JPEG vs. TIFF / 8 bit vs. 16 bit...the tests

[slackercrurster]

Children playing 1900 L.O.C.

Screen shots of crops.

JPEG

TIFF 16 bit

TIFF 8 bit

 

[wiltw]

The web is the great neutralizer in folks being able to evaluate differences in photos...when forums have limits like 1600 pixels in each direction, and JPG image display files are limited to 8-bit monochrome renditions on the web, any differences in files often vanish!

What do YOU see as a difference, if any, slackercrurster?!

 

[loccdor]

Monitors are 8-bit per channel. The purpose of using 16-bit is to edit the brightness range of the original scan (increase its contrast) without seeing the effect of reducing 8-bit values into something smaller like 7-bit (for a 2x contrast boost) 6-bit (for a 4x contrast boost) 5-bit (for an 8x contrast boost).

It's most useful for films of low contrast, such as color negative, high-latitude black and white negative. It's less useful (but not useless) on films of high contrast, like positive slide film, microfilm, or high-contrast negative like Pan F.

 

[koraks]

Screen shots of crops.

8 bit for web output is ample. The difference is relevant once you start doing curve manipulations, especially if they're very pronounced. If you do a straight scan of a negative of print in 8 bit space and the scanning software internally does the curve adjustment needed to get a full-scale output, and that output is going to be the final product, there's no advantage to doing 16 bit scans. Likewise, if you shoot digital and you print digitally, there's in a great many instances (but not absolutely always) no benefit to going beyond 8 bit jpeg. In such cases, the main issue with jpeg is the possibility of compression artefacts.

So your comparison IMO shows that in a great many (but not all) cases, 8 bit JPEG is indeed perfectly adequate.

 

[Pieter12]

All depends on the purpose of the scans. Web posting? Printing? How big? JPEG artifacts start showing at large sizes. The 16-bit TIFF is going have more leeway for fine-tuning adjustments.

 

[blee1996]

I don't want to re-scan my negatives due to the huge time/effort, therefore I always scan at the highest optical resolution and 16 bit TIFF. Scan once, and forget about it. There are always improvements in algorithms in the future to extract every last bit of information out of a good scan.

 

[Kodachromeguy]

I don't want to re-scan my negatives due to the huge time/effort, therefore I always scan at the highest optical resolution and 16 bit TIFF. Scan once, and forget about it. There are always improvements in algorithms in the future to extract every last bit of information out of a good scan.

I agree with blee. Scan at highest optical resolution and bit depth for your own files. Storage space is cheap now. Later convert to 8-bit and resize for web display or printing (ink jet).