As much a note to self as a blog post published for others’ benefit, but this is how I ingest photos, tag them, and cull them.

Output

• unedited photos, converted to DNG for size savings and consistency across camera brands
• photos tagged with people and basic objects by a transformer model

Goal here is to save time with tagging later. That said, I still manually tag for “theme”, since Digikam doens’t (yet?) offer tooling to tune a model to identify your own tagging priorities. Still super useful.

Process

1. Convert images to DNG with Adobe DNG Converter (free as in beer)
2. Cull images in DxO Photolab
3. Use digikam to tag faces and basic objects (libre and also free as in beer)
4. Edit, rate, and add more refined tags in Photolab

Converting to DNG

DNGs are both super well supported, and reduce file sizes of the RAW images. This is done with the highest fidelity (so no lossy compression).

For a new camera, I’ll also embed the original RAW image just in case, but given a couple months with a camera and no issues in editing, I’ll stop doing this. It has the nasty side effect of increasing the resulting image size not by 2, but like, 2.5, since the RAW is much larger than the DNG it will produce without the reembed.

Culling in Photolab

I do this in Photolab instead of digikam mostly because I sometimes need to futz with exposure or crops to see if the image is worth keeping, which means I’m effectively starting the developing process. Rather start in Photolab where I’ll also finish, than hop back and forth between editors and waste work.

Tagging

digikam offers fantastic facial recognition, and decent object recognition for tagging.

Saves me a ton of time when dealing with personal photos. I’d bet this would also be hella useful for event photographers.

Editing etc

Back to Photolab, where I also often use the DxO Film Pack.

Also, do start ratings, thematic/project labels, etc.