Video recordings offer great advantages for qualitative social science research. They do, however, come with tricky limitations. The paper picks up on the debate about prudent use of video recordings in sociological research, sounding out the specific potential of this instrument of inquiry, analysis and presentation (of findings). Looking at research on "professional vision" in science studies we try to outline the constructive nature of these practices, which manufacture visual traces into evidence. In order to avoid implicit realism we argue for taking seriously practice-trained vis-ability as a necessary anchor point for the use of video recordings (in qualitative social research). Such an approach allows the researcher to take advantage of technical conservation and equally to reflect its principles of construction. Only in this way is it possible to see through the camera and at the same time past it.