alanorth/cgspace-notes
1
0
Fork 0
mirror of https://github.com/alanorth/cgspace-notes.git synced 2025-01-27 05:49:12 +01:00
Code Issues Projects Releases Wiki Activity

Add notes for 2023-03-30

Browse Source
This commit is contained in:
Alan Orth
2023-03-30 16:59:20 +03:00
parent 479bb9684a
commit a2875a3811
31 changed files with 172 additions and 36 deletions
Download Patch File Download Diff File Expand all files Collapse all files

66
content/posts/2023-03.md
View File

@ -582,4 +582,70 @@ $ for q in 70 80 90; do ssimulacra2 img/10568-103447.pdf.png /tmp/10568-103447-l
- Using `-define webp:method=6` (versus default 4) gets a ~0.5% increase on ssimulacra2 score
## 2023-03-29
- Looking at the `-define webp:near-lossless=$q` option in ImageMagick and I don't think it's working:
```console
$ for q in 20 40 60 80 90; do magick convert -flatten data/10568-103447.pdf\[0\] -define webp:near-lossless=$q -verbose /tmp/10568-103447-near-lossless-q${q}.webp; done
data/10568-103447.pdf[0]=>/tmp/10568-103447-near-lossless-q20.webp PDF 595x842 595x842+0+0 16-bit sRGB 80440B 0.080u 0:00.043
data/10568-103447.pdf[0]=>/tmp/10568-103447-near-lossless-q40.webp PDF 595x842 595x842+0+0 16-bit sRGB 80440B 0.080u 0:00.043
data/10568-103447.pdf[0]=>/tmp/10568-103447-near-lossless-q60.webp PDF 595x842 595x842+0+0 16-bit sRGB 80440B 0.090u 0:00.043
data/10568-103447.pdf[0]=>/tmp/10568-103447-near-lossless-q80.webp PDF 595x842 595x842+0+0 16-bit sRGB 80440B 0.090u 0:00.043
data/10568-103447.pdf[0]=>/tmp/10568-103447-near-lossless-q90.webp PDF 595x842 595x842+0+0 16-bit sRGB 80440B 0.080u 0:00.043
```
- The file sizes are all the same...
- If I try with `-quality $q` it works:
```console
$ for q in 20 40 60 80 90; do magick convert -flatten data/10568-103447.pdf\[0\] -quality $q -verbose /tmp/10568-103447-q${q}.webp; done
data/10568-103447.pdf[0]=>/tmp/10568-103447-q20.webp PDF 595x842 595x842+0+0 16-bit sRGB 52602B 0.080u 0:00.045
data/10568-103447.pdf[0]=>/tmp/10568-103447-q40.webp PDF 595x842 595x842+0+0 16-bit sRGB 64604B 0.090u 0:00.045
data/10568-103447.pdf[0]=>/tmp/10568-103447-q60.webp PDF 595x842 595x842+0+0 16-bit sRGB 73584B 0.080u 0:00.045
data/10568-103447.pdf[0]=>/tmp/10568-103447-q80.webp PDF 595x842 595x842+0+0 16-bit sRGB 88652B 0.090u 0:00.045
data/10568-103447.pdf[0]=>/tmp/10568-103447-q90.webp PDF 595x842 595x842+0+0 16-bit sRGB 113186B 0.100u 0:00.049
```
- I don't see any issues mentioning this in the ImageMagick GitHub issues, so I guess I have to file a bug
- I first [asked a question on their discussion board](https://github.com/ImageMagick/ImageMagick/discussions/6204) because I see that the near-lossless option should have been added to ImageMagick sometime after 2020 according to another discussion
- Meeting with Maria about the Alliance metadata on CGSpace
- As the Alliance is not a legal entity they want to reflect that somehow in CGSpace
- We discussed updating all metadata, but so many documents issued in the last few years have the Alliance indicated inside them and as affiliations in journal article acknowledgements, etc, we decided it is not the best option
- Instead, we propose to:
- Remove `Alliance of Bioversity International and CIAT` from the controlled vocabulary for affiliations ASAP
- Add `Bioversity International and the International Center for Tropical Agriculture` to the controlled vocabulary for affiliations ASAP
- Add a prominent note to the item page for every item in the Alliance community via a custom XMLUI theme (Maria and the Alliance publishing team to send the text)
## 2023-03-30
- The ImageMagick developers confirmed [my bug report](https://github.com/ImageMagick/ImageMagick/discussions/6204) and created a patch on master
- I'm not entirely sure how it works, but the developer seemed to imply we can use lossless mode plus a quality?
```console
$ magick convert -flatten data/10568-103447.pdf\[0\] -define webp:lossless=true -quality 90 /tmp/10568-103447.pdf.webp
```
- Now I see a difference between near-lossless and normal quality mode:
```console
$ for q in 20 40 60 80 90; do magick convert -flatten data/10568-103447.pdf\[0\] -define webp:lossless=true -quality $q /tmp/10568-103447-near-lossless-q${q}.webp; done
$ ls -l /tmp/10568-103447-near-lossless-q*
-rw-r--r-- 1 aorth aorth 108186 Mar 30 11:36 /tmp/10568-103447-near-lossless-q20.webp
-rw-r--r-- 1 aorth aorth 97170 Mar 30 11:36 /tmp/10568-103447-near-lossless-q40.webp
-rw-r--r-- 1 aorth aorth 97382 Mar 30 11:36 /tmp/10568-103447-near-lossless-q60.webp
-rw-r--r-- 1 aorth aorth 106090 Mar 30 11:36 /tmp/10568-103447-near-lossless-q80.webp
-rw-r--r-- 1 aorth aorth 105926 Mar 30 11:36 /tmp/10568-103447-near-lossless-q90.webp
$ for q in 20 40 60 80 90; do magick convert -flatten data/10568-103447.pdf\[0\] -quality $q /tmp/10568-103447-q${q}.webp; done
$ ls -l /tmp/10568-103447-q*
-rw-r--r-- 1 aorth aorth 52602 Mar 30 11:37 /tmp/10568-103447-q20.webp
-rw-r--r-- 1 aorth aorth 64604 Mar 30 11:37 /tmp/10568-103447-q40.webp
-rw-r--r-- 1 aorth aorth 73584 Mar 30 11:37 /tmp/10568-103447-q60.webp
-rw-r--r-- 1 aorth aorth 88652 Mar 30 11:37 /tmp/10568-103447-q80.webp
-rw-r--r-- 1 aorth aorth 113186 Mar 30 11:37 /tmp/10568-103447-q90.webp
```
- But after reading the source code in `coders/webp.c` I am not sure I understand, so I asked for clarification in the discussion
- Both Bosede and Abenet said mapping on CGSpace is taking a long time and I don't see any stuck locks so I decided to quickly restart postgresql
<!-- vim: set sw=2 ts=2: -->