Digital Light & Color

I just started getting interested in HDR and other stacking techniques, hence the question:

When one wants to stack carefully made, but hand-held images, theoretically it is possible to use multipoint alignment in Composite (using one image as a master and then aligning/registering all others to it). Did people tried it already, and if so, what's the practical advice? i.e. how many control points? placed where? what interpolation method to use, with what radius?

Cheers!

p.s. Can I be explained how the multi-point alignment actually works: the 3? first points are used for global shifting, rotating, scaling and sheering?, and additional points improve local interpolations? What happens with perspective adjustment (it needs 4 points, I think)?

Best to use a tripod, but if you must hand hold, 2-point (shift/rotate) alignment should suffice in most cases. If there is significant lens distortion however, try to remove this first. In any case, very precise positioning of control points is required.

Maciej,

I'm always looking for the easiest way and have used two alternative to PWP for image alignment.

I've done a fair bit of focus stacking using the rather wonderful, and free, CombineZP. The first step in the focus stacking process is image alignment and it is possible to export aligned images from within CombineZP for processing within PWP. However, CombineZP is 8 bit only.

The other method I've used, which does handle 16bit, is to use the free SNS-HDR Lite to align and render the images saving as a 16 bit per channel TIFF. You can choose and LDR output to limit what SNS-HDR Lite does beyond alignment.
Den has described how to use SNS-HDR 'from within' PWP.

Mike.

Thanks a bunch!

Would anyone know what the '-panorama' switch in SNS-HDR Lite does? Also, if I create shortcut to the SNS-HDR.exe, how do I specify the command line options/parameters?

I have not explored the " -panorama" preset but to specify command line parameters:

(1) right click on the the shortcut icon and select "Properties"

(2) change the 'Target' box of the 'Shortcut' tab to included the desired parameters, for example, my main desk top application icon Target Property for SNS-HDR Lite is:

"C:\Program Files\SNS-HDR Lite\SNS-HDR.exe" -srgb -x1 -default -tiff16

(3) click 'Apply'; then 'Close'

Incidently... the web pages have been re-organized to include a forum and the changelog link is now:
http://www.sns-hdr.pl/changelog.htm

Thanks Den! I think I know what -panorama does: it makes sure that both ends of the image that will be joined in a 360deg panorama look the same to minimize stitch visibility.

In your settings, aren't -x1 and -default switches used implicitly by default? i.e. would omitting them make any difference?

One more question: the -sRGB switch: what exactly does it do if RAW files are the input - how different is the output image if it's omitted?

In your settings, aren't -x1 and -default switches used implicitly by default? i.e. would omitting them make any difference?

Yes and No... I put them in to help me remember what the default options are and that I will get: sRGB colors, full size, 'default' ambience, and 16bit tiffs. I have additional desktop shortcuts for other ambience presets such as 'low dynamic range' and 'dramatic':

"C:\Program Files\SNS-HDR Lite\SNS-HDR.exe" -srgb -x1 -ldr -tiff16

"C:\Program Files\SNS-HDR Lite\SNS-HDR.exe" -srgb -x1 -dramatic -tiff16

Also I have encountered errors when stacking differing image versions created from the same base image where the reality is 1:1 registration... on those rare occasions, the shortcut Target Properties include " -da" and " -dd" to disable alignment and de-ghosting.

One more question: the -sRGB switch: what exactly does it do if RAW files are the input - how different is the output image if it's omitted?

I have assumed that the " -sRGB" switch is the same as the " -o 1" color option for 'dcraw9.07' that SNS-HDR Lite v1.3.12 uses when converting RAW files... but this assumption has not been varified nor its effect if omitted.

I do not use SNS-HDR Lite as a RAW converter or 'stacker' of RAW files. I will convert RAW files with PWP or Canon's DPP or DCoffin's 'dcraw' to 16-bit sRGB tiffs and then SNS-HDR Lite stack and/or tone map them... the resulting processed tiff images will have better color and less texture [luminance noise], especially in a resulting SNS-HDR Lite processed image's aggressively tone-mapped low contrast image areas such as sky/clouds... ...even at that SNS-HDR Lite processed images may need more saturation and/or noise reduction...

The refine feature in the composite transform could be more helpful in aligning if its search radius could be made larger. I find it hard sometimes to get close enough for refine to work. In the same vein, it would be very nice if one image could be nudged one pixel at a time while watching the outcome in absolute difference mode.