Why uncompressed TIFF is used?

Linas

Forum Enthusiast
Messages
381
Reaction score
0
Location
Dublin, IE
Folks,

I do not understand why uncompressed TIFF is used in cameras. Ok, jpg compression is lossy, but why tiff is uncompressed at all, while non lossy compression can be used (like LZW)? In this case files would be 2 times smaller (or 15% smaller if there are so many details). The huge file sizes and lag between shoots make this format allmost useless (except those cases, of course, where you can wait).

Thanks,
Linas
 
That's right!
It sounds like a candidate for firmware update...

Together with the possibility to set upto 1/1150 sec manually and maybe a few bugs fixed (manual focus on infinity sometimes not focusing) etc...
I wonder if it's real or just a naive dream of another hot-blooded customer...

Martin

ps: I have been using my C2100 for 5 months already but I've started shooting TIFF images just now. (jpeg artefact? ...what's that? :)
Folks,

I do not understand why uncompressed TIFF is used in cameras. Ok,
jpg compression is lossy, but why tiff is uncompressed at all,
while non lossy compression can be used (like LZW)? In this case
files would be 2 times smaller (or 15% smaller if there are so many
details). The huge file sizes and lag between shoots make this
format allmost useless (except those cases, of course, where you
can wait).

Thanks,
Linas
 
Folks,

I do not understand why uncompressed TIFF is used in cameras. Ok,
jpg compression is lossy, but why tiff is uncompressed at all,
while non lossy compression can be used (like LZW)? In this case
files would be 2 times smaller (or 15% smaller if there are so many
details). The huge file sizes and lag between shoots make this
format allmost useless (except those cases, of course, where you
can wait).

Thanks,
Linas
Linas;

Canon has come out with a format that is less lossy and much more compact than Tiff, howeverit is proprietary. Olympus is using a format that already exists and is not proprietary. LZW is another proprietary product, and unless you pay to use it, you cannot market it commercially. The Canon uncompressed format is not usable by some of the editors, because the software developer has chosen not to pay to include a special hook in their program that allows them to interpret this format. Bottom line, it is all about money. Some applications require an uncompressed graphic that has no artifacts (for those who are not sure what an artifact is, in very simple terms, it is a best guess at what should have been at a particular pixel location, because the decompression of a graphic that has been compressed is not 100% accurate), therefore Olympus has provided a way to achieve that in the digital cameras that they sell. Tiff is public domain (costs nothing to use), and supported by just about every photo software package in existance.
MDiamond
 
This from the infameous dead pixel test document.

"Dead Pixel Test – Documentation

This tool allows you to check your camera for dead pixels. It also reports hot pixels which depending on the length of the exposure and the threshold used in the test can be attributed to normal noise.

How it works

Make a lens cap shot. Close the lens with your lens cap, also make sure to close the viewfinder so no light can get into the camera. Now make exposures with different shutter speeds ranging from 1/30, 1 second, 2 second and more.

Best results will be with TIFF files or other non-compressed or loss less compressed files as JPEG artifacts will introduce more dead/hot pixels than you really have."

I guess tiff will give the highest quality you can acheive with a digital camera. Has any one any pictures showing the difference between a .jpg and a tiff file? Is it visable?
Folks,

I do not understand why uncompressed TIFF is used in cameras. Ok,
jpg compression is lossy, but why tiff is uncompressed at all,
while non lossy compression can be used (like LZW)? In this case
files would be 2 times smaller (or 15% smaller if there are so many
details). The huge file sizes and lag between shoots make this
format allmost useless (except those cases, of course, where you
can wait).

Thanks,
Linas
Linas;
Canon has come out with a format that is less lossy and much more
compact than Tiff, howeverit is proprietary. Olympus is using a
format that already exists and is not proprietary. LZW is another
proprietary product, and unless you pay to use it, you cannot
market it commercially. The Canon uncompressed format is not
usable by some of the editors, because the software developer has
chosen not to pay to include a special hook in their program that
allows them to interpret this format. Bottom line, it is all about
money. Some applications require an uncompressed graphic that has
no artifacts (for those who are not sure what an artifact is, in
very simple terms, it is a best guess at what should have been at a
particular pixel location, because the decompression of a graphic
that has been compressed is not 100% accurate), therefore Olympus
has provided a way to achieve that in the digital cameras that they
sell. Tiff is public domain (costs nothing to use), and supported
by just about every photo software package in existance.
MDiamond
 
ps: I have been using my C2100 for 5 months already but I've
started shooting TIFF images just now. (jpeg artefact? ...what's
that? :)
Sure glad your subjects will hold still for that long and that you have that much patience.

