Converting film-negatives to usable images is not as trivial as one might think.
I have spent some time constructing a flow that can do it correctly.
If this is a topic that has your interest, you may have seen how there is an abundant amount of articles on the internet on how to do colorfilm-conversion. That's good, until you discover that there seems to be as many different methods (yielding different end-results) as there are articles.
Some people will, to avoid drowning in articles, opt for a plugin (there is at least one very popular one, as far as I could find. I haven't tested it, but it gets good reviews. It's payware)
Until now
Because now you can use PhotoReactor
Unfortunately the champagne stays in the wine-cellar for just a bit longer. There is one major drawback to the PhotoReactor method I have come up with; bit depth.
Say you have scanned your film-negative in linear mode and import it into 8bit-PhotoReactor. In that case you will have between 5 and 7 bits available to work with*. Obviously this will not give as clean an image as if you could import it into a 16bit (or more) PhotoReactor.
*
blue ~5bit
green ~6bit
red ~between 6 and 7bit
*
However, the results can still be useful, but the bit-limitations will show (more on some negatives than on others, which is related to the color of the mask)
There's one more issue you might want to pay attention to, at least if you want extreme precision, and that is the film-response curve. In this flow there is no correction for individual film-response curves (for normal use this is not a problem, since these curves are very similar to each other, but for some connoisseurs it might be relevant)
Now, where this flow WILL (hopefully) excel, is in the fact that I have included TWO methods rather than just a single one
Method #1 will balance colors to give you a color-neutral image (a correct conversion)
This means blacks will be black, and whites will be white.
Method #2 will balance only the middle-grays, which has the neat side-effect of introducing a dual-tone effect that can be very aesthetically pleasing. The shadows will trend toward bluish tones, and the highlights towards golden/yellowish tones (you probably know this 'teal/orange'-like tone-scheme if you watch movies)
The strength of method2 depends on the film-negative's inherent color-balance. The more orange the mask is, the stronger the blue shadows will be. So this method can be hit or miss depending on the film-negative used.
Using the flow can be tricky, but here's a rundown:
First you must acquire a scanned image of your film-negative. This can be either linear or with the standard gamma2.2 applied.
The first node (levels node) in the flow applies a gamma of 2.2 to the source-image, so if your scan is NOT linear, then deactivate this first node (or correct the gamma setting such that you get an image with gamma 2.2)
(In the attached flow-snapshot the source-image is a linear scan of a Kodak film-negative)
After this first node, the flow splits into 2.
For a neutral image you follow the top flow-path.
For a tinted image you follow the bottom flow-path.
The procedure for both paths is similar. Here's the one for the neutral image:
Connect the top-scope (labeled 'neutral') to the layers-node with the multiply blend-mode selected (drag the output from the multiply node to the scope)
Now you will see that the red, green and blue histograms on the scope are unbalanced.
To counter this we will apply a counter-mask (a color that nulls out the orange mask of the film-negative)
We do this by adjusting the highlights cutoff in the red and green levels-nodes found on the left side (in each levels node the highlights cutoff is the control at the bottom on the right-hand panel)
The point is to adjust the highlights cutoff until the RGB bars on the scope align at each others top (basically you will move the red and green bar to align with the blue bar), which will place somewhere below the value of 128 on the scope (depending on the mask color of the film-negative)
When these 3 bars are aligned you move onwards in the flow-path to the levels-node at the top (just left of the output-node) where you adjust the shadow and highlights INPUT-levels to select the part of the histogram containing your image.
Finally you adjust the whitebalance (the kelvin-node) if needed (to do this connect the scope to the kelvin-node, as is shown in the flow-snapshot, so you can see if the RGB bars are aligned at their tops)
That's it
For method2 you do almost the same, except you do not balance the RGB bars at their tops but at their center (connect the linear-burn blend-mode node to the scope) AND you adjust only the green and blue channels (for film-negatives with orange colored masks)
Method2 is not setup to include a whitebalance-node, since the goal is to introduce a slight golden tint at the highlights, but if you wish you can probably do so.
Method2 will generally result in stronger, more saturated, colors, although they will obviously be non-neutral.
Choose whichever method you prefer with the film-negative you are converting.
A technical note; All films have different response-curves* that aren't completely linear, so the simple gamma adjustment performed in this flow will not be a perfect gamma-conversion.
*
Even if you know the film's published response-curve, this curve can change over time depending on how the film has been stored, how old it is, which batch it is from, etc.
However, to do a 'perfect' gamma-conversion you can setup 3 curve-nodes, one for each color-channel, instead of the first levels-node and do this specific gamma-conversion that way.
PhotoReactor makes it possible
), but here they are: