Raw data format

from Wikipedia, the free encyclopedia

As raw data format or raw file ( english raw "raw") refers to the particular file format for digital cameras and digital cinema cameras , in which the camera data after digitization largely without processing to the storage medium writes. Since the respective format reflects the raw data of the camera, it differs from manufacturer to manufacturer and often also from model to model, sometimes also between different versions of a camera model. The raw data are proprietary formats , the structure of which is partially known only to the respective manufacturer.

The "raw data" are sometimes referred to as "digital negative" because they contain the original data of the recording. Although the basic functions of digital image sensors from different manufacturers and models do not differ significantly from one another, no standard has been established for raw data - sometimes “raw” and “standard” are generally considered to be contradictory.

use

After recording and storage, the digital raw data is usually in a proprietary format, i.e. That is, the structure of the files is not disclosed by the camera manufacturers. In most cases, the data can then only be processed with the software of the respective manufacturer, or with programs from third-party providers who work with the manufacturer or who have decrypted the manufacturer's formats through reverse engineering . Current professional image processing software has integrated raw converters for reading in raw images. The market leader Adobe Inc. is trying to establish its own format DNG (Digital Negative) as a manufacturer-independent standard.

The raw data is usually supplemented by the camera's firmware with additional data, such as B. Exif data or a preview image in JPEG format that can be used inside the camera or on the computer for image control.

Other common formats for image storage often do not allow all of the information supplied by the sensor to the camera to be saved. JPEG variants allow e.g. B. only 256 brightness levels (8 bit) per color channel , while raw data formats usually contain 10, 12 or 14 bits of brightness information, which enables 1,024 to 16,384 brightness levels. Due to the characteristics of the Bayer sensor , which is usually used , in which there is a filter for one of the three basic colors in front of each pixel , complex interpolation (so-called demosaicing ) is necessary in order to calculate the resulting colors for each pixel at full resolution.

Juxtaposition

Most camera models can be switched between image recording as manufacturer-specific raw data and recording as a “developed” image in a standard image format; Often only JFIF ("JPEG") is offered as the standard image format . Therefore, a comparison is made below between raw data and JFIF. Other standard image formats sometimes have significantly fewer disadvantages, but must first be created from the image information supplied by the camera.

Raw data Image developed by the camera (JPEG)
flexibility The camera-side image parameters focus , aperture , exposure time and light sensitivity are inevitable in the raw data .

All other parameters can later be selected as required during the raw conversion, including the interpolation algorithm used.

In addition, the preprocessing steps such as B. white balance , color saturation or applied color filter (for black and white photography), color space adjustment , contrast enhancement , sharpening , noise reduction , and JPEG compression rate as well as individual corrections (e.g. of skin tones).

Subsequent correction with image processing programs is also possible on preprocessed JPEG images, but subject to limits and always associated with loss of image information and details. The interpolation used by a Bayer camera can no longer be changed later.

picture quality By storing the (uninterpolated) sensor data with 10, 12, 14 or 16 bits per pixel, a more precise brightness resolution is given. All details captured by the image sensor are completely retained. 8 bit color depth means a maximum of 256 gradations per color channel. Subsequent tonal value corrections create “gaps” in the histogram , i.e. loss of information. In addition, the always lossy compression , in addition to the loss of image details, creates typical compression artifacts , especially in low-contrast areas , the amplification of which (e.g. by contrast curves or unsharp masking ) must be taken into account and avoided in the following post-processing steps.
File size Despite the compression method used, the files are often much larger than their JPEG equivalents, especially since most manufacturers also contain the JPEG image generated by the camera as a preview in addition to the raw data. The writing of the files consequently takes longer, which is why a large buffer inside the camera and a high writing speed are necessary in order to achieve an acceptable serial image speed in raw mode and to maintain it over a longer period of time. Depending on the setting of the JPEG compression, the image files can be considerably smaller and z. B. can easily be sent by email or MMS if configured accordingly.

Even with the best JPG quality set, memory cards (depending on the subject) usually hold at least three times as many images as in the raw format, and with reduced quality, a multiple of them. Because of the smaller amount of data and the much longer camera buffer, the writing speed does not play such an important role, and even slower cards can “keep up” with the camera for longer in series mode. The decisive factor here is the speed of the camera's internal processor.

compatibility The raw formats of the individual manufacturers differ from one another and are not compatible. There are also differences and deviations between models from the same manufacturer. Therefore, special raw converter software, either individually or integrated as a plug-in in another program, is necessary for use. Even though currently (as of October 2013) most of these programs can handle almost all raw data formats on the market, there is no guarantee that this will always be the case. One possible way out would be the use of a documented, uniform and manufacturer-independent storage format (e.g. Adobe's digital negatives ), another would be the complete and freely accessible documentation of the raw data formats by the respective manufacturer. Practically all existing image editing and processing programs support the open JPEG standard and can read and write files, regardless of their origin. However, there may be restrictions in the evaluation, processing or writing of embedded Exif information, such as manufacturer-specific maker notes , geo-tags or preview images.
post processing The raw data must always be post-processed in order to be displayed or edited. The main step (with Bayer data), the interpolation or the resolution of the mosaic pattern ( demosaicing ), as well as the noise suppression can take significantly more time than the image signal processors optimized for it need for the same step.

