----- Meeting Notes (2011-12-27 12:24) -----PNG, or Portable Network Graphics, is a lossless, compressed bitmapped image format. It was released in its first incarnation in 1996 as a truecolor compatible replacement for the 256 color GIF format.PNG uses lossless compression, meaning that the quality of the graphic information stored does not decrease with the compression. This can be compared to a lossy format like JPG which is prone to quality degradation and unwanted artifacts.Let's take a look at the most interesting aspect of the PNG format: the compression.PNG is compressed in two stages. The first stage, the pre-compression, makes use of a process called filtering, or prediction. If we take an image, like this oneand zoom in to get a good view of the pixels, we see that there are large areas of the same color, or value, in the image. Let's call this pixel here X, and these we can call A, B and C. The filtering component of PNG uses the values of byte-sized chunks of pixels A, B and C to predict the values of the corresponding chunks of data defining pixel X. The predicted value is then compared to the actual value of pixel X, and the difference is encoded.After pre-compression, the resulting data is further compressed using the DEFLATE, lossless data compression algorithm which looks for repeating strings of binary digits in a file and replaces them with shorter reference "symbols".PNG is most commonly used in web design, because it retains all the sharp edges and details of the original picture. It is less often used for photos, since the compression algorithm is not as well adapted to the far less predictable patterns of photographs. Also, unlike GIF images, PNG is a single-image format and does not support animation.Saving an illustrator image at 1425x1245 pixels as a BMP, an uncompressed image format, results in an image file of 4,2 Megabytes. Saving the image instead as a PNG leaves us with a 715KB file, a compression ratio of approximately 6:1.FLAC, or Free Lossless Audio Codec, is a relatively new, free, open source audio codec allowing lossless compression of digital audio. Unlike the popular mp3 format, no information is lost in the compression of an audio file to FLAC format.FLAC is a fixed-point sample format and works with bit resolutions from 4 to 32 bits per sample at sampling rates from 1hz to 655,350. It also handles up to 8 channels of audio, allowing for advanced surround effects. Like MP3 files, FLACs also allow for metadata, such as information about Artist, Album, Release Year and other text information, and even album art to be stored within the file.FLAC compression comprises three components. First, a linear prediction algorithm predicts and then compares adjacent sample chunks. The relative difference between the prediction and the real sample is then encoded. Second. the Golomb-Rice algorithm is used to . Run-Length encoding is also used on blocks of identical samples, storing them instead as a single data value that is repeated x number of times. This primarily saves space when encoding audio tracks containing periods of silence. A wave file encoded to FLAC will typically be reduced to 50-60 percent of its original size, which can be compared to standard quality mp3 encoding which reduces original size by about 90%.For these reasons, FLAC is primarily used by audiophiles and music collectors wanting to store full quality audio while using less disk space. As disk space costs decrease and Internet connection speeds increase, it's likely to become a more popular format for digital audio, also among mainstream users. It has already become a favored audio format of the online torrent scene.H.264, Mpeg-4 Part 10 or AVC, Advanced Video Coding, is one of the most commonly used video compression formats today. It is ubiquitous on the web, supported by YouTube, Vimeo and other streaming sites, as well as satellite and cable TV, Blu-ray Discs and a multitude of other media. It is used for both low bitrate media, like low quality YouTube clips and streaming, and for High Definition Video.H.264 makes use of a plethora of video compression algorithms, techniques and tools, too many to present within the time constraints of this video. But let's have brief a look at two compression features of H.264:Block-Size Motion Compensation, or BMC, partitions the individual video frames into smaller blocks of pixels. Suitable reference frames, or key frames, are fully encoded and the blocks of both subsequent and previous frames are compared to the blocks of the reference frames. When parts of a block in a frame match a block of the reference frame, only the differences between the two blocks is encoded in that frame. The blocks of pixels can also be moved around to compensate for objects moving across the screen. Entropy Encoding is also used in H.264. It works by searching the code for frequently recurring data patterns, which it then replaces with simplified strings of code. To make this process as efficient as possible, the most commonly recurring data strings are represented by the shortest string of code, and vice versa. The resulting quality of an H.264 encoded video depends on the encoding settings used and the source material. An uncompressed 640x360 video with a frame rate of 24 frames per second and 16-bit color will require approximately 33 Megabytes per second. With H.264 encoding, the file size can be brought down to approximately 64 kilobytes per second without significant quality loss. This amounts to a compression ratio of 515:1, which makes H.264 the most efficient video compression standard available today.
Relatively New Free / Open Source Lossless Audio Compression
0100 to 00000000000000000000000000100000 bits / sample 1 To 655 350 samples / second Up to 8 Channels Artist / Song Name / Album / Year / Copyright
1. Linear PredictionPredicts and compares adjacent samples. Records the relative difference. 2. Golomb-Rice Algorithm Encoding Searches the code for commonly repeated chunks of data and replaces them with shorter “symbols”. 3. Run-Length Encoding Replaces identical blocks of code appearing in order (silence for example) with 1 block repeating X number of times
Compression Ratio Uncompressed Video: 33 MB /s @ 640x360 24 FPS 16-bit Color Compression Ratio: 515 : 1 Compressed Video: 64 KB/s @ 640x360 24 FPS 16-bit Color
THE ENDVideo Presentation createdby: Henrik LarssonImage Source for compression Techniques images:Screenshot from “Elephants Dream”Copyright 2006, Blender Foundation / Netherlands Art Media InstituteURLS:http://en.wikipedia.org/wiki/File:Motion_compensation_example-original.jpghttp://en.wikipedia.org/wiki/File:Motion_compensation_example-difference.jpghttp://en.wikipedia.org/wiki/File:Motion_compensation_example-compensated_difference.jpg