Bit depth and Sharpening ?

[Draken]

Ever since i started saving my .tiff files in 16 bit they seem to come out not as sharp as my 8 bit jpgs i find i have to apply twice as much sharpening to my tiff files before saving them for them to retain their sharpness. Is this because they are 16 bit as aposed to 8 ?

any help would be great :-D

cheers,

Mike.

 

[Ls3D]

Doubt it,... are you shooting RAW and have the RAW converter setup to open in PS at 16bits?

BTW - I would do any sharpening in the RAW dialog.

It is not uncommon to re-sharpen an image after resizing. I try to match what I 'know' is in the full resolution image.

I suppose a greater bit depth could be perceived as softer, where an image with fewer bits and more compression could have a higher dynamic range, due to fewer pixels representing the image. But I keep picturing the linear nature of bits per pixel and this is not exactly intuitive.

I've heard that as we pack more photo receptors into a sensor of the same size, the dynamic range will be compromised unless address with logarithms (no big for major manufactures),.. so the bit depth and relative sharpness could be similar to the physics of sensor density.

-Shea

 

[usayit]

So you are starting with a raw file, then applying sharpening, then saving to tiff/jpg, and finally comparing the tiff and jpg. Is that correct? or are you comparing the jpg from camera with a RAW converted to tiff? I doubt it is bit depth.

 

[Draken]

Yeah shooting as RAW ofcourse and opening in ACR then post processing then saving 1 x tiff and 1 x jpg and jpg seems alot sharper unless i apply twice the sharpness to my tiffs

 

[Garbz]

Post an example for all to see.

 

[JerryPH]

I'd want to see an example of once each sharpened TIIF vs JPG as well. This is certainly NOT happening here. If anything, the moment I drop from 16-bit to 8-bit, I see, very clearly, a small softening of the JPG on screen using PS CS3.

 

[SilverGlow]

Perhaps the worse thing one can do is sharpen very early in post processing. I would never apply sharpening in the raw dialog. In addition, the problem with sharpening in raw is that the sharpening is applied to the entire image, instead of just in the areas required. Sharpening where it is not needed damages the image, and can introduce digital artifacts.

Sharpening should be done as the last step because the amount and method of sharpening is a function of how the image will be used. This is why a completely mastered image must never be stored sharpened.

It is far better to archive mastered images with no sharpening.

So when does one sharpen? Just before you output the image. For prints sharpening is a function of the size of the print. For publishing on the web, sharpening is applied different still. Instead of applying sharpening to the entire image, it would be more beneficial to apply it only to those parts that require it, like the eyes, teeth, whatever is required per the composition. And to do it with the most benefit, apply it to a duplicate layer, then use a mask to paint in the effect only on the parts of the image that require it. This sounds complex, but with a little practice, one can do all this very fast and easy, and to stunning results.

If you archive a mastered image already sharpend, then you have applied an amount that cannot be undone, and will not necessarily be appropriate for all forms of output and all sizes of output for that image.

Moreover, it is best to archive images in the highest bit depth possible. You might be asking, why save a 8 or 12 bit image in 16 bits? Because when you apply sharpening and other tweeks later, a deeper bit depth will accommodate the rounding errors that come out of digital image math. Round errors mean combing, blocking and other bad digital artifacts.

As to bit depth, this has less effect on sharpening then pixel density. Although a lesser bit depth could give the perception of more "sharpness" because as one drops in bit depth, contrast often increases, and that increase in contrast is perceived as "sharper".

 

[SilverGlow]

Doubt it,... are you shooting RAW and have the RAW converter setup to open in PS at 16bits?

BTW - I would do any sharpening in the RAW dialog.

It is not uncommon to re-sharpen an image after resizing. I try to match what I 'know' is in the full resolution image.

I suppose a greater bit depth could be perceived as softer, where an image with fewer bits and more compression could have a higher dynamic range, due to fewer pixels representing the image. But I keep picturing the linear nature of bits per pixel and this is not exactly intuitive.

I've heard that as we pack more photo receptors into a sensor of the same size, the dynamic range will be compromised unless address with logarithms (no big for major manufactures),.. so the bit depth and relative sharpness could be similar to the physics of sensor density.

-Shea

Why do you think bit depth is like the number of pixels? It is not. You can have the exact same pixels between two images, and different bit depths. In addition, there is no correlation between bit depth and compression.

 

[Ls3D]

I don't think that, which is why I eluded to 'bits per pixel' and 'linear nature' of the problem.

IME - When stepping down to jpg or another compressed format, the apparent dynamic range can increase with the same number of pixels as the encoding prefers similar pixels values and rapid changes. Run length encoded, bilinear, trilinear it is all sampling down.

And,. any oversampling is welcome in my book.

-Shea