dynamic range

see also:

• exposure
• high dynamic range (HDR) photography with Olympus OM-D cameras
• HDR image formats:
  • http://www.anyhere.com/gward/hdrenc/hdr_encodings.html 
• Using luminance masks in Photoshop to blend images to increase dynamic range

• see also: 
  • digital camera sensors and photographic film in a digital age
  • http://www.ronbigelow.com/articles/dynamic-range-1/dynamic-range-1.htm 
• this is the ability of a film or digital camera to capture a range of image intensities that will then be translated to the range of black to white in the final output.
• Slides have less photographic range than negatives. Slides have perhaps about 5 or 6 f-stops of total scenic range, compared to perhaps 9 or 10 f-stops for negatives. 
• the narrower the capture dynamic range, the more important accurate exposure becomes:
  • expose a slide half a stop off and the results are objectionable. Do that with negatives, and you never even realize it.
• digital camera dynamic range:
  • this is primarily determined by:
    • the characteristics of the image sensor (the results you get when shooting in RAW mode)
      • number of bits A-D converter
      • amount of noise produced by the sensor
    • modifications by in-camera image processing (or on computer RAW development processing) which will usually result in some loss of dynamic range (often 1-2 EV which is usually in the highlight region). 
  • digital sensors have a linear analog output from each sensor site such that the output voltage is proportional to the amount of photons (light) hitting it. This analog output is then converted to a numeric digital value via the camera's analog-digital converter chip which then forms the “RAW” image file (although some cameras (eg. Nikon dSLRs) have in-built processing that refines the image file further before outputting the end RAW file.
  • in digital data terms, maximum possible dynamic range of sensors is determined by how many “bits” of data can be saved with each image, in general, the larger the sensor, the more dynamic range:
    • YouTube: 8bit vs 16bit - Bit Depth EXPLAINED In-Depth
    • most prosumer cameras capture 9 to 10 bits which equates to 5-6 EV dynamic range (ie. equiv. to slide film)
    • digital cameras may capture:
      • 8 bit jpeg (24 bits per pixel):
        • each colour channel can have up to 256 values
        • this is the most common final image format such as jpeg files
        • this is the reason you should really shoot in RAW mode if you want to post-process your files - the limited bit depth of 256 values will result in posterisation artefacts if jpeg files are significantly post-processed.
        • cameras and RAW development software convert their 10-14bit RAW data & map it to the 256 values of an 8 bit jpeg file (see below).
      • 12 bit RAW (36 bits per pixel):
        • most Micro Four Thirds system and older digital SLR's
        • full frame cameras in compressed or burst modes may drop to 12 bit ^{1) 2)} but there is little difference in image quality that is discernible if other compression artefacts are not introduced such as spatial filtering by Sony which “ate stars”
        • each colour channel can have up to 4,096 values
      • 14 bit RAW (42 bits per pixel):
        • most current full frame digital cameras
        • each colour channel can have up to 16,384 values
      • 16 bit RAW (48 bits per pixel):
        • this currently is only the medium format digital camera systems cameras costing upwards of $A10,000 although they may not be true 16bit range.
        • each colour channel can have up to 65,536 values
      • 32 bit HDR (96 bits per pixel):
        • HDR files generated from multiple exposures via HDR software
  • RAW image file processing:
    • the digital camera RAW files image file needs to be processed to derive an image that can be viewed, this involves:
      • gamma encoding correction to the desired color space (eg. sRGB has gamma=2.2)
      • de-mosaicing (Bayer array interpolation) - each photosite on a sensor is sensitive to only one RGB colour channel and thus the data from adjacent photosites must be interpolated to give a 3 color channel RGB pixel.
      • sharpening - although lower end cameras often over-sharpen to give better looking small files but produce artefacts that can be hard to remove, thus it is better to minimise in-camera sharpening and leave it to the last step in image processing.
      • white balance adjustment
  • image files and dynamic range:
    • for details on tonal quality, dynamic range and digital see here
    • RAW files can be converted to either 8 bit JPEG or TIFF files, or to keep as much information as possible, to 16 bit TIFF.
    • note that jpeg files are only 8 bit so you have lost range by converting to jpeg.
    • 8 bits mean that the data per color channel for each pixel can have a value within the range of 0 to 255.
    • if one uses a linear 8-bit image file (ie. gamma = 1 such that data spread is even for each part of the exposure range) then to fit a dynamic range of 8EV equally, each 1EV in dynamic range would only have 256/8 = 32 possible values per channel which would not give a sub-optimal image quality.
    • Gamma encoding spreads the data and shifts the results towards the more sensitive end of the human visual range. The stages are first to convert the analog signal to a 10-bit or higher internal values (depending on the hardware), then apply the gamma function to these values, before finally rounding down to 8-bits. This preserves much of the dynamic range, and minimises the perceptible quantisation errors in 8-bit values. 
    • Thus if we look at the brightest 1EV exposure zone in an image, if it was shot in 12bit RAW using sRGB color space, this zone would have available to it 2048 brightness levels per channel, when this is directly converted to a 8-bit jpeg, this same exposure zone has those 2048 levels squeezed into only 69 levels.
    • If we look at the exposure zone 3 stops down from the brightest, and thus closer to skin tones, in 12bit RAW there are 512 levels which are squeezed into only 37 levels in an 8-bit jpeg.
    • if we under-exposed our skin tones by 2 stops, we would have only 128 levels in RAW which are squeezed into a measly 20 levels in jpeg with resultant marked loss of tonal range - what could have been 512 levels of tonal range has been converted to only 20!!
    • THUS GENERAL RULES OF DIGITAL FOR GOOD SKIN TONAL RANGE:
      • shoot in RAW mode at the camera's base ISO and don't under-expose.
  • effect of ISO on dynamic range:
    • why don't we all just shoot with camera set at camera's base ISO (usually 200) and under-expose as need be instead of changing the ISO upwards?
      • if one assumed there was no in-camera sensor boosting taking place, then if you double the ISO, this would equate to under-exposing by 1EV (then boosting it by 1EV in Photoshop to give correct exposure appearance), but the result would be you would lose 1EV of shadow detail, and perhaps worse if shooting in jpeg mode, due to gamma encoding as above, you move your most important part of your subject exposure to a lower data range which has less data elements available for the same amount of exposure range, thereby degrading you image.
      • In actual fact, there is some sensor boosting going on when you select a different ISO in the camera, and this reduces the dynamic range loss from 1EV to about 0.5EV for each doubling of EV and reduces the 
    • this maximum dynamic range value only applies to the ISO range where the camera best performs at which is usually ISO 200. As you increase the ISO, noise adversely impacts shadow detail and thus the shadow region dynamic range reduces perhaps by 0.5EV for every doubling of the ISO on digital SLRs.
    • dynamic range of the highlight region remains relatively unaffected.
  • to get maximum dynamic range, shoot in RAW mode at the lowest ISO setting, expose correctly to avoid blown-out highlights and then select a RAW developer program and settings that give the best dynamic range (this may include reducing EV by 1-2 stops to capture the highlights from being blown out and reducing the contrast setting, but color accuracy may be affected with extreme settings).
  • digital effects use of Photoshop can improve over-exposed or under-exposed regions of an image but it cannot restore image detail once it is lost due to insufficent dynamic range.
  • to get even more dynamic range, one can bracket exposures by +/- 2 f-stops and then combine the images to give a high dynamic range image:
    • high dynamic range (HDR) image production and tone mapping from multiple images of differing exposures (ie. bracketed exposures such as -2, 0, +2 stops using a tripod):
      • Photomatrix or Photomatix  
        • free basic version, but need pro version ($US99) to do tone mapping.
        • may be of use to reduce noise in astrophotography
      • Photoshop CS2
      • see tutorial here
      • Savad's method:
        • you'll need 3 exposures (-2, 0, +2) from the camera, adjust the light blending in photomatix, push the black up a bit, turn micro contrast down to max, micro smoothing to zero. push the whites a bit, the rest is up to you and the effect you want. if the light smoothing is to the right more, it will be more real, to the left, much less real. Use PS to finish.
    • high dynamic range landscapes from 2 images using gradient tool in PS:
      • see tutorial here
      • one image exposed for the sky, the other for the foreground
      • combine as layers in PS then add a layer mask on the over-exposed image and use the linear gradient tool and drag the line from the top of the image to the horizon, then adjust the opacity slider for effect
    • pseudo high dynamic range from a single RAW file - suitable for motion images but gives more noise:
      • Photomatix 2.2 or later 
        • Automate-Batch Processing select one RAW file & hit run; need PS white balance afterwards.
      • FDRTools 

