mirror of
https://github.com/flynx/ImageGrid.git
synced 2025-11-03 04:40:10 +00:00
42 lines
1.6 KiB
Bash
42 lines
1.6 KiB
Bash
|
#!/bin/bash
|
||
|
|
|
||
|
|
SCALE=0.21
|
||
|
|
FACTOR=4.76
|
||
|
|
ALGORITHM=bicubic
|
||
|
|
|
||
|
|
COMPRESSION=90
|
||
|
|
|
||
|
|
PATH=$PATH:`pwd`/vips-dev-7.32.0/bin/
|
||
|
|
|
||
|
|
for f in */DCIM/hi-res\ \(RAW\)/*jpg ; do
|
||
|
|
# this yields images with EXTREAM aliasing artifacts, with all supported algorithms
|
||
|
|
# NOTE: vips appears not to support lanczos scaling...
|
||
|
|
##vips im_affinei_all "$f" "${f/hi-res\ /preview }:${COMPRESSION}" $ALGORITHM $SCALE 0 0 $SCALE 0 0
|
||
|
|
|
||
|
|
# NOTE: -n (no sharpening) gives a bit too blurry results...
|
||
|
|
# NOTE: this also renames the files...
|
||
|
|
##vipsthumbnail -s 900 "$f"
|
||
|
|
|
||
|
|
# this gives lots of warnings but appears to be OK... but not too fast.
|
||
|
|
# so far, this is the best result...
|
||
|
|
# NOTE: this, being "not too fast" actually is the fastest when comparing to I_View or PIL...
|
||
|
|
# - I_View ~2-3 minutes
|
||
|
|
# - PIL ~6 minutes
|
||
|
|
# - im_shrink <2 minutes
|
||
|
|
vips im_shrink "$f" "${f/hi-res\ /preview }:${COMPRESSION}" $FACTOR $FACTOR
|
||
|
|
|
||
|
|
# this is different in that it uses a shrink factor while opening the image, thus
|
||
|
|
# a different scale factor...
|
||
|
|
# - ALLOT faster
|
||
|
|
# - blurry in comparison
|
||
|
|
##vips im_shrink "$f:4" "${f/hi-res\ /preview }:${COMPRESSION}" 1.19 1.19
|
||
|
|
|
||
|
|
# apply different amounts of read scaling...
|
||
|
|
# still quite extreme aliasing...
|
||
|
|
# IDEA: it's so extream, might be a good idea to first blur the image some and then scale...
|
||
|
|
##vips im_affinei_all "$f:2" "${f/hi-res\ /preview }:${COMPRESSION}" $ALGORITHM 0.42 0 0 0.42 0 0
|
||
|
|
# artifacts, better than all previous attempts at im_affinei_all, but worse than plain im_shrink...
|
||
|
|
##vips im_affinei_all "$f:4" "${f/hi-res\ /preview }:${COMPRESSION}" $ALGORITHM 0.84 0 0 0.84 0 0
|
||
|
|
done
|
||
|
|
|