#38793 closed enhancement (fixed)
Twenty Seventeen: Compress the header image (mostly visibly lossless)
Reported by: | Presskopp | Owned by: | joemcgill |
---|---|---|---|
Milestone: | 4.7 | Priority: | low |
Severity: | normal | Version: | 4.7 |
Component: | Bundled Theme | Keywords: | 2nd-opinion has-patch |
Focuses: | Cc: |
Description
To save bandwidth or to be SEO friendly, or just because it needs less space :),
I suggest to exchange the header image with a compressed version.
tinjpg.com can do so and here's the result
compressed image is 143KB (vs 209KB)
@joemcgill likes the idea, he said:
https://wordpress.slack.com/archives/core-themes/p1479168211000348
Ran DSSIM on the two images and they’re 99.8% visually similar, which is nearly indistinguishable
Attachments (5)
Change History (19)
This ticket was mentioned in Slack in #core-themes by joemcgill. View the logs.
3 years ago
#3
@
3 years ago
- Milestone changed from Awaiting Review to 4.7
- Owner set to joemcgill
- Status changed from new to reviewing
#4
@
3 years ago
This seems fine to me. For what it's worth – Image Optim was used to compress the image before its commit: https://imageoptim.com/
#6
@
3 years ago
@Presskopp
The header.jpg can be image optimized even further down to 130kb or under 50kb.
#7
@
3 years ago
Is it a competition now :) ?
I see artifacts in your examples, of course we can discuss if we need to keep all the pixels. I made another attempt, which for me is optically still ok (112kb).
#9
@
3 years ago
@Presskopp
Both the 97kb and 112kb versions look good. I see nothing wrong with making sure content in a default theme is optimized.
#10
@
3 years ago
- Priority changed from normal to low
- Resolution fixed deleted
- Status changed from closed to reopened
Thanks @Presskopp and @lukecavanagh for working on this. While I think the version we used in [39248] is probably adequate, there's no harm in trying to optimize further as long as we're not degrading the quality significantly. I'll take another SSIM measurement of these against the original to make sure we're not over-compressing.
#11
follow-up:
↓ 12
@
3 years ago
- Keywords 2nd-opinion has-patch added
Some measurements here using DSSIM against the original before [39248] (lower scores are better):
Original (209KB)
header.jpg (143.7KB) – 0.002244
header(75).jpg (112.2KB) – 0.003032
header-tinypng.jpg (131.3KB) - 0.003752
header (75)_mini.jpg (97.0KB) - 0.004268
header.2.jpg (47.9KB) – 0.008374
Based on these numbers, I think header (75).jpg is the best option, but would like a second opinion from @davidakennedy, @karmatosed, or @melchoyce before compressing this image further.
#12
in reply to:
↑ 11
@
3 years ago
Replying to joemcgill:
Based on these numbers, I think header (75).jpg is the best option
Agreed — the other versions have visible artifacting.
#14
@
3 years ago
Side note:
Interesting, the difference:
header(75).jpg (112.2KB) – 0.003032
header (75)_mini.jpg (97.0KB) - 0.004268
because the 2nd was calculated out of the 1st using http://www.jpegmini.com/, seems it's more intrusive than I expected.
I'm sorry of course it's tinyjpg.com doing the job - thank you!