Why compression?

• Because graphics files can be huge
  
  
• 640 x 480 x 24 bits = approx. 1 megabyte per video resolution image
  
  
• 30 images / second = approx. 30 megabytes per second of motion video
  
  
• An 8 x 10 photographic image at 300 dpi equals 2400 x 3000 pixels, or about 23 times the file size of a video image at equivalent color depth
  
  
• System resources (e.g. memory, disk space) and bandwidth are finite
  
  

General compression catagories

• Lossless vs. lossy
  
  
  • Lossless - reconstructed file will be identical to original, uncompressed file
    
    
  • Lossy - reconstructed file will be missing some ``inessential'' data that were contained in the original, uncompressed file
    
    
  • Avoid using lossy compression in the middle of workflow

  
  
  

• Symetrical vs. asymetrical

  
  

  • Asymetrical compression may take longer to compress than to decompress
    
    
  • Motion video compression is an example (e.g. MPEG)

RLE (Run Length Encoding)

• ``Runs'' of similar data values are reduced to a single value and a description of the length of the run
  
  
• Lossless
  
  
• Supported in many image formats
  
  
• Efficiency is dependent on image
  
  
  • AAAAAAAAAAAAAAA
  • 15A
  • 5:1 compression ratio
    
    
  • AAAAAAbbbXXXXXt
  • 6A3b5X1t
  • 2:1 approx. compression ratio
    
    
  • Xtmprsqzntw1fbP
  • 1X1t1m1p1r1s1q1z1n1t1w111f1b1P
  • 1:2 compression ratio

  
  

• Good for most graphics, where long runs of similar pixels may be found
  
  
• Not so good for ``busy'' photographs

LZW (Lempel-Zev-Welch)

• Lossless
  
  
• Used in GIF and TIFF, among others
  
  
• Dictionary based encoding algorithm
  
  
• Encodes at least as well as RLE
  
  
• Noisy images tend to reduce compression effectiveness

JPEG (Joint Photographic Experts Group)

• Lossy
  
  
• Discards information not easily detected by human eye
  
  
• Uses luminace/chrominance schemes (YUV)
  
  
  • Human eye is more sensitive to luminance changes,
    less sensitive to color changes
    
    
  • JPEG downsamples chrominance component before encoding
    
    

• Advantages of JPEG
  
  
  • Supports 24 bit colorspace
    
    
  • Compression ratios of 20:1 to 25:1 without noticeable artifacts
    
    
  • Very useful for photographs

  
  

• Disadvantages of JPEG
  
  
  • Not very useful for many graphics
    - introduces artifacts into smooth color areas

  
  

• JPEG user strategy
  
  
  • Quality factor of 1-100
  • Higher quality means lower compression, and vice versa
  • Start with 75
  • If no artifacts, reduce quality factor
  • If artifacts, increase quality factor

Back to color outline