offers a unique DNG export capability, offering tight integration with RAW applications supporting the DNG format, such as Adobe Photoshop®.
Please note that DxO Raw Engine is not available for Canon EOS D60, Nikon D100, Nikon D2H cameras.
DxO Raw Engine performs raw-to-RGB conversion, ensuring very sharp images while minimizing conversion artifacts.
Most digital cameras offer the possibility of shooting JPEG; some, typically the higher-end ones, offer the possibility of saving the Raw format. The JPEG format produces smaller, ready-to-use images, but it’s lossy 8-bit structure suffers from 3 main limitations: |
|
1. No exposure latitude
|
Sensors capture images at 12 or more bits per pixel. Once truncated to just 8 bits (for JPEG conversion), the latitude for exposure correction is greatly reduced, making incorrectly-exposed shots irrecoverable. |
|
2. No white balance latitude
|
If the white balance was perfectly set by the onboard software, the JPEG is usable as is. However, if any white balance problem occurs (due perhaps to an error in use, misinterpretation of scene colors or software limitations), correction of a JPEG file is extremely difficult and often produces unacceptable banding in the resulting image. |
|
3. Limited processing time
|
Because photography is often a real-time activity, it is vital for camera designers to minimize camera response time. One critical factor in camera responsiveness is the requirement to be able to handle a sequence of images, which in turn means the processing involved in producing JPEG images has to be optimized for speed, precluding the use of the most advanced CPU-intensive algorithms available. |
Demosaicing (also known as ‘Color Filter Array interpolation‘) is the most difficult step in Raw-to-RGB conversion because it has to recover information that has been lost: by design, Bayer sensors only capture 1/3 of the RGB data (each photosite is either Red, or Green or Blue). |
 |
The reconstruction of the missing 2/3 of the data faces two conflicting challenges: resolving the smallest possible details while limiting Raw conversion artifacts such as: |
|
1. Moiré patterns
|
 Moiré pattern example
|
| Sampling of frequencies above the Nyquist frequency produces aliasing, a folding of those frequencies into the lower part of the spectrum. The effect is called moiré by analogy with the effect created when one transparent pattern overlays a second pattern (as in moiré silk). |
|
2. Stairstepping
|
 Stairstepping example
|
| Diagonal straight lines become jaggy. |
|
3. False colors
|
 False color or color aliasing example
|
| Because there are twice as many green photosites as red or blue, aliasing can appear at lower frequencies in the red or blue channels than in green, leading to the appearance of false colors |
|
4. Halos (ghosting)
|
 Halo example
|
| Halos are created around very bright zones in the image. |
DxO Labs has designed a breakthrough "non-local" demosaicing algorithm that significantly reduces artefacts compared to competitors’ solutions, while delivering very sharp images and respecting manufacturer's choice and settings for color rendering. Most importantly, this RAW conversion enables the full DxO Optics Pro corrections to operate in 16-bits, providing ultimate image quality while retaining the potential to retouch images. |