PDA

View Full Version : Scanner nearly a densitometer



swmcl
14-Mar-2016, 00:31
Hi all,

I've made a discovery today that I'm quite happy about so I'll share my learnings ...

I replied to an old thread about a densitometer and scanner and I've played a little since then and have come up with a little bit more real-life experience to add to the pot.

OK. I have a flatbed scanner. I have the BetterScanning kit for the scanner. This kit came with some black plastic sheets that one cuts a hole in for the film being scanned. I cut my holes just a little bit bigger than the film sheet itself, say 2mm or so on the length and also the width.

I have not that much experience with scanning but I recently took a photo of a bunch of frangipane flowers. These flowers are velvety to the touch and are a striking white especially when their backgrounding leaves are a dark green - perhaps even more so when I photographed them in very diffuse light on a refreshingly overcast day recently. Suffice to say we have a photo with a genuine white and also one where the shadows are pretty darn dark.

The densities on film are about from 0.20 or 0.25 through 1.45 or 1.5 I guess. I can't measure a minute area on film. I think the round aperture on the densitometer is around 4mm diameter.

So.

I tack the film by its edges such that it is surrounded by a black plastic mask with only a teeny bit of a gap to the film edges and perform the preview on Vuescan.

The settings on Vuescan are very plain. No post-processing as much as possible. No sharpening, no colour correction, nothing. I also make the first scan the one to have the most information. So that means 16-bits per channel and lots of ppi etc. Whether there is more information is of little interest as I can cull in photoshop.

IF I select an area to do the final scan that is only including the actual film - thus cutting off the identifying notches and not including the sticky tape etc. then Vuescan takes my flowers and says, "Oh! This density must be the 'white' of the photo ..." and gives the flower tips a value around 240 or so out of a possible 255. It also takes the clear portions of the film and gives them a value of 2 or 3 or so.

However ... IF I select and area for the final scan that goes well beyond the edges of the film and includes a good chunk of the black mask things change quite substantially. All of a sudden the digital values assigned to my flowers is now around 150 to 160. The histogram has been shifted left quite substantially.

IF I create a mask with a large hole in it nearby to the film cutout so that the scanner's light bar can make a good portion of the image as a 'white' then what happens is I raise my f+b up from 2 or 3 to about 9.

So by including a good amount of what is truly black in my final scan (the mask) and also including a good amount of area without any masking at all, Vuescan all of a sudden shows me values from say 9 through 157 or so. Previously, the software gave values from 2 to 245 or so. This meant I couldn't get any separation on my highlights, they were burnt out and I couldn't make the changes I wanted.

The values of 9 through 157 are very close to the expected results from my scans of a Stouffer step tablet. It means in Photoshop that I can level adjust appropriately without blowing out any values. The initial scan is much closer to reality and leaves room for those scans that have some real density up towards 2.0 on the densitometer. The low value of 9 is also appropriate as the film does have some density after all ! This can also be changed in Photoshop with levels adjustments.

I feel that I have learnt how to make a more 'absolute' scan and feel better that any manipulations are my choice to make. It leaves me feeling very uneasy when the software makes assumptions even if I end up fiddling things such as to produce the same sort of output.

Cheers!

Steve

JaZ99
14-Mar-2016, 01:07
If you are interested in using scanner as a densitometer, check it out: https://sites.google.com/site/negfix/scan_dens