Sophisticated software interpolation algorithms and often multi-level wavelet or motif-based noise suppression result in a high system load and considerable memory requirements due to the calculations carried out in 16-bit integers or floating point.

Due to the compromise that the manufacturer has to make in order to perform the interpolation quickly enough and yet with acceptable quality, problems or clipping effects occur at the edges, which is why subsequently converted raw images may be slightly larger than the JPG files calculated internally in the camera .
Loss-free post-processing is possible, but limited to certain processes (rotation by a multiple of 90 ° and cropping by a multiple of 16 pixels). With all other changes there is a generation loss, since JPEG uses lossy compression. As a storage format, it is therefore disadvantageous during post-processing, and a lossless format (e.g. TIFF or PNG ) should be used.

Manufacturer specifics

Raw data pre-processed by the camera

In contradiction to the strict design of raw image files as an image of the raw data of the image sensor, manufacturers are increasingly processing image files inside the camera as well, the details of which, however, similar to the internal image processing of JPEG files, are usually not disclosed. The following reasons are possible:

  • Some sensors do not implement variable signal amplification. Exposure indices (arbitrarily high) are subsequently simulated by applying a multiplier to raw image data. This is evident from the fact that at higher ISO values ​​many intermediate brightness values ​​never appear in the image. So-called breaks then arise in the brightness or color gradient.
  • The assumption that (non-manufacturer) raw conversion software has or uses less information about the camera and the lens currently being used, thus making insufficient use of the potential for error correction.
  • A masking of weaknesses of lenses (e.g. sharpness) and sensors ( hot pixels or e.g. noise).
  • Hardware-based methods may be faster than software-based calculation methods (e.g. noise reduction at camera level), but even then they are associated with a loss of detail.
  • With modern sensors that work with phase autofocus on the chip, the areas used must be interpolated.

File formats

Just as the data from the various manufacturers is stored in different file types, the way manufacturers deal with the disclosure of their formats is just as different. Sigma discloses all data on its format (this no longer applies since Sigma SD15 and its products, such as Sigma Merrill and Sigma dp2), while most others such as Olympus and Canon do not document their formats or only incompletely. Nikon uses partially encrypted information within the file in its NEF format (Nikon Electronic Format), which programmers can, however, automatically decrypt in their own applications with the software development kit available free of charge . At the end of 2005, however, Nikon disclosed the encrypted white balance information, which means that Nikon-independent software can also decrypt the data. The NEF format is also used for film scanners and image processing software from Nikon, as it can contain processing steps and other settings in addition to the actual image data. Some film scanners are equipped with an infrared channel for removing dust and scratches . The HDRi raw data format can record this infrared raw data as an additional 16-bit channel.

With the introduction of new camera models, new proprietary file formats are occasionally created within the same camera series from one manufacturer (e.g. Canon CRW and CR2). However, there is no guarantee that any conversion software will be available; H. Archived raw data may at some point no longer be read by future programs.

For this reason, an interest group has been formed under OpenRAW.org that calls on camera manufacturers to unrestrictedly disclose the raw data formats, which would enable the user to access his raw data for many years to come, without being in possession of the previously functioning software process and, if necessary, be able to write a program to support its now outdated format. The software manufacturer Adobe is also pursuing this goal with the introduction of the DNG format.

TIFF

Some cameras offer storage in the lossless TIFF format as an alternative to the JPEG format . This is important when there is no way to save the raw data directly. TIFF has the advantage that there is no lossy JPEG compression and that the color depth is not limited to 8 bits. Bayer filters, white balance, tonal value correction and noise suppression are also applied to the image data, as is sometimes (depending on the manufacturer) the reduction to 8 bits per color channel (24-bit color depth).

Digital negative

Based on the film strips in analog photography, raw photos are sometimes referred to as “digital negative”. This term is used by Adobe Inc. to refer to its own patented open raw data format. The Digital Negative (DNG) format was developed with the aim of replacing the proprietary formats used by camera manufacturers.

Filename extensions (selection)

Raw data support in the amateur area

The camera manufacturers regard the raw data format as a professional function; it is standard in digital SLR cameras and mirrorless system cameras as well as in some semi-professional compact digital cameras. Often only limited software for simple raw data conversion is offered for the cheaper camera models.

The years 2001 to 2004 were characterized by the fact that, in addition to the digital SLR cameras, more and more compact, semi-professional digital cameras were equipped with raw support. In its early days, the raw data format was apparently not yet seen as a strategic added value. It was only in later years that this feature was specifically used as a distinguishing criterion, when many manufacturers were already offering raw data format support below the professional models.

Canon installs its raw-capable DIGIC image processors in many digital camera models. Activists have been working on the CHDK firmware upgrade since 2007 . In addition to extended recording modes, it enables a number of compact cameras to save raw files.