• this is the EV range from the darkest shadows in which you wish to see detail to the brightest highlights in which you wish to see detail.
• see exposure
• the scene dynamic range depends on:
  • subject dynamic range:
    • eg. bride in white, groom in black creates an extreme subject dynamic range
  • lighting dynamic range:
    • ie. the difference in exposure for the main light source vs the light source lighting the shadow region.
• if you take a photo on a sunny day with the sun directly behind you so there are no shadows, then the dynamic range will match quite well with the cameras ability to capture it, but the lighting may not be how you would like it.
• if you shoot with light source to the side of in a backlit situation, then the resultant shadows will cause extreme scene dynamic range which may need to be managed, either by:
  • exposing for the highlights and let the shadows come out black
  • exposing for the shadows and let the highlights come out white
  • exposing somewhere in between and having black shadows and blown out highlights
  • modify the scene dynamic range as below.
• if you take photos on a heavily overcast day with minimal shadowing, the scene dynamic range may be quite small and the resultant image may appear quite “flat” and boring
  • the scene can be given contrast in a variety of ways such as:
    • adding colour objects to give colour contrast
    • using Photoshop to stretch the contrast so you get black blacks and white whites, but this process may add noise.
• the scene dynamic range can be effectively modified by either:
  • additional photographic lighting basics or flash technology to fill-in or lighten up the shadows, or,
  • use of gradient &/or polariser  photographic filters to darken bright highlights such as the sky
    • see dealing with high contrast scenes
  • use of colour  photographic filters to modify the contrast range in B&W photography

• see also:
  • http://www.photo.net/learn/drange/ 
• this is the range of optical density on a given image and becomes important when using a scanner to copy it as the scanner ideally needs to have sufficient image capture dynamic range to cover the image dynamic range.
• Image density is measured from image brightness with optical densitometers, and ranges from 0 to 4, where 0 is pure white and 4 is very black. More density is less brightness. Density is measured on a logarithmic scale (similar to the Richter Scale for earthquakes). Density of 3.0 is 10 times greater intensity than a density of 2.0.
• The minimum and maximum values of density capable of being captured by a specific scanner are called DMin and DMax. If the scanner's DMin were 0.2 and DMax were 3.1, its Dynamic Range would be 2.9. DMax implies unique image tone values are distinguishable, and not hidden by electronic noise.
• 24 bit scanners might have a dynamic range specifications near 2.4, needed for photo prints. 30 bit scanners might be near 3.0, needed for negatives. The best 36 bit scanners might approach 3.6, better for slides.
• A printed magazine image has a dynamic range well less than 2.0, maybe half of that (1.7). The blackest ink still reflects some light, the white paper is not so bright that it blinds us, and the difference is relatively small. Photographic color prints have a dynamic range of less than 2.0 too. Film negatives might have a range up near 2.8. Slides may be near 3.2. These are not precise numbers.