storing pictures

Willem Ferguson willemferguson at zoology.up.ac.za
Mon Oct 26 00:42:56 PDT 2015


On 25/10/2015 23:48, Anton Lundin wrote:
> One idea i got for not bloating the git repo, is actually chucking the 
> pictures in a separate git branch, completely separate from our 
> "dive-data" branch. We could even put thumbs in another branch. That 
> way we could choose if we would like to just download the dive-data, 
> thumbs, or even full pictures, just by keeping them in separate lines 
> of history. Example is that you only like your dive-data on your cell, 
> but dive-data + medium resoulution thumbs on your tablet, meanwhile 
> you got your dive-data and highres on your laptop. Another direction 
> to go is to use https://git-annex.branchable.com/ //Anton 

Once images are stored remotely, the thumbnails become much more 
important. What were your ideas about thumbnails? I have been looking at 
some of my images, and if these are stored remotely, then a "thumbnail" 
size of around 500 x 300 px gives sufficient detail to see what is 
important and to decide whether to download the remote image or not. I 
have been crunching the images from 20 mb raw to around 16 kb jpeg 
images (500 x 300 px) using a huge compression factor. But, the question 
is: is 16 kb per image small enough? Even with this size, a collection 
of several (tens of) thousand photographs would bloat the dive log 
significantly. What thumbnail size would be small enough to keep on-line 
permanently? Maybe a user-specified preference? Now, Subsurface works 
with three image sizes: thumbnail-on-profile, enlarged thumbnail and 
full-sized image. Maybe a useful approach would be to use the enlarged 
thumbnail and generate the thumbnails-on-profile on the fly, using the 
enlarged thumbnails? This would provide maximum independence on the 
remote collection. In this respect Anton's proposal above makes sense to 
keep the images and thumbnails separately.
Kind regards,
willem



More information about the subsurface mailing list