TIFF files too big

[ChrisR]

Canon DPP is creating 16 bit TIFF files from 18MP CR2 images to be about 105 Mbits.
Should I be using 8 bit tiffs?

I've done the conversion in Photoshop CS4 before but that's fallen over so I'm using an older version which can't convert them.

Nikon 12MP D700 produces tiffs about 32MB, But Capture ( Nikon's raw converter) makes then about 70MB - compares ok with 105 for the Canon's 18MP.

8 bit 18MP tiffs are 52MB from DPP.
(The DPI setting seems to make no difference to the size)

(Edit typo)

[Charles Krebs]

Chris,

If you are going to use a color space larger than SRBG, then 16 bit is a good idea. If you are going to use a color space larger than AdobeRGB then 16 bit is highly recommended.

If you anticipate doing some rather extreme global contrast changes (as is pretty common in microscope work) then the larger bit depth is a good idea.

[rjlittlefield]

Canon DPP is creating 16 bit TIFF files from 18MP CR2 images to be about 105 Mbits.

That's right on target. 16 bit TIFF takes 6 bytes per pixel, 2 each for RGB. 6*18MP = 108 MB, minus a bit of fluff because "mega" is actually 1,048,576, or 1024K if you're inclined to count that way.

8 bit 12MP tiffs are 52MB from DPP.

I suspect a typo. 18 MP, perhaps? 3*18 MP = 54 MB, minus the fluff.

The DPI setting seems to make no difference to the size

Right. DPI is just a number in the file header. Pixel counts are what matter.

--Rik

[ChrisLilley]

Looks like these are uncompressed TIFF. Tiff also supports a couple of lossless compression methods - lzw and gzip - but of course both your TIFF writing software and your TIFF reading software need to support these. PNG is another alternative, which has lossless compression and supports 16 bits per component RGB. One of the design goals of PNG was to be a bit better than TIFF for 8 and 16bits per component RGB images.

[stevekale]

An 8 bit file will get torn apart during any extensive image editing. As Charles notes, if you are using a broad colour space (I use ProPhoto RGB) then 16 bit is a good idea.

[Pau]

An uncompressed 16 bits image of course has more information than a lossy compressed 8bits jpg, but for me the question is: do I need all that info or some of it is not relevant?

Monitors and most printers are unable to show more than 8 bits/channel, so in most cases the extra info in the final image is not relevant.
The main advantage of 16 bit is the ability to work in the image.

My computer is very outdated and working with such tiff files becomes very slow. In my workflow I do all the adjustements in 14/16 bits (mainly in RAW with Canon DPP), if needed some ones in 16 bit TIFF in PS CS3 and finally I save the image in jpg with max quality, retaining the raw original as backup. After fulling some hard drives with +100Mb slide scans few years ago, now I only store TIFF or PS images in few special cases.

(and note that PS alllows to save Tiff files with LZW losless compression)