Am Montag, den 21.07.2008, 23:11 +0200 schrieb DigiSus: > Hello > Hey!
> Thanks Felix for implementing variant B. :-) > You're welcome! :-) > As to calculation of the 35mm equivalent; it is not so bad/difficult. I > think the problem is rather to get the needed data (and it is not 14 > parameters). > Pretty much yes. The 14 (actually 15) parameters I was talking about are the ones that could be used by the implementation in ExifTool (in the worst case). But I've taken a closer look at them and 5 of them are specific to Canon cameras and one is available on Olympus cameras only (MakerNote). I think my comment on that matter sounded harsher than I intended. > Let me give a few points so you can think about what could make sense to > implement. > > 1) The "standard" format for images on a film role from the ancient > analog world is 24x36mm, having a diagonal of d_full=43.27mm. > Ah, yes that explains the 36 in the different formulas. For some reason I applied the 35mm here as well all the time. :-) > 2) Digital cameras have very differently sized CCD sensor chips, all of > them usually smaller than the full format. > > 3) To calculate the crop factor/focal length multiplierFLM (different > names, same thing), one needs the diagonal d_ccd of the actual CCD > sensor chip. > > 4) Then, it is quite easy: FLM = d_full / d_ccd > > 5) With that the measured focal length (EXIF:FocalLength) times the MLF > gives the equivalent focal length in 35mm film camera. > This is pretty much the technique behind ExifTool. The interesting part is determining d_ccd. > Problems/Solutions: > I did not find the CCD size in the EXIF spec. Am I wrong? > Right. Exif currently has no tags for the sensor size. Some Canon cameras seem to carry this information in their MakerNote and Olympus has a tag for d_ccd directly. Implementations normally try to calculate it by comparing the FocalPlaneX/YResolution tags and the image size. But the Exif specs also warn that the FocalPlaneX/YResolution tags aren't necessarily the pixel resolution of the image sensor. So, I'm not sure how accurate this really is. Felix _______________________________________________ Eog-list mailing list [email protected] http://mail.gnome.org/mailman/listinfo/eog-list
