Scanned film better bit depth than digital camera?

[mike_dodd]

Comparing a digital image from 20d with similar image from colour

slide film scanned on ls4000 the 20d image �falls apart� much sooner

when you try some moderately extreme channels adjustments to produce

a good b/w image.

For example to get a very dark sky for good contrast with the fluffy

white clouds the same amount of channels adjustment on the 12(16) bit

20d file ends up with blocks of black and grey where the dark area of

sky is supposed to grade down smoothly to the lighter horizon.

Whereas the 14(16) bit scanned slide has smooth grade without these

blocks although if you zoom in further you can see the film grain.

The bottom line here is that for the same size print film is better

than digital under these circumstances.

 

My question is: am I doing something wrong or is it just a fundamental

limitation of images from current digital cameras that they are only

12 bit compared to some scanned images which have a higher bit depth?

 

The adjustment is in photoshop on 16 bit tiff files from both the 20d

(via a raw converter to convert the 12 bit file out of the camera into

16 bit, and using the Nikon software to convet the 14bit scanner file

to 16bit tiff). I have tried this on a range of files from digital

cameras always with the same result of this nasty blocking once you

push them too far whereas with the scanned film this does not happen.

This is actually a rather common occurrence since I frequently want

to achieve results similar to using orange or red filters to increase

the contrast in b/w landscape photos.

[Ed_Ingold]

The LS-4000 images have twice the nominal resolution as the 20D. You might try resampling the 20D before using the channel mixer. This will blend the digital noise, possibly making it less obtrusive. You might also apply the Photoshop noise filter.

[ellis_vener_photography]

It seems like you are comparing raw fro mthe 20D t othe LS4000 TIFF files. Is this correct?

[godfrey]

Three possibilities come to mind:

<br>

- the 20D RAW images are accidentally being converted to 8bit@channel <br>

- the 20D frames are underexposed, reducing the amount of adjustability available in RAW

conversion. <br>

- the RAW conversion is not optimized to produce the best results

<br><br>

To get the most dynamic out of a digital capture, you must save in RAW format and

expose such that important highlight details are just short of saturation. You then use the

RAW converter to adjust the RGB curve's white point and gamma to render all the other

values correctly, and output into a 16bit@channel RGB file. Digital capture sensors like

that in the 20D generally have at least the dynamic range of a transparency film (6 to 9

stops of dynamic range on the digital, vs 5-7 stops on the very best transparency films, in

my experience) but require that you think in different terms for optimizing exposure for

the subsequent RAW conversion process.

<br><br>

A film scanner is working with the film image, which is already an RGB rendering, and

converting it to a digital 8bit@channel or 16bit@channel representation. Presuming that

the exposure is correct for a particular film, the scanning application should do a good job

of keeping all the values in a useful range.

<br><br>

So what it sounds like to me is that your digital capture exposure and subsequent RAW

conversion process is not yet being exploited to its limits.

<br><br>

Godfrey

[roger_smith4]

Read:

http://groups.google.com/group/comp.periphs.scanners/browse_frm/thread/c575bd68701bfb6d/cffb62f4c7b97749?hl=en#cffb62f4c7b97749

[mike_dodd]

thanks for the responses. however the photos were taken at 100asa on the digital camera so very little issue with digital noise and the images were correctly exposed, no problems with particularly light highlights or very dark shadows. and it was a proper 12 bit conversion not 8 bit by mistake. the areas that have broken up are in the mid tones which is why i think it is something to do with 12 bits not being enough. For example if it is a blue sky and the channels mixer is taking mainly red then could the actual number of levels of grey be rather few?

I would have thought there is some fancy trick in photoshop to get around this problem.