RAW power

RAW is now a mainstream and accepted method of capture, but many photographers do not understand what it really is and how it can improve their work.

Manufacturers take RAW very seriously and in the past 12 months, two major software companies have introduced their own flavour of RAW workflow software while others have previewed major updates.

In the past few weeks Canon has updated its own established Digital Photo Professional (DPP) software to version 3, and with the EOS-1D Mark III, camera development has reached the stage where RAW is no longer any impediment to speed.

The Technology

All the cameras in the current EOS range use CMOS sensors, a technology, unique to Canon, introduced to the photo market by Canon in the EOS D30 in 2000, in place of the more familiar CCD sensors.

CMOS chips contain rows of light sensitive photodiodes, each with its own analogue to digital converter and amplifier to increase the electrical signal generated. These photodiodes, in conjunction with their associated microlenses, comprise the 'pixel site' that is the basis of the digital image.

Light hitting the micorolens is magnified on to the photodiode where an electrical charge is generated, converted to a voltage and amplified. Because each photodiode can create its own voltage signal, it allows each row of photodiodes to be read separately and therefore, multiple channels of sensor data can be read at the same time. In recent models of the EOS-1 series, eight data channels can be read simultaneously, allowing dynamic capture performance.

From the sensor, the image data is transferred to the DIGIC processor (Dual “DIGIC III” processors in the EOS-1D Mark III), a dedicated image-processing chip that handles a range of specific tasks:

• Signal processing
• Memory card control
• Camera LCD display
• Image parameters (colour, contrast, sharpness and JPEG compression)

 

The board with dual "DIGIC III" processors that handles camera control and image processing in Canon's EOS-1D Mark III camera.

For JPEG output, the 12- or 14-bit RAW file (depending on camera) is processed by the DIGIC processor(s) and saved with finalised exposure, white balance, sharpening and lossy JPEG compression.

Except for the Picture Styles customisation of the image parameters, the photographer has little control of the final JPEG image that is saved out. More importantly the 12 or 14 bits of data per pixel are reduced to just 8 bits by the processing, making any subsequent changes to the image a lossy and potentially damaging operation (See Fig 2).

By contrast, images shot as RAW contain all the information captured by the sensor. This is losslessly compressed and written to disk. The only extra information is in the file header which describes to the computer how the image was captured together with the basic EXIF data about camera type and exposure parameters.

Processing

RAW files need to be processed into usable image files from the raw data that they are captured as. Processing of the RAW file is done on the computer using a RAW converter like Digital Photo Professional 3, Lightroom, Aperture or Capture One.

By handling the processing in software it is possible to optimize the image in almost unlimited ways before output. Colour correction, white balance, exposure, contrast optimisation and noise reduction can be done in a near-lossless environment before creating a standard JPEG, TIFF or PSD file that may still be fine tuned if required.

By using the full set of camera data in the RAW converter, it is possible to do work on an image that simply isn't possible with a JPEG file. The image below was captured as a JPEG file but was underexposed by 2 stops. It can apparently be recovered, until a closer look is taken at the banding in the hair, under the chin and at the histogram.

The banding is a result of the few levels in the shadows being stretched by the exposure compensation, until gaps appear. By contrast, the RAW version of the file (to right) can easily be recovered and makes a good quality image without banding or visible damage.

The Advantage of RAW

It quickly becomes apparent that a 12-bit RAW file from a camera like an EOS 5D, containing 4,096 levels per colour is more robust and can be manipulated far more than a JPEG containing only 256 levels per colour, before damage to the image becomes obvious.

Maximum Image information

New technology from Adobe and others has been bought to bear on extracting the maximum detail from RAW files and improving workflow. This has resulted in huge gains for photographers both in their capability to extract details from images and from the fundamental changes to workflow.

In image terms, the latest innovations are highlight recovery and control, and shadow tools. These tools are effectively extending the dynamic range that can be used by allowing recovery of highlight and shadow detail that would be lost from a normal exposure. Canon’s own EOS-1D Mark III is aiding this process by capturing in 14 bit for enhanced gradation and having a setting for highlight tone priority. This expands the exposure level to the maximum allowable highlight level, improving graduation from the greys to highlights and expanding dynamic range by around 1 stop.

Back at the dawn of RAW in 2000, processing, even on the most powerful workstations was slow, clumsy and strictly on a one-at-a-time basis. The situation has now changed dramatically with real workflow and the possibility to effectively batch process large numbers of images rapidly, and without the need for the largest desktop computers.

Table 1 shows how many discrete colour levels can be contained in a file of given bit depth.

Table 2 and idealised histogram, showing numbers of discrete colour levels from highlights to shadows in RAW and JPEG files. The numbers in the top row show exposure in stops going downwards from the highlights.
The red area of the histogram shows the first stop, containing 50% of the image data.

Pros and Cons

The technical and workflow advantages to capturing RAW files as opposed to JPEG are very clear, but there are factors that still discourage some from utilizing the format.

Common complaints are:

• Less images on a card
• Images have to be processed
• RAW files are too big to transmit quickly

The cost of camera storage has recently dropped by a significant amount. A 2Gb card that was around €500 2 years ago is now (May 2007) around €35 and a quality 500Gb Firewire / USB2 hard drive can now be picked up for around €225, so the pricing argument no longer makes sense.

To avoid processing all files, it is possible to shoot both large JPEG and RAW. The latter can be used when any adjustments need to be made.

RAW files are big to transmit, but again, shooting JPEG + RAW provides the best of both worlds and allows the RAW to be utilized when re-use requests come in or adjustments need to be made to the file. With the recent software improvements, it is now as quick to adjust and process out a RAW file as it is to open up the JPEG and adjust it in Photoshop.

On the camera front, the EOS-1D Mark III has changed the landscape and offers both the ability to shoot at 10fps in RAW with a substantial, 30 frame buffer, but also the facility to shoot the new small RAW (sRAW) which offers all the advantages of RAW, but half the file size. Highlight Tone Priority allows a priority to be given to extra tonal information where it matters most and where half of the image data is contained − in the highlights.

RAW software has evolved rapidly to the point that it is now straightforward to get a more detailed and less noisy image from RAW images shot 7 years ago.

Despite all the changes, it has been said that we are still in the Stone Age of digital capture. The future looks interesting!