When to use 8 bit? When to use 16 bit? Regarding TIFF files.

I know that you can either use 8 bit or 16 bit when using TIFF files. What are the differences? What would you use the different formats for, for example, landscapes, portraits, or sports?

An 8bit file can use 256 distinct tones to represent a colour at each pixel position. 16bit can represent 65,536 values (that's 1111111111111111). In theory it makes for a smoother image with more scope in editing, but to be quite honest I can never tell the difference. It's like everything else in digital photography (and problem film too) individual small improvements may add up to a better image.

16 bit TIFFs are at least twice the physical size; eg, 40MB against 20MB.

Hi Pono,

You almost certainly won't see it on screen as I believe most screens are 8 bit (if that).

The reason to use/save with 16 bit files is when you intend to edit the image further. If you do it from 8 bits, you soon begin to se problems due to 'rounding errors' resulting in smooth tone transitions becoming stepped like contours on a map. How visibly pronounced this is depends on the picture content and what sort of processing you were doing.

If the output is just for display on a monitor or printing after all processing is done, 8 bit is fine which is why these things are typically done with jpgs, which are also 8 bit.

Cheers,

With some editing software, some procedures only work with 8 bit images.

If processing images with large areas of gradually changing tones, such as blue skies, you can soon see the difference between working in 8 or 16bit. With 8bit you start to see posterization, or banding, far sooner than with 16bit.

Quote:

---

Originally Posted by Geoff F

With some editing software, some procedures only work with 8 bit images.

---

Indeed, Geoff :(

One reason why I took the plunge with CS5 over Elements.

Is there not something about Analogue to Digital conversion (A/D) lurking in the background? Do digital cameras actually have A/D converters that can do 16 bits? I had the impression that 12 or 14 was the maximum. Thus, a TIFF 16-bit file will be based on say 14 bits of data. That's certainly better that 8-bits for precision, but not actually 16-bits. For most purposes this makes no real difference, it's just the pedant in me that says it might make a "bit" of difference now and then.

Cheers

David

Quote:

---

Originally Posted by David

Is there not something about Analogue to Digital conversion (A/D) lurking in the background? Do digital cameras actually have A/D converters that can do 16 bits?

---

I believe that some MF digital backs do 16 bit, but it's a bit of a moot point, as Photoshop only does 15 anyway.

What the image starts as (12 or 14 bits) is, after almost any PP editing, irrelevant.

All the pictures are digital, so essentially just a series of numbers, think of the value of that number at one single pixel and consider what happens when you levels/brightness/whatever adjust it, then apply some other effect, all you are doing is adding, subtracting and in some cases multiplying and dividing that number - this can easily result in more bits.

A crude anaology is to divide 6 by 4 and say the answer is 1 - which it is if you are using whole numbers (integers) only, because of the digit-cropping error loses the 0.5.

So; divide 6/4 = 1, now consider what happens if we multiply by 4, it = 4, but if we had more digits it would (and should) be 6 (6/4 = 1.5, 1.5 x 4 = 6).

If you consider that 7/4 and 5/4 all digit-crop down to 1 and multiply back up to 4 (instead of going back to 5 and 7) you can see how the posterisation effects creep in.

That's why 15/16 bits is useful over 8 bits when any image editing is intended.

By using 15/16 over the 12 or 14 generated by the sensor, I guess it gets a bit more accurate still, but more importantly, it is standardised (and called "16 bit" regardlesss what goes on 'inside').

Since the output media is going to be display or print with typically far less dynamic range than 15/16 bits can produce, that's why, when all you're gonna do next is display or print, 8 bit (even jpg) is perfectly adequate.

Well that's my belief anyway, based on video digitisation processes I was trained on waaay too long ago to remember the precise figures. I am not learn-ed in the international standards on jpg/tif, et al, so I may be slightly off in the detail, but I believe the concepts are correct as a working approximation to understanding.

There are lots of why's and where-for's I have left out above to keep it simple, if you have any questions, just ask.

Final thought; another place to look is Sean's tutorials, I bet he's got it right - and his research will be 20+ years more recent than mine :)

Cheers,