Stefan,
I will try to answer in English since I am not good in German, hope this is OK.
I have register same frames twice (actually twice for L channel, and than repeated whole procedure for RGB as well...), first time registered on stars, second time on comet. Then I have integrated both series with aggressive sigma clipping, as you have already guessed. Since during imaging, comet moved more than its size, first series gave me star field (almost) without comet, while second series give me just a comet, without stars.
Then I have combined L with RGB channel, both registered on stars to get colour image of the star field. And than same for L and RGB registered on comet to get colour comet image. And at the end, I have combined comet over star field, so the comet position is approximately on May 5th 1:00 CET ...
Regards,
Davdi