Can you post an example (a small crop) of the same scene shot in both TIFF and SHQ and point out the differences?

John
 
Michael,

As i have noticed tiff includes more hot pixels than SHQ jpg. It's probably because it takes longer to write (more noise from other camera components), or probably jpg eliminates those pixels (when there is not much). There was a thread about that:
http://www.dpreview.com/forums/read.asp?forum=1008&page=1&message=1339607

From other side tiff does not include artifacts (maximize your jpeg up to 300% and you will see them, more on HQ, less on SHQ...)

Cheers,
Linas G.
 
Mike,

Ok, LZW compression is proprietary, but other compression types (non lossy!) can be used (tiff header, if i remember correctly, includes information, how data is compressed inside the file)... Other non lossy and non proprietary formats can be used also... PNG for example:
http://www.w3.org/Graphics/PNG/

So... whats the reason to have UNCOMPRESSED TIFF, if there can be NON LOSSY compressed TIFF (Ok, not with LZW, but there are other compression methods...), or PNG or whatewer...

It looks, that Oly and other brands does not care much about that...

Cheers,
Linas
 
Michael,

As i have noticed tiff includes more hot pixels than SHQ jpg. It's
probably because it takes longer to write (more noise from other
camera components), or probably jpg eliminates those pixels (when
there is not much).
The time to write the TIFF to the memory card will not introduce more hot pixels. That's like saying saving a Word document to a floppy instead of the hard drive can cause more errors in the file. Neither of these operations should cause error unless there is a hardware problem.

The reason to use TIFF instead of JPG is that JPG compression and affect the appearance of the hotpixels by applying compression to them.
From other side tiff does not include artifacts (maximize your jpeg
up to 300% and you will see them, more on HQ, less on SHQ...)
That is mostly true. It depends on the content of the image if JPG artifacts will be visable when you enlarge the image. Any large area of similar tone (a blue sky for example) can show artifacts. Othere more complex areas should not be visably different.

John
 
Linas;

While I totally agree with the need for a lossless, compressed format in digital cameras, I'm not sure that PNG is the format to go for. I checked the link you included and in ther first paragraph describing the features of the format was the following line:

"Sample depths range from 1 to 16 bits."

Sorry, but 16-bit color (65,536 possible simultaneous colors) just isn't adequate for digital photography. It's better than GIF, which can only do 256 simultaneous colors, but it just isn't the same as 24-bit color (16,777,216 possible simultaneious colors).

The other thing I'm wondering is this. Since most optical scanners have moved beyond 24-bit color into 32-bit, 40-bit and even higher color depths, why haven't digital cameras? I know that the file sizes will balloon, but for serious digital photographers (which probably doesn't count us Uzi owners), I would think that the greater color fidelity might be as desirable as lossless file formats and higher pixel counts.
Mike,

Ok, LZW compression is proprietary, but other compression types
(non lossy!) can be used (tiff header, if i remember correctly,
includes information, how data is compressed inside the file)...
Other non lossy and non proprietary formats can be used also... PNG
for example:
http://www.w3.org/Graphics/PNG/

So... whats the reason to have UNCOMPRESSED TIFF, if there can be
NON LOSSY compressed TIFF (Ok, not with LZW, but there are other
compression methods...), or PNG or whatewer...

It looks, that Oly and other brands does not care much about that...

Cheers,
Linas
 
John,
The time to write the TIFF to the memory card will not introduce
more hot pixels.
Maybe yes, maybe no...

It's not because it's TIFF, it's because TIFF is bigger... so it takes longer to write it to SM card...camera components will be busy for longer time... this will generate more electromagnetic noise and more heat... as a result this will impact CCD.
That's like saying saving a Word document to a
floppy instead of the hard drive can cause more errors in the file.
I'm sure that to write word file to floppy can cause errors a lot more often
than to write it to hdd... :-))
Neither of these operations should cause error
Yess!
unless there is a hardware problem.
Here it is... CCD accumulates noise... and it look like that it depends on camera uptime (it will take more time to take 5 sequential shots with tiff than with jpg) ... :
http://www.dpreview.com/forums/read.asp?forum=1008&page=1&message=1382712
The reason to use TIFF instead of JPG is that JPG compression and
affect the appearance of the hotpixels by applying compression to
them.
Here i agree with you!