Meanwhile, support smartphone - apps that store photos in a raw data format. The prerequisite for this is that the hardware and operating system allow access to the raw data of the camera.

Raw data processing

Comparison of the basic raw data processing in the digital camera and with the raw converter in the PC

A typical software for raw data conversion is u. a. the following functions are available:

Raw data conversion programs can save the data after Bayer interpolation (so-called demosaicing ) and conversion into the RGB color space , assuming a preliminary white balance, without extensive further corrections in 48-bit TIFF files (16 bits per color per pixel). The corrections can then be carried out in other image processing programs.

software

The camera manufacturers usually offer their own (proprietary) program or a software development kit for importing raw data images and for post-processing. Typically you can then the images z. B. export to JPEG format. For Windows operating systems, Microsoft provides the “Microsoft Camera Codec Package” which enables common RAW formats to be displayed in Windows Explorer and the Windows Photo Viewer . Numerous image processing programs and conversion programs offer the import of raw data images and subsequent image processing. The better known are:

Raw data processing programs
Commercial software Free software
ACDSee (Windows, macOS) GIMP (UFRaw-based, Windows, macOS, unixoide)
Adobe Photoshop (Windows, macOS) IrfanView (Windows)
Adobe Photoshop Elements (Windows, macOS) Darktable (Windows, macOS, Linux ; FreeBSD , Solaris )
Adobe Photoshop Lightroom (Windows, macOS) DCRaw ( platform-independent )
Affinity Photo (Windows, macOS) digiKam (DCRaw-based, Windows, macOS, unixoide )
Apple Photos (macOS) F-Spot (UFRaw-based, unixoid)
Bibble (Windows, macOS, Linux ) FastStone Image Viewer (Windows)
Capture One (Windows, macOS) PhotoLine (Windows, macOS) - image software from Google
Corel Photo-Paint (Windows, macOS) Rawstudio (DCRaw-based, macOS, unixoide)
Digital Photo Professional (Windows, macOS) - Canon’s corporate program
DxO Optics Pro (Windows, macOS)
GraphicConverter (macOS)
HDR projects (Windows, macOS)
PaintShop Pro (Windows)
PhotoImpact (Windows)
PhotoLine (Windows, macOS)

Others are:

Software development kits

Some camera manufacturers offer software development kits for software developers that allow developers, among other things, to access the image files in raw data format:

  • Canon EOS SDK
  • Nikon Software Development Kit

hardware

There are also some hard drive based mobile image storage devices that can decode raw files and display them on the built-in screen. This means that sorting and organizing is possible without a PC or camera.

cinematography

Digital recording technologies are also increasingly used in the production of cinema films, and accordingly each camera has its own raw format. With CinemaDNG there is also an open format for recording film data, which comes from Adobe and is used, for example, by the Swedish camera manufacturer Ikonoskop or by Blackmagicdesign from Australia.

literature

  • Christoph Künne : Digital Negatives. Camera Raw . Addison-Wesley, Munich 2005, ISBN 3-8273-2314-2 , ( Photoshop basic knowledge 4), ( Edition DOCMA - dpi ).
  • Mike Schelhorn: Adobe Camera Raw. Develop, edit and organize digital negatives. (For Photoshop CS / CS2 and Elements 3.0 / 4.0. Plus: Introduction to Adobe Lightroom) . 2nd updated and expanded edition. Addison-Wesley, Munich a. a. 2006, ISBN 3-8273-2450-5 , ( dpi ).
  • Uwe Steinmüller, Jürgen Gulbins: The Art of RAW Conversion. Edit RAW files with Adobe Photoshop CS2 and leading RAW converters . dpunkt, Heidelberg 2005, ISBN 3-89864-351-4 .
  • Andrea Trinkwalder: Raw mass. Higher color depth, fewer errors. Better pictures thanks to raw data . In: c't Magazin für Computer-Technik 2004, 16, ISSN  0724-8679 , pp. 152-157.

Web links

Individual evidence

  1. ↑ FAQ on dcraw
  2. ↑ Luminous-Landscape.com: Article "Digicams vs. DSLRs " (engl.)" What about makers such as Canon that have admitted that they apply sharpening even to Raw files on some of their cameras "
  3. ↑ Pixelfehler , Wikibooks Digital Imaging Methods , Chapter Light Conversion , accessed online on October 6, 2013
  4. ↑ Adobe Now Supports Canon CR3 and C-Raw Files from the M50 . In: Dan Carr Photography . April 3, 2018 ( dancarrphotography.com [accessed June 6, 2018]).
  5. ↑ CHDK-Wiki firmware attachment for Canon compact cameras with DIGIC processors
  6. ↑ Microsoft camera codec package on the provider's website
  7. ↑ Canon Digital Camera Software Developers Kit General Information. In: Canon Developer support. Canon USA, Inc., accessed May 4, 2017 .
  8. ^ SDKs for Digital Imaging Products. In: sdk.nikonimaging.com. Nikon Corporation, 2017, accessed May 4, 2017 .