Following Monday's incident, I have decided to urgently bring forward + accelerate my plans to start developing a replacement solution to my Picasa-based Photo Management workflow. As this incident had highlighted, I do somewhat urgently need to find a reliable automated solution to migrate my almost 2 decade library of non-destructive edits over to a future-proof solution I fully control!
As a result, I made a bunch of promising progress on my Picasa Reverse Engineering project last night - i.e. the first night of a new journey!
Progress on Night 1:
* 1) Located again the partial notes of other folk who have attempted this.
* Notable Links:
- https://gist.github.com/fbuchinger/1073823/9986cc61ae67afeca2f4a2f984d7b5d4a818d4f0
- https://sites.google.com/site/picasaresources/picasa/how-picasa-works
* Main takeaway is that I now have notes on how the cropping stuff is stored + decoded (I.e. their weird "rect64" hex-encoded relative offsets format), so now I just need to code up my own usable version
* Between that and being able to identify the starred items, we're already a major way towards having enough of the really basic + important repeatable info extracted
* 2) Spent time probing every tool + slider I usually use for my edits (which is really not that many of them... hence why this is even a feasible project to consider, beyond the archival importance 😜)
As a result, I should have enough info to attempt a parser of the .ini settings files tonight, which is a major first step in getting anywhere with this!
Of course, these are the easy parts:
The hard parts are to actually map those parameters into pixel operations that can be reimplemented in other tools...
(The most critical of all these is the algorithm for the "Fill Light" tool, which I haven't been able to replicate by simply munging sliders + curves in most standard image editors to date... who knows... I may have to turn to "AI" techniques to crack this one)
~~~
In the context of the wider project, the other todo's at this stage that I had been tinkering with over the weekend already:
(These had been as more of a speculative prep stage, vs a "this is happening now" type of project execution stage) include the following (Of course, all these are for after the critical task of being able to liberate the image edit history for the library)
* 1) Try to figure out how the database format works, as there's some magic there that let's them quickly load the thumbnails better than any other tool out there
* 2) Figure out how to implement a similar image gallery widget, with similar performance + nice layout.
OR at the very least, figuring out a new pixel-perfect design for how I'd like to work with these... as the existing alternatives don't hit the right balance, even after tweaking IMO (which is again a big reason why we're even having this debate)
(NOTE: On a high level, I do have my wishlist of next-step tools I would like to have that I've posted about before - e.g. the sequence grouping/tagging + batch-edit tools... - so supporting those is also a major part of the design considerations)
No comments:
Post a Comment