Cheers,
Linas
 
Jerry,
I'm not sure that PNG is the format to
go for. I checked the link you included and in ther first paragraph
describing the features of the format was the following line:

"Sample depths range from 1 to 16 bits."
Yeah, you are right... i have put "png" here just to show that other non lossy and non proprietary formats exist (to be honest i was thinking about those formats and the first idea was PNG... i did not realized that it's 16 bit only...)
Sorry, but 16-bit color (65,536 possible simultaneous colors) just
isn't adequate for digital photography. It's better than GIF, which
can only do 256 simultaneous colors, but it just isn't the same as
24-bit color (16,777,216 possible simultaneious colors).
Agree.

Cheers,
Linas
 
The time to write the TIFF to the memory card will not introduce
more hot pixels.
Maybe yes, maybe no...

It's not because it's TIFF, it's because TIFF is bigger... so it
takes longer to write it to SM card...camera components will be
busy for longer time... this will generate more electromagnetic
noise and more heat... as a result this will impact CCD.
The CCD is only sensitive to noise during the actual exposure. The electronic process of reading out the CCD is very rapid and is not affected by the file type that you are going to be using.

Broadly speaking, the sequence of operation in a digital camera goes roughly like this:

1) Close the shutter

2) Read out and discard the CCD

3) Open the shutter

4) After the exposure time, close the shutter

5) Read out the CCD to the internal high-speed buffer (DRAM)

6) Apply in-camera sharpening, white balance and exposure compensation algorithms to the image in the buffer

7) Apply selected compression algorithm to the image in the buffer

8) Transfer the final file to the storage medium

So if step 7 uses a lossless compression, step 8 will take a lot longer since the storage medium (SM in our case) is a lot slower than the internal buffer. However noise can only enter the image during steps 3 and 4. True, the camera and batteries may generate a small amount more heat becaise of the extra activity during step 8, but that should disipate before the next shot.

Not all digital cameras have a separate shutter. In that case steps 1,3, and 4 wiuld be replaced by "4. Wait for exposure time."

I did not include such items in the above sequence as reopening the shutter after step 5 and continuously sampling the CCD to drive the EVF, exposure, auto-focus, and white balance systems. These activities will be going on in parallel withe the processing of the image. Again, they are not related to the type of compression.

Fritz
 
Fritz,

...
7) Apply selected compression algorithm to the image in the buffer

8) Transfer the final file to the storage medium

So if step 7 uses a lossless compression, step 8 will take a lot
longer since the storage medium (SM in our case) is a lot slower
than the internal buffer. However noise can only enter the image
during steps 3 and 4.
And if step 7 does not use compression at all (this is the case),
this will take even more longer... and will generate more heat...

Noise will enter the image on next shot (if break between shots will not be long enough).
small amount more heat becaise of the extra activity during step 8,
but that should disipate before the next shot.
Should, but it does not... if there is not enough time....

Again... this is not about TIFF as such format... This is about HUGE UNCOMPRESSED TIFF SIZE... If tiff would be non lossy compressed (loosy is jpg, so we do not need loosy tiff) heat (i think it's because of heat) impact would be smaller...

From where this noise (that depends on camera up time, not on ISO settings or shooter speed) can come?

The only two things that come into my mind are the HEAT and electromagnetic fields...

Thanks,
Linas
 
PNG has a 24 bit format, the problem is that most browsers at this time do not support PNG
Robs
I'm not sure that PNG is the format to
go for. I checked the link you included and in ther first paragraph
describing the features of the format was the following line:

"Sample depths range from 1 to 16 bits."
Yeah, you are right... i have put "png" here just to show that
other non lossy and non proprietary formats exist (to be honest i
was thinking about those formats and the first idea was PNG... i
did not realized that it's 16 bit only...)
Sorry, but 16-bit color (65,536 possible simultaneous colors) just
isn't adequate for digital photography. It's better than GIF, which
can only do 256 simultaneous colors, but it just isn't the same as
24-bit color (16,777,216 possible simultaneious colors).
Agree.

Cheers,
Linas
 

Keyboard shortcuts

Back
Top