Cory Papenfuss wrote:
As a long-time user of the TIFF format for somewhat unrelated
purposes, I feel most inclined to comment on this note:
Tiffs don't have any of the post-processing advantages of RAW
Which is obviously untrue, since TIFF (unlike JPEG) won't usually
compress data by throwing actual image data away, and also has the
capability of storing 12 bits-per-channel (well, actually, it will
have to be be 0-padded 16-bit, but...)
Tiffs are a tricky subject, because they can have different
attributes like you say.
Yes that's the advantage *and* disadvantage of the format...
I would be surprised if in-camera produced TIFFs are 16 bits. They
would be extra-huge if they were. Pentax's RAW files basically *are*
TIFFs (zero padded 12->16 for the -D, and packed 12->12 for the -DS).
The difference is that only a single plane of 6megapixel, 12-bit data
is necessary in RAW, but an interpolated version for R, G, and B is
needed for TIFF.
I started wondering now if you might actually store the "raw"
representation in a TIFF file, since different colour models (not just
RGB) are supported, but perhaps not quite. You can save data as
luminance/crominance, but I guess that's not exactly the same thing. And
only RGB images would be considered baseline TIFF, I think.
Ignoring compression (as Pentax likes to do), a 16-bit linear RGB
TIFF file would be 2 times as large as the 8-bit gamma-correct TIFF, 3
times as large as the padded RAW, 4 times as large as the packed RAW.
Yes. I suspect that the difference would be a lot smaller with
(lossless) compression, though...
If one didn't do white-balancing on the 16-bit linear TIFF, the only
thing RAW has as advantage would be the potential for a more advanced
Bayer interpolation.
Large tiffs may be an acceptable compromise. It's much bigger
than it needs to be, however, and does lose a slight bit of
flexibility.... not nearly so much as 8-bit gamma tiff or JPEG, though.
My point exactly. While the referenced text suggests that TIFF has no
advantages over JPEG...
- Toralf
