There is at least one more issue here. The JPEG thumbnails generated by the camera may not match the output of the RAW file as it emerges from the libraw pipeline. Many cameras will adjust the JPEG files it generates (whitepoint, contrast expansion, etc.) in their own way. Other cameras (esp. point-and-shoot RAW cameras) have barrel distortion that's corrected in software. (Shotwell can't correct this at this time.) If we relied on the JPEG thumbnail as the baseline image, all adjustments you made would be against that image. Then, when you exported the RAW photo, we would run those color adjustments against the RAW output and not the JPEG, and you'd get different results -- sometimes drastic differences.
In other words, you'd be editing a post-processed JPEG and applying those transformations to a pre-processed RAW file. The mimics, however, are the pre-processed RAW files stored in a format more conducive to interactive editing, which is why we generate them. -- Jim On Mon, Aug 16, 2010 at 8:07 AM, Adam Dingle <[email protected]> wrote: > On 08/15/2010 10:25 PM, stesind wrote: > > Hi, > > > > Asap as I start latest svn shotwell it starts to create JPEGs in the > > mimics folder. Each of them has a size of 2-3MByte. This is nearly the > > size of delevoped JPEGs. In my libary are a lot of CR2s. Every time > > shotwell eats a lot of my system performance. The mimics folder has a > > size of 15GByte. Why is it creating this files? I think it is totally > > useless since JPEGs are stored in the CR2 Raws. They should be used. > > > > stesind, > > First of all, note that it can be *very* slow to generate a display > image from a RAW photo - it can take several seconds, which is an > unacceptable amount of time for opening a photo. So Shotwell needs to > have a full-size JPEG available for each photo in the library. This is > why it generates the mimic JPEGs. > > Now, we agree that it would be nice if Shotwell could use a JPEG > embedded in a RAW file rather than generating its own JPEG mimic - this > is http://trac.yorba.org/ticket/1771 . But not all RAW files contain > embedded JPEGs which are large enough for all editing needs, and so even > after we implement this capability the user will be able to turn it on > and off. For example, CR2 photos from my Canon S90 camera contain > embedded JPEGs which are 1600x1200 (to see the sizes for your camera, > run 'exiv2 -p p foo.cr2' on one of your photos). My monitor is > 1920x1080, so it would probably be reasonable to use these embedded > JPEGs for displaying photos in Shotwell as long as I never need to zoom > into a photo. But, still, this should be a user option. If the user > enables the use of embedded JPEGs, then Shotwell will not generate the > mimic JPEGs, but if you try to zoom into a resolution larger than the > embedded JPEG then Shotwell will have to display a blurry photo (at > least for several seconds; perhaps it could render the RAW image in the > background and then update the display). > > adam > > _______________________________________________ > Shotwell mailing list > [email protected] > http://lists.yorba.org/cgi-bin/mailman/listinfo/shotwell > _______________________________________________ Shotwell mailing list [email protected] http://lists.yorba.org/cgi-bin/mailman/listinfo/shotwell
