I use RT for most of my raw processing these days but struggle with the metadata handling. Any tips are welcome. Below i outline my workflow and what I want to achieve.
I photograph mainly buildings/spaces and family. The former is work related and occasionally published in books/magazines and goes on the web. Some are also printed for personal enjoyment. The latter get displayed on a private website for family members who live far away. Again some are printed.
Shoot DNG or DNG+jpg
Cull the worst in camera
Mount sd card and cull again using Geeqie
Transfer the images using Rapid Photo Downloader. Folderstructure is $HOME/img/yyyy/mmdd_jobcode/
Tag/title/caption images using Digicam or Geeqie. Data is saved as .xmp files next to the raw file
Open directory in RT and auto apply (Dynamic profile) a pp3 which provides a conservative tone curve, colour management settings, lens profiling and some disabled further settings (color propagation, 4:3 crop default etc)
Customize settings further mostly straighten, exposure curve tweaks and perhaps crop.
Star the images that might be good enough
Apply websave.pp3 to starred images which sets resize and post resize sharpening. Web is more urgent since grandmothers and architecture students need imgs quick!
Send starred images to queue with a %p1/web/%f path. This creates a web sub folder and outputs images there
Tag all images in the web folder again as the website generator creates tags and captions based on metadata. (ikiwiki with tweaks to various plugins including img, album, page templates etc.) The metadata contains faces, places, architects names, dates of construction, places and captions titles.
Create gallery page with basic info about the building or just a monthly page for the private stuff.
Return to RT at a later date and disable the web resize using the batch edit tab
Send images to queue again using a converted subfolder instead of the web folder above. This must be done manually before clicking queue. A full resolution version is now available
Tag all images in the converted folder again
Print/email/upload these files
Most of the time I try to get stuff onto the web asap. Partially because people are waiting for the photos and partially because I will have a backlog unless I do.
So the two issues I have are the bolded and italicised above. Having to retag after every export and managing the resizing. Using exiv2 -l ../ -S .DNG.xmp in *.jpg from the web/converted folder is doable but it screws with the rotation so I usually just Geeqie tag everything again The reason for having to retag is that RT doesn’t read xmp sidecar files and ignores the metadata.
The resizing is just conceptually and habitually difficult as I mentally see it as part of the export stage and not part of the image processing proper. Darktable get this right.
Is anyone working like this? Am I missing a trick or can a different mental model solve my issues. I know that using darktable solves my issues but pixelshift and highlight recovery have turned out to be pretty important. RT is also more responsive when working with tone curves and while zooming.
Edit for more info: My websites have about 5000 of my photos taken over 15 years or so.
One solution would be to tag your final images using ExifTool using digiKam’s original xmp sidecar files as source. IIRC ExifTool has an “exclude” option which you could use to prevent rotation being copied over. If not, you could sed-delete the rotation line from the XMP file.
It may be not the best solution but it could save you one tagging operation. Why not create the full resolution version containing the tag on the first place and then resized them using tool like convert from ImageMagick to create the web version ?
Yes that’s probably the best way. The downside is that I have to re tag whenever I produce a new output file due to actual processing changes rather than just resize. This is however more rare, although not as rare as I would like… it does happen that the initial processing needs tweaking. It also feels wrong not to have the metadata associated to the raw, but I guess that’s just a habit.
So, maybe the full-resolution one needs to be further back in the processing chain, so it is where you start any new processing. Now I’m not a RawTherapee expert, but from what I do understand, if you used that image as the baseline for any processing, you could use one of the previously created output images’ .pp3 to reapply processing to it, then change resize (or any of the other processing) and save as a new output/pp3 combination. ??
My proposition is a derivative of my own Workflow. I produce a Master file after all processing is done in RT and, if necessary Gimp at full resolution. This Master file contain as many pixel as it can get after cropping in RT. I then use it to produce any other file version for web, print,… by downsizing it.
I use ImageMagic because I also take advantage of the operation to add white margin for printing.
The downside is that I have to re tag whenever I produce a new output file
I have solved the tagging problem for my scans of slides and negatives by creating args-files for EXIFtool-GUI, which copy the relevant tags from the tagged images to their derivatives.
I have scanned all my images into raw files (TIF), which are 260MByte each. From these I have created in a batch job a down-sized JPG with standard image processing. These small images are for two purposes: From them I can judge (especially for negatives) which images are worth detailed processing and second, I tag them, because that is much faster than tagging the big raw data files. Once I am done, I use the args-files to copy selected tags to the raw files and the processed images in a batch job. So the hard work of tagging has to be done only once.
For the scans, tagging the raw image files is not dangerous, if one takes care not to overwrite relevant tags. But I do not know for raw images from digital cameras.
Thanks for the tips everyone. I’ve tried them out (except exiftool-gui which is windows only). I think I’m going to try using darktable again and only use RT for pixel shift. I must from now on always underexpose even more than I already do
I’m reluctant as I find adjusting exposure/tone curve clunky and inconvenient in darktable (80% of my manual processing…) but I’ve also been reminded how much better the whitebalance works for my Pentax K-3 II in darktable.
Both software are really great but I find there’s quite a workflow cost to switching between them and I’d like to avoid this. It seems that I’m the promiscuous type tough and I struggle to stay software faithful.
The reason for having to retag is that RT doesn’t read xmp sidecar files and ignores the metadata.