Much appreciated Godfrey ! Kenneth Waller
----- Original Message ----- From: "Godfrey DiGiorgi" <[EMAIL PROTECTED]> To: "Pentax-Discuss Mail List" <[email protected]> Sent: Friday, June 16, 2006 10:34 AM Subject: Re: Where Do All the Pixels Come From (was: Shooting Digi in JPEGMode) > Much of this has been answered already through several emails, I > thought I'd try to bring it all together and add a little more of the > mathematics... > >> ... So, if JPEG loses, or throws away, a lot of information, why >> are the files >> when converted to TIFF (or PSD) so large? Where does the extra >> info come from? ... > > In the camera... > The image is not created in JPEG format then converted to TIFF. The > order of operations is > RAW sensor data -> 8bit RGB rendering -> Compressed 8bit RGB rendering > > The RGB rendering is what the camera uses to create the TIFF file. It > is larger than the RAW file because the RAW file uses 12bits to > describe each photosite state where the TIFF file uses an [r,g,b] > triplet of three 8-bit bytes to describe each pixel, where the number > of pixels is the same as the number of photosites. That's 24 bits vs > 12 bits to describe the same thing, so the uncompressed TIFF file has > to be at least twice as large. There is additional overhead in the > TIFF file's structure as well. > > The JPEG rendering is the compressed 8bit RGB rendering. It's > resulting smaller file size is a matter of compression coding, > packing the [r,g,b] pixel array values into a more compact form of > numbers that can be 'unpacked' back into a reasonably accurate > rendering of the RGB image according to an algorithm. > > Out of the camera... > Taking a RAW file and performing RAW conversion to an 8bit TIFF file > does the same thing that doing this in the camera does. If you > convert a RAW file to a 16bit TIFF file, each of the pixels is > assigned an [r,g,b] value with 16bit values instead of 8bit values, > which allows for more precision in manipulation ... the resulting > data size is twice again as large as an 8bit TIFF file. Taking a JPEG > file and converting it to a 8bit TIFF file simply reverses the JPEG > packing back into the expanded, simple 8bit per channel [r,g,b] > pixel description array. > > Precison 8bit vs 16bit: > Say you look at a pixel value from an 8bit TIFF file and you get a > triplet like [128, 128, 128]. That represents the amounts of R, G and > B that are added together to produce the total intensity and color > value of that pixel, on a scale of 0-255 possible values in each color. > > If you were look at the same pixel in a 16bit TIFF file rendering of > the 12bit RAW data, the relative values of the channel assignments > would be the same, but you have 16x as many numbers available to > describe the values in the original RAW file which is then scaled to > a representation in a discrete numeric space 8x larger (0-4095 in the > RAW data, 0-32767 in the 16bit RGB channel (the topmost bit is not > used so it's actually 15 bits of data)). Only 4096 of the values in > the 16bit space are actual photosite RAW values, they're fitted into > the larger space because current computing machinery manages 16 bit > numbers with greater efficiency than 12 bit numbers, in general, > *and* because as you perform Real or Discrete valued operations on > these numbers, there are more numbers to represent the results, thus > greater precision and less likelihood of clipping or round off errors. > > An illustrative example would be the 'digital' volume knob on many > modern automobile radios. As you turn the knob, the display might > display a range from 0 to 10, or it might display a range from 0 to > 100. The actual analog amplitude of the volume is the same, but it is > represented from none to max in two different resolutions... with the > 0-10 representation you only get to set one of ten steps, with the > 0-100 representation, you can set a lot more precisely the exact > volume you want, with 10x the steps between values. > >> ... Further, when viewing a high quality JPEG in Photoshop, it >> shows the file >> size in the status bar to be about the same as the TIFF TIFF (or >> PSD) file >> made from that JPEG. ... >> And why does Photoshop show the smaller JPEG file to be the size of >> the larger TIFF or PSD file. > > Photoshop's description of image size is a description of the size of > the uncompresed, actual pixel value array. If the array is packed in > a JPEG or 'compressed LZW' TIFF file, it unpacks the values into an > uncompressed array before reporting the size, which means that a PSD, > TIFF or JPEG image with the same number of pixels and the same bit > depth will show as the same size. > > Godfrey > > > On Jun 15, 2006, at 6:30 PM, Shel Belinkoff wrote: >> ...Taking an image shot in highest quality JPEG on the DS results >> in a file >> size of 1,900kb. Doing absolutely nothing to it but converting to >> a TIFF >> results in a file size of 17,600kb. Converting that file to 16-bit >> doubles >> the size. Now, making the same shot using RAW results in a file >> size of >> about 10,000kb, and converting it to TIFF results in a file size of >> approximately 35,000kb. >> >> I've noticed the same behavior with my little Sony. It will >> produce a TIFF >> and a JPEG simultaneously, and when the JPEG is converted to a TIFF >> it's >> the exact same size as the original TIFF. >> >> Further, when viewing a high quality JPEG in Photoshop, it shows >> the file >> size in the status bar to be about the same as the TIFF TIFF (or >> PSD) file >> made from that JPEG. >> >> So, if JPEG loses, or throws away, a lot of information, why are >> the files >> when converted to TIFF (or PSD) so large? Where does the extra >> info come >> from? And why does Photoshop show the smaller JPEG file to be the >> size of >> the larger TIFF or PSD file. > > > > > -- > PDML Pentax-Discuss Mail List > [email protected] > http://pdml.net/mailman/listinfo/pdml_pdml.net -- PDML Pentax-Discuss Mail List [email protected] http://pdml.net/mailman/listinfo/pdml_pdml.net
