Photoshop post processing

Question

I'm hitting a limitation with the way Lightroom handles TIFF files.
I have nearly two thosands scanned slides saved as TIFF. In addition to the normal R, G and B channels, each TIFF file includes an additional "infrared" alpha channel. This channel is created by the scanner using an infrared light to isolate dust and scratches on the film surface. I can then use a Photoshop action to read this infrared channel and eliminate most of the dust and scratches from the TIFF image. This is a step I do when exporting the TIFF as a JPEG for instance. I do not want to overwrite the original TIFF file. It is like my "raw" scan and I want to save it in case I come across a better de-dusting technique in the future.
Now the problem. Ligthroom is my tool of choice for cataloging and editing photos. It can read TIFF files but any edit applied to one during the export stage discarts the infrared alpha channel. Only the R,G,B channels are kept. This means that I can't use a Photoshop droplet or action to remove the dust as a post process step. I currently do not see a solution short of either dropping Lightroom, or keeping duplicate images (the original raw and the de-dusted version).  Keeping duplicates is not an appealing option. It would take hundreds of GB of extra space to start with.
Has anyone come across a similar scenario? How can one handle Photoshop post processing that relies on alpha channel (or layers).
Thanks

Answer 1

Hi Jaap,
Here are the steps that I use to implement my own ICE dust removal. I'm on Photoshop CS3 mac by the way.
1) Copy the alpha channel (infrared) into a new channel named "Dust".
2) Adjust the level on the Dust channel to isolate only the dust. You should get something like below. Note the markers. I would overlap them but I've spread them very slightly here to show where they are. I always err on the side of removing all image details from this channel even if that means that some of the dust/scratches won't make it in. Of course, you can also edit the channel and manually add (painting in Quick Mask mode) any missing dust or remove any image detail left.
3) Select the RGB channel
4) Load the Dust channel as a selection (Select -> Load Selection, Channel: Dust, Invert: checked)
5) Expand the selection by 1 pixel (Select -> Modify -> Expand)
6) Apply the Dust and Scratches filter (Filter -> Noise -> Dust & Scratches, Radius: 6, Threshold: 0)
The Radius to use depends on the resolution of your scan. Play with it and try to find the smallest radius that will remove most of the dust. If you set the radius too large, the color used to fill in will average surrounding pixels farther away, which isn't as accurate for fine details. A resolution of 6 seems to give me good results with a 4000 ppi scanner like the Nikon CoolScan LS 5000.
7) Work's done. Clean up by deselecting (Select -> Deselect, or simply Command D on a mac)
I've created an action for step 1 and another for steps 3 to 7. Step 2 is fully manual unfortunately. I don't think there is a way to set the markers programmatically like in the image above since the histogram varies for each image.
I also keep my alpha channel intact in case I find some better way to create the dust channel in the future. I'm a firm believer in the non destructive workflow. I won't scan these slides again. It may not even be feasible soon. Film scanners are being retired from the market much faster than I would have thought just a year ago.
Hope this helps