Discussion in 'Digital Discussion & Q&A' started by SamuraiAZ, Jul 13, 2008.
How big are they usually?
8 -10mb from a Nikon D80. That is in Nikon's NEF format. Others will vary from brand to brand. But about there.
good to know, I just ordered a d60
The ones from My Samsung GX-20 are ~23mb each.
Depends on the camera and even on the image.
Yes... Nikons NEF(RAW) files are compressed on the smaller DSLR's ... which explains the difference from xscodes 20meg on the Samsung.
Someone please correct me if I'm wrong, but I think uncompressed RAW's of whatever format are roughly 20meg depending on the image??
I think it's dependant on the sensor/camera. My RAW files (Canon 20D) are 6.8 to 9 MB.
A photographer I know is using a 1Ds mk III (22 MP), and the RAW files are 30-40 MB...he had to upgrade his computer's processor to handle them without bogging down too much.
well that makes perfect sense ... the larger the sensor/mp the more information there is to store....
i thought there was some form of compression applied as well.... i remember asking why my 8meg nef's were turning into 28meg monsters when I converted them to tiff and compression on the nef's was the source of discrepency.
That doens't sound right. I've always heard that RAW files were not compressed...at least not the image part of the file. The other info might be. TIFF and PSD files are always larger than the RAW file.
The way I understand it...a RAW file isn't technically an 'image file' like JPEG & TIFF are. It's the RAW data from the sensor...which needs to be 'converted' before it can become an image file. We can see & preview the RAW files because they have an embedded JPEG image. So once you use the RAW data to create an actual image and save it as an uncompressed image (TIFF)...then you see how big of a file it becomes.
There is suprisingly little official information on this.... I had a hard time googling it... lots of forum chatter but nothing from reliable source....
The compressed NEF is optional I guess on the big camera's D200 and upwards. I don't really understand how something compressed can be lossless but whatever. I'm back to shooting jpeg so i don't have to think about it. I think (but am not sure) that on Nikon's smaller DSLR's the compression is automatic and not optional like on the D200 upward.
As described by Ken Rockwell -
[FONT=Arial, Helvetica, sans-serif]Compressed Raw [/FONT] [FONT=Arial, Helvetica, sans-serif]Since raw data is just data it can be compressed losslessly to save a lot of space. Since it is mostly random you get the usual 2:1 compression per Lempel-Ziv and recover every single bit later. [/FONT] [FONT=Arial, Helvetica, sans-serif] Nikon uses lossless compression of the data, but is clever enough to apply some of it's curve-shaping to curve 12-bit data to 10 bit data before storage. This mapping scares innocent laypeople. It's invisible. As you can see on my D200 examples, JPG compression is invisible unless you set it all the way down. The only potential defect in this mapping is the addition of noise far less than the noise already present in the sensor, so it doesn't matter. [/FONT]
As described by Thom -
[FONT=Verdana, Arial, Helvetica, sans-serif]Nikon's Compressed NEF compromises highlight bits slightly. [/FONT][FONT=Verdana, Arial, Helvetica, sans-serif]Nikon claims that the old Compressed NEF format (the D3 and D300 now support another, lossless, form of compression) is visually lossless. Basically, they play off the ability of our eye's inability to resolve small differences in bright areas by throwing away some information. If you can't distinguish between a value of 14,230 and 14,238, why store values of 14,231 through 14,237? In practice, this works without penalty unless you make large changes to highlight data. Where I see small, resolvable differences is in something like a wedding dress detail after large amounts of post processing and sharpening are applied. But in general, you can shoot Compressed NEF without worry.[/FONT]
If you google Canon Raw Compression you get a lot of hits too. I guess they do it as well.
Let me give you an example:
"The sixth sheep's sheek's sixth sheep's sick" could be compressed as:
"sixth,*,shee,^ The * ^p's ^k's *p's ^ sick" That's 3 bytes less just by inspection, and it is lossless because the original data is fully recoverable to what it was. 3 bytes is also not much. Another example would be RunLegth Encoding:
becomes "12W1B12W3B24W1B14W". This type of encoding is supported by the TIFF file format and again is lossless meaning that the exact original can be recovered.
Obviously the pattern recognition by words would not work well on images, but none the less some pattern recognition can reduce the filesize. Consider the above RLE compression applied to an image with a lot of clipped highlights. Also think of ZIP files. If zip were lossy compression like JPEG which approximate the original data the program simply would not run.
Now as to the file size differences, consider the above statement by ken rockwell mentioning that the RAW data without compression is 12bit (That's a 16mb file on the D200). TIFF being an image format is usually 8bit or 16bit, so when an uncompressed RAW at 12bit is converted to an uncompressed TIFF at 16bit file size naturally follows.
Also if you have a head for numbers I suggest reading this: http://www.majid.info/mylos/weblog/2004/05/02-1.html
Ken Rockwell may believe that it makes no differences because his flawed examples don't illustrate the problem, and that his word from some Nikon rep is better than the actual engineering examples, but it quite literally means the difference between the highlight recovery slider in ACR and Lightroom working or not working for borderline cases, and could mean one less forum poster saying, "nice job but you shouldn't clip the highlights".
Another reason for the size difference: Raw files only have one channel of data because there is no colour information in the file. Once they are converted to colour TIFF or PSD files they have three channels of data (eg RGB etc).
Separate names with a comma.