HDR Wavelet Tool

[sreilly]

In Harry's excellent video on using the HDR Transformation Tool, the example data is from a one shoot color camera. I'm working on my M95 image that is taken with a ST10XME mono camera using a color filter wheel and at this point the image is only an RGB image and the luminance data has yet to be applied. My gut tells me that I sould not be using HDR until the luminance data is combined.
So far the master red, green and blue images have been made and combined as a RGB image using MaxIm and saved as a IEEE floating point image. I have a fits plug-in for PS that can read color 32 bit fits files but usually convert to 16 bit as PS has functions such as curves that only work at 16 bit or lower bit depths. My usual routine is to process the RGB data seperately in Photoshop CS4 using curves, levels and sometimes saturation, and then process the luminance data with levels, curves, selective high pass sharpening and then combine the two layers into one making final adjustments to the combined image.
So in PI, I've got the raw 32 bit RGB image. Should I be combining the red, green, and blue 32 bit images in PI? Should the luminance be added at this point? My thinking is that tools shuch as HDR are probably better applied to the stronger luminance data rather than the weaker color to suppress image noise? I've been taught over the past 12 years that the color data is for colorization of the image and the details are all in the luminance thus detail processing is done on the luminance layer only. In fact, I've been using a Gaussian blur to dampen the color noise prior to adding the luminance layer. Now  it seems that is about to change or not? Should I be combining the LRGB in the first steps and if so, does PI aling when the color is bin 2x2 and the luminance is bin 1x1 or do I need to resize the RGB data first?
This will be a good start but more questions will be following based on these answers.

Thanks,

[Nocturnal]

Hi Steve,

I recommend you combine your image in PI using the LRGBCombination process. This ensures that all 32 bit detail from your stacks is preserved. You can then apply HDRWT at whichever point you desire. I typically do this after stretching and noise reduction but before saturation curves. Note that the HDRWT dialog gives you the option to work on luminance only. I typically do that but I recommend you experiment a little to see what works best for you.

A good trick is to draw a preview around your target area and apply HDRWT to that. When you find a combination of parameters you like but want to experiment more, create a copy of your preview and now work with that one. Repeat until you're satisfied you're done playing. Now switch back and forth between previews to find the one you like best. Load the history explorer for that preview (right click menu). Drag and drop the HDRWT icon to your desktop or directly to the main tab of your image. It'll now get applied to the entire image. This is one of those PI features I'm confident you won't find anywhere else and really facilitates exploratory processing without requiring you to take notes to remember process settings.

Other people may provide more insight in the mathematical reasons for doing HDRWT earlier or later. That's a bit beyond me, frankly

[sreilly]

Sander,

I just tried this but not sure what I'm doing wrong. I've opened the master red, green, blue and luminance images and as these are simply combined 32 bit stacked images, only the very brightest of stars show, all 3-4 of them. Do I stretch each individual image before registration or what? They are also 1/2 the size of the luminance image. What, if anything needs to be done for registration?

Thanks,

[Nocturnal]

Steve,

if your images aren't registered you can do this with the StarAlignment process. Pick one of your images (use L) as your reference and drag the process icon (bottom left corner triangle) onto the other 3. This will create new images for each of them, registered to the reference. Then combine. The combine image won't have a ScreenTransferFunction yet so the linear image will be dark. With the new image active, click the 'A' icon on the STF window for an automatic STF.

Let me know how it works for you