Home > Photography > Postprocessing
Once your exposures are aligned (probably using the method described in Aligning Multiple Exposures in Hugin), the hard part is over.
Import and Crop
- Open your image-editing software, and use it to open the TIFFs that were output by Hugin.
- Choose one of the TIFFs to be the background layer. I use the "RGB" (human-visible light) file, but it doesn't really matter.
- Copy and paste the other exposures as new layers into the image you selected in the previous step. I make the "NIR" file the second layer and the "UVA" file the third, but again it's just a matter of personal preference.
Newer versions of Photoshop® combined with newer versions of Hugin will cause you grief when performing this step if you attempt to simply copy and paste the additional images (unaltered) into new layers on top of the first image. This is because Hugin creates a transparent (zero-value alpha channel) border area, and Photoshop® snaps the selection down to the largest non-transparent area. The result is that the additional images will be centered on top of the base layer, which is not the result you want.
To avoid this, before copying and pasting the additional images, use the magic wand tool to select the border area around them, and then fill it with a bright (easily-recognizable) colour. I usually use 100% red. This will ensure that when you perform a select-all/copy/paste, the images are aligned correctly, and as a side benefit, make it easier to crop the images accurately in the next step.
- Find the layer that has the smallest coverage (the widest black borders) and crop the image down to remove those borders.
- Repeat the previous step for the other layer(s) if necessary. Make sure none of the layers have any border left.
- Save the file in the native format of the image-editing software.
- If there are any hotspots, create duplicates of the layers that contain them and perform corrections (for example, using the methods described in Reducing and Eliminating Hotspots). Be sure to do this before creating a bunch of false colour variations to avoid having to update them all.
Multispectral Colour Balancing
Before proceeding with the false colour variations, my next step is to balance the relative brightness/contrast/levels/etc. of the exposures. There are several ways to do this, but this is the one I use. Note that the situation becomes much more complicated if there are more than two additional spectral bands beyond the ones we can see.
- Copy and paste the RGB/human-visible image into a new file.
- Apply adjustment layers to get the contrast and brightness the way I want.
- Save the file (in my case, as "RGB.TIF" for use in The Mirror's Surface Breaks).
- Collapse all of the layers into one.
- Desaturate the image (convert it to greyscale).
- Paste the near infrared version in as the red channel.
- Paste the ultraviolet-A version in as the blue channel.
- Use a levels adjustment layer to tweak the red and blue channels around until the colours look right (or, if I happen to have a white-balance grey card shot of all three bands, until the card does appear grey in this composite version). Important: do not change the values for the green channel at this point.
- Make a note of the settings in the levels adjustment layer.
- Discard changes and close the file.
- Copy and paste the near infrared exposure into a new file.
- Apply a levels adjustment layer with the settings I noted for the red channel in the temporary composite.
- Save the file (in my case, as "NIR.TIF", for use in TMSB).
- If I don't think the image looks "right" for greyscale/tinted purposes (as opposed to false colour combinations), create an alternately-adjusted version, and save that separately (in my case, as "NIR2.TIF", again, for use in TMSB).
- Repeat the previous four steps for the ultraviolet-A version of the image.
At this point, you can either proceed with manually combining the greyscale channels (described below), or use The Mirror's Surface Breaks to do all of that and more.
Traditional False Colour
You'll need to repeat this section for each variation you want to produce (NIR-R-G, G-B-UVA, etc.). There are some shortcuts that can be taken, but there is currently a lot of manual work involved.
- Copy and paste the "RGB" (human-visible light) layer from the combined file into a new file.
- Shift the existing channels as necessary. For example, if you are creating the NIR-R-G variation, first copy the green channel into the blue channel, then copy the red channel into the green channel. If you are creating the G-B-UVA variation, first copy the green channel into the red channel, then copy the blue channel into the green channel, etc.
- Switch back to the window with the combined file. Copy the layer for the near infrared or ultraviolet-A (depending on which variation you are making).
- Switch to the new image, and paste the layer you copied in the previous step into the corresponding channel. For example, if you are creating the NIR-R-G variation, paste the NIR layer into the red channel. If you are creating the G-B-UVA variation, paste the UVA layer into the blue channel.
For variations that involve combining spectral bands (e.g. NIR-RGB-UVA), there are two possible methods. If the bands are already set up as individual channels of a single layer (e.g. RGB), use the process above, but instead of shifting the channels, desaturate the layer after pasting it into the new file.
If you want to create a secondary combination (e.g. NIRR-G-B, NIR-RG-BUVA), you'll need to create an intermediate file:
- Copy the channel or layer that corresponds to the first component of the combined output. For example, if you want an "NIRR" layer, copy the contents of the NIR layer from the combined file.
- Create a new file and paste in the copied layer or channel as a new layer. It should appear as greyscale.
- Switch back to the combined file and copy the second component. In the case of the "NIRR" example, this would be the red channel from the RGB layer.
- Switch to the new file and paste in the copied layer or channel as a new layer.
- If you are combining more than two spectral bands, repeat the previous two steps until you have as many layers as there are spectral bands to combine.
- For each layer in the new file except the background, set the opacity/transparency/alpha to something other than 100%. For example, to create an equal mix of near infrared and red, leave the background (NIR) layer alone but set the opacity of the red layer to 50%. If you are creating a "NIRRG" combination and want it to be mostly NIR but still include some of the red and green luminance, you could set the red and green layers to 25% opacity each, or set red to 25% and green to 10% to favour red over green, etc.
- Flatten the new image (or combine the layers).
- Use the result as you would any of the primary spectral bands in the last step of the previous section. In the "NIRR" example, you would copy the contents of the flattened composite created this section into the red channel of the false colour image.
- Copy the greyscale version of a spectral band (or combined spectral bands) into a new file.
- Copy one of the coloured variations of the image (the "RGB" layer, the tinted version of one of the other individual spectral bands, a false colour variation produced using the above technique, etc.) into the new file as a new layer.
- Set the compositing mode of the second layer to Color.
At this point, the result can be treated as if it were any other photo - adjust the levels, curves, hue/saturation, etc. I always do this in adjustment/effects layers so I can go back and change my mind later with minimal effort.