Quote:
Originally Posted by KevinH
And I really doubt anyone is going to be super zooming in on any of these images in Sigil's user guide.
|
To read the text in the labels?
To read the text in the menus?
To read the text in the UI?
Look at the "Book.xhtml" text in the "super zoomed in" version.
It's as if you've rubbed your eyes in vaseline.
Quote:
Originally Posted by KevinH
But can you and most people actually see a difference without "super zooming"? Seriously? I can not.
|
Yes.
And when you've converted hundreds of ebooks, and know where to focus (flat colors, edges of text), you can tell the fuzz/artifacts from miles away:
https://www.scantips.com/basics9jb.html
https://photo.stackexchange.com/ques...one-about-them
PNGs are lossless and excel at screenshots + charts/graphs/artificial images.
It's the reason why you want to use PNG >> JPG in this case!
Quote:
Originally Posted by KevinH
So we must agree to disagree here.
It is just that image file sizes on the web and in browsers trumps all of that for me as long as at normal display sizes no big differences are noticeable. Again, speed of loading, caching, reload, memory footprint, disk footprint, download size, bandwidth used all are greatly improved.
|
We don't live in the age of Kindle 1 anymore.
Devices can now:
- "click-to-zoom" (to see the full quality image).
- have plenty of RAM.
- have plenty of space.
It would be like purposefully taking screenshots, shrinking them down to 800x600, blurring them... for absolutely no reason.
And we are talking 11 MBs... 11 MBs.
And 4 MBs of that was taken up by 1 4k + 1 2k image.
A single podcast is larger than this.
It's giving me flashbacks of the "Inline Greek JPG" + "Tables as Images" days when I first started making books.
I am back in Nam!
Quote:
Originally Posted by KevinH
Which is probably why Photoshop has a "save for the web" setting to begin with.
|
That was because Adobe inserted a massive amount of metadata into the images.
You'd have a 1MB JPG with like 3MBs of garbage embedded in it.
"Save as Web" then removed the metadata + did some other potential settings (interlacing/progressive).
"Save as Web" for PNGs would also turn on higher compression instead of no/low compression + remove the Alpha channel if it's not used, etc.
Quote:
Originally Posted by KevinH
But can you and most people actually see a difference without "super zooming"?
|
The "super zooming" was just to easily show the artifacts. And just how rotten/lossy Indexing the wrong types of images is.
Screenshots should stay PNG (RGB) throughout the workflow.
Lossless changes should be done (cropping, compression). Never destructive (resizing).
- - -
IF filesize then becomes an issue, then you go in descending order:
- Throw away the Alpha.
- + Losslessly compress.
- PNG compression tools have gotten much better in the past 7+ years.
- As you can see, I was able to compress another 3.2MBs (~22% drop) out of those PNGs.
If it's still an issue, then:
- Index (if possible, within reason).
- Lots of text + little color? Yes.
- Entire colorful screenshot of Sigil? No.
If the image is too complicated/ugly using that method, then go back to source and:
- crop to focus on relevant portion
- recreate at lower resolution
- Like those original 4k images. Someone can re-screenshot at 2k or 1080p (or whatever).
If that's not possible, then:
- PERHAPS saving as a JPG.
- Like the current cover may be smaller as a JPG.
- PNG is usually larger for "photographic" images.
- At least the JPG would still be millions of colors (vs. <256 Indexed PNG).
But always keep the original lossless source.