Upon conversion, NLP analyzes the photo and calculates black/mean/white points for R, G, B each (and additionally, smart white balance values). These are stored in the database, and – when using Roll analysis – also in the JSON file for that roll.
For some photos which contain predominant colors (think sunsets, etc.), the white and black values might become quite extreme, and the resulting color cast is not really correctable with the sliders.
Playing around and analyzing different crops of the photo does not seem very reproducable (since re-analyzing the photo with its final crop might destroy everything).
So here is an idea to mitigate this:
How about making the analysis result visible and editable, maybe in the Advanced tab?
-
Initial analysis result could be refined (or completely changed in creative ways) to better match the image in extreme situations
-
By also allowing copy & paste for these 9 values (black/mean/white for R/G/B) like for the other adjustments, this would allow for an easy “Match” feature (copy the analysis values from one photo to others) while still being able to further adjust the values on the destination photo if necessary.
This would enable reproducible results independent of the current crop of an image, and avoid “use a different crop and hope for the best” situations when dealing with difficult color casts.
Also, this way (with the actual analysis values being editable and reproducable alongside the photos), Roll analysis would be much more robust: upon initial roll analysis the values are stored in the photo metadata, without the need of maintaining roll data in JSON files, etc. (I don’t use Roll analysis at all, because I want to be able to replicate the result when scanning only a single photo, without still having access to the whole roll).
What do you think?