Camera "Scanning" -- Use Di-Chro Head to undo Color Mask?

ColSebastianMoran said:
Godfrey, thanks for the illustration and explanation. I agree, this gives color that's close enough. Nice motorcycles, nice color.

I took the jpg of your negative to look at the histogram, the red channel is well to the R of the others. Red will clip first when shooting a color negative in a camera-scan.

My aim with the cyan/magenta light is to bring all three channels up into the same range on the histogram. So that the inverted, color corrected histogram will have fewer gaps. Any thoughts on this idea?
Click to expand...

Actually, I began to wonder if I hadn't made it too complicated. I looked at the raw histogram of the captured negative image, no processing applied:

35126502052_ece34be7a8_o.png


Hmm. Nothing even close to clipping. (The values have to be up at the end of the histogram on the right to be clipping...) This makes sense: most negative emulsions are low contrast to help overcome the contrast gain of duplication. With that kind of a histogram, the last thing you need to worry about is red clipping.

From that I used a white balance sample on the rebate to pull out the mask:

35126502262_de0379fba8_o.png


and then added exposure to get the rebate up to 87-90% (near white but without saturation) in all channels.

35126502562_98990616db_o.png


Next I used the point curve in Tone Curve panel to do a linear inversion:

35126503032_4810246acb_o.png


Modified that to do a linear inversion with white point and black point clipping (note: since the curve is inverted, white clipping is on the left and black clipping is on the right):

35126503352_211b1e1169_o.png


After that I put an S curve in to jazz up the highlights and push down the shadows for a normal contrast range:

35126503742_e06ffe94dd_o.png


The result is an even better overall color fidelity than my quickly edited CCP:

34482736703_90929c80ef_o.jpg


So there you have it: six steps to bring a negative to positive easily and quickly.

There's a bit of a residual cyan in the image due to the wide-open blue sky. It's easier to process it out by exporting to TIFF and then editing the TIFF, since the controls will all be in positive image manipulation at that point.

G
 
Thanks for this. I'm new to camera as scanner and will likely be scanning a bunch of older color neg material this coming year. Recently scored a BEOON that I'll set the M9 to. I still have a colorhead from the LPL I used when I wet printed, will see what I can rig up. Or what gels I can add to my light table.
 
My preference when opening scans is to choose the file via Open in the finder bar of PS. The resulting box allows one to choose Camera Raw as the file type, and it is very easy to correct color this way. That last image opened this way, and choosing the car with the white balance dropper would be my preferred method.
 
sepiareverb said:
My preference when opening scans is to choose the file via Open in the finder bar of PS. The resulting box allows one to choose Camera Raw as the file type, and it is very easy to correct color this way. That last image opened this way, and choosing the car with the white balance dropper would be my preferred method.
Click to expand...

That's as good a technique as any. :) I use LR exclusively, and do essentially the same thing.

G
 
Off-topic here, but my Washi arrived yesterday! I'm going under the knife tomorrow, but should be up and about enough to shoot the Washi next week. Will re-visit your recommendations before loading the Rollei.
 
Godfrey said:
34482736703_90929c80ef_o.jpg


So there you have it: six steps to bring a negative to positive easily and quickly.

There's a bit of a residual cyan in the image due to the wide-open blue sky. It's easier to process it out by exporting to TIFF and then editing the TIFF, since the controls will all be in positive image manipulation at that point.

G
Click to expand...

Much much better. I had to wait until I got in front of my x-Rite colour corrected iMac before I could comment.
 
Godfrey said:
35126502052_ece34be7a8_o.png


Hmm. Nothing even close to clipping. (The values have to be up at the end of the histogram on the right to be clipping...) This makes sense: most negative emulsions are low contrast to help overcome the contrast gain of duplication. With that kind of a histogram, the last thing you need to worry about is red clipping.
...


The result is an even better overall color fidelity than my quickly edited CCP:

34482736703_90929c80ef_o.jpg



There's a bit of a residual cyan in the image due to the wide-open blue sky.
Click to expand...

Unusual that the sky was so much brighter than the sun. It looks to me like the cyan is from the red left in the capture by your light source, unlike Colonel Moran's. Also, the cylinder head peeking out of the fairing of the duck on the right is seriously badass.

https://www.youtube.com/watch?v=qsB6IBN8gHc
 
I find it horrible looking. It might be better with the mask tuned out with the light source though. I've found most of the C41/digital camera attempts horrible looking up til now. Who knows?
 
Workflow for scanning negatives is very easy - you make a photo or scan in linear gamma 16 bit and apply log curve (I can send) then invert and set black/white point convert to 8 bits. For most films this is all so only three steps. Some films require midtones correction with levels or curves eg. Ektar has too much magenta. All depends on scanner spectral response too. If your photo is in gamma 2,2 space (windows) then log curve approx. is gamma 3 so you need add gamma 0,8 (lighten image). I wrote plugin in 2004 for that that works in older PS.

Orange mask removal through mixing light or exposure is a good thing that maximize bit depth and prevent banding.
 
If your soft does not allow change file output gamma then look for monitor or output icc profile. You can create sRGB gamma 1 profile in Photoshop save and copy from Windows spool folder to your soft. All sensor with one exception (new cmosis log sensors) are linear but software always apply inverse gamma 2,2 for viewing on screen (old transfer function of CRT displays) and because of 8bits storage. If you scan image without taking care then it will be too light after inverting. Image on negative is in logarithmic space that's why we apply inverse transfer curve. Please don't use Photoshop's Levels for gamma change because from version 6 it has slope limiters so can be used only for balancing midtones.

An altrenative is to apply inverse gamma 0,8 to lighten image and have a kind of log output. Beneft - faster less storage.
 
jzagaja said:
Workflow for scanning negatives is very easy - you make a photo or scan in linear gamma 16 bit and apply log curve (I can send) then invert and set black/white point convert to 8 bits. For most films this is all so only three steps. Some films require midtones correction with levels or curves eg. Ektar has too much magenta. All depends on scanner spectral response too. If your photo is in gamma 2,2 space (windows) then log curve approx. is gamma 3 so you need add gamma 0,8 (lighten image). I wrote plugin in 2004 for that that works in older PS.

Orange mask removal through mixing light or exposure is a good thing that maximize bit depth and prevent banding.
Click to expand...

Jack can you elaborate? Maybe give a point by point on how do you make a photo in linear gamma 16bit?
 
brbo said:
Click to expand...

Do you know how to get consistent results with colorperfect? I have it and feel its a complete waste of money. Nearly impossible to get consistent results. It's also VERY confusing to use, completely unintuitive. All the help files are just as confusing. If you can get consistent results with it please share. I would really like to know
 
You must log in or register to reply here.
Back
Top Bottom