Adding Support for New Raw Formats

From RawPedia
Revision as of 12:32, 2 September 2018 by DrSlony (talk | contribs) (Revised, updated)
Jump to navigation Jump to search

Introduction

Supporting a raw format requires generally two things:

  • Being able to decode the raw file, so that the program has access to the image data and metadata stored within each raw file. This is handled by a program embedded inside RawTherapee called "dcraw".
  • Being able to interpret the image data. This can be further broken down into more items:
    • Measuring the black and white levels of the decoded image data, as every camera's idea of what is the blackest black and whitest white is different. This article covers this point.
    • Determining where the valid image lies on the "canvas" - the "raw crop".
    • Correctly interpreting the colors. This is done by generating an input color profile (DCP or ICC), as described in the "How to Create DCP Color Profiles" article.

You will have to shoot the photos, but you do not heave to understand or carry out the measurement - we can do that for you.

What are black and white levels? A sensor is made of millions of tiny photo-sensitive elements called photosites or sensels. Each one measures the intensity of the light which falls upon it, and records that intensity as a number - the more light, the higher the number. The bulk of the raw file consists of these recorded measurements. Each photosite has a level below which it cannot sense any light - there might be some light falling upon it, but this light is too weak to register a signal. This is called the black level, and it is not always 0. There is also a level beyond which the photosite will not register a change in light intensity even if the light does keep getting brighter - this is the white level. A photosite which cannot record any brighter light is said to be fully saturated, and in post-processing this state is called clipping.

Some camera models use the same white and black levels regardless of other settings, while for other models these levels depend on other factors, such as ISO sensitivity. We need photos from across the whole range of ISO values to determine this.

Some cameras have built-in noise reduction, often called LENR - Long Exposure Noise Reduction. It could affect the white and black levels. If you enable it, typically it only kicks-in when the exposure time is over 1s. The steps explained further on will explain when to turn it on and off.

The white levels for some camera models change depending on the aperture, but generally this only happens for wide-open apertures. To avoid this being a problem, set the aperture to f/5.6 or higher unless instructed otherwise.

What and How to Shoot

Shoot raw photos in manual (M) mode. If your camera has several raw modes, use the full one, uncropped, lossless compression if possible.

Each photo must be completely overexposed all across the frame. As such, it does not matter what you shoot since everything will be white anyway, but its easiest to achieve this while pointing the camera at the sky or at a white light bulb. It does not matter whether the sky is sunny or overcast, but don't point it at the sun as you might damage the sensor.

It does not matter what lens you use, but it will be easier to make the whole image overexposed if you do not use a wide angle lens.

Needed Photo Sets

We have broken down the image requirements into three sets, where each subsequent set would improve the quality of support, but would also require more effort from you and from us. In most cases, you only need to photograph the first two sets.

The sets:

  1. If your camera has LENR, turn it OFF. Set the aperture to f/5.6 or higher. Take a series of photos, one photo for every ISO value your camera supports, making sure not to exceed an exposure time of 1 second, using an aperture of f/8. As an example, you could end up with about 8 photos for ISO 100, 200, 400, 800, 1600, 3200, 6400 and 12800. Most cameras include intermediate ISO values, e.g. ISO160 or ISO320, so if you wanted to improve white level accuracy you would need to photograph these intermediate values as well.
  2. If your camera has LENR, turn it ON. Take a second series of photos, as described above but this time making sure that the exposure time in all cases is at least 2 seconds, not less. That's another 8 or more photos. In most cases these two sets should be enough.
  3. Some cameras scale raw values for larger apertures, particularly Canon and Nikon models. The only way to know whether your camera does this for sure is to take a photo and measure it. Take one photo using your lens's widest aperture, e.g. f/1.7, at ISO100 with LENR turned OFF, and send it to us along with the rest of the shots. If we detect that there is raw scaling (or if you detect it yourself if you do your own measurements) then we will ask you shoot a series of photos using an exposure time of less than 1 second, from the widest aperture your lens supports down every 1/3 of a stop until such an aperture where raw scaling is no longer performed. This could mean many photos. Handling raw scaling caused by large apertures is not very important so don't feel daunted by it, you don't need to do it even if your camera does do raw scaling, but if you have the time and bandwidth then it would be better to check for it.

At the very least, you should end up with a series of about 8 photos from point 1. It is recommended that you take photos for both points 1 and 2, leading to 16 or more photos, plus the one raw scaling test photo from point 3. If it is found that your camera performs raw scaling, you could additionally take the needed series described in point 3, but since this could potentially mean many photos (over 50) it is not expected.

Compress all these photos, upload them to filebin.net and send us the full link either through our GitHub page or in the Forum.

Completely clipped photos can have amazing compression, don't forget to compress them (7-Zip, ZIP, bzip2, whatever) before uploading! As an example, 10 completely clipped Sony 7M2 raw files with LENR disabled weigh 234MB but if you ZIP them you get a 1MB file.

More Information

For documentation detailing the required photos and instructions how to measure them, read the comments inside the camconst.json file: https://github.com/Beep6581/RawTherapee/blob/dev/rtengine/camconst.json

Storing and Reading This Information

RawTherapee looks for information regarding how to interpret the image data (the black and white levels and some other details) in three places:

  • In the dcraw code which is embedded inside RawTherapee,
  • In the raw file itself,
  • In a text file on your system which is installed with RawTherapee called camconst.json

The needed information is gathered from all three places, and camconst.json is prioritized above the others.