More > JRiver Media Center 31 for Windows

Black bar detection not updating tag correctly

<< < (8/12) > >>

mattkhan:
realised I set limit incorrectly, have to rerun the analysis  ::)

Hendrik:
Setting limit to a number > 1 requires it to be bitdepth aware (<1 is a percentage). Which I suppose is what you figured out? :)

mattkhan:
Yes exactly, surprised it actually a) accepts such values and b) produces generally sane results but which still can differ just by varying those values (from 1 absurd value to another)

mattkhan:
corrected my use of limit & reran same analysis, now with 16px tolerance

461 are ok
129 are not

sensitivity counts

limit: 34
skip: 68
both: 18
neither: 45

go down to 8px and you get

317 are ok
273 are not

go up to 32px and it barely changes from 16px

need to spend a bit of time looking at specific differences to see if there is a recommendation to make

Hendrik:
Note that we use "round=4" as a parameter (used to be even round=2 before .25, but i bumped it up one notch to reduce some variance), while the default is round=16, which is likely why you are seeing many small-pixel differences, as it rounds the rectangle differently (and also why 16px difference is the sweet spot, as the difference should in many cases be at most 12px)

Navigation

[0] Message Index

[#] Next page

[*] Previous page