I was impressed when I saw this story yesterday and really liked the idea behind Sprite Lamp, so I figured out the algorithm and wrote a program that replicates its basic functionality of generating normal maps for 2D sprites lit from four directions: http://i.imgur.com/H1H0R8k.png. The program will need some more work before it can be practically used but it's the same basic idea.
I intend to release it as a command line tool under a free license; I do not intend to compete with Sprite Lamp by building an artist-friendly GUI or implementing lots of fine-tuning. The idea has been used in games before and seemingly isn't patented but Snake Hill Games are, as far I can tell, the first to offer this functionality in a stand-alone tool. I would like to see them succeed but I also feel that having a FOSS implementation of this algorithm (mine or otherwise) would be a benefit to the community.
Should I wait until the end of the Kickstarter campaign (edit: or longer) to release it?
And as you say, it targets a completely different set of users. Sprite Lamp, is building something for artists to use, while the people who want a command line tool, would likely try to build it themselves rather than try to fit the .NET gui into their pipeline.
And additionally, it's likely that your release won't spread fast in the same channels at all, not really even showing up on the radar of many potential contributors of Sprite Lamp.
My point is, the ethics of the decision really only matter if you are going to have any affect on the kickstarter campaign. (Which I think even then you'd be in the proper right to release your code) But your unlikely to affect the kickstarter at all. So, IMO, I think you should release now, and ride the wave a little bit for people who like the idea of Sprite Lamp, but don't like the idea of the interface.
At least one HN user has shown interest in this tool, so as a compromise in case Snake Hill Games does ask me to wait until the end of the Kickstarter campaign I've set up a Google Group  for those who might want to test it before the public release. Depending on the circumstances I'll post a private download link or a GitHub link in the group once it's ready.
If Sprite Lamp can fit a similar niche to Texture Packer then it can be successful. For example, if it implements multiple normal-determination algorithms, has loads of parameters to tweak, accepts and outputs many different formats, and has a stable and great looking GUI.
I'd say release it. There is nothing wrong with competition, capitalism works because of it. Sprite Lamp will have to provide a better product if they want to earn money from it.
Better to be confronted with a FOSS version now than when they have invested a lot of time and money into their own product.