Screenshotting Chinese Web documents for WeChat

Reasons for sharing a screenshot instead of a link:

1. Technical problems on the receiver's side prevent proper display of the page;
2. The server is unreliable or rapidly changes;
3. The page itself is innocent but the recipient's network infrastructure blocks the domain it's on because of something else on that domain or by mistake.

Reasons not to use screenshots:

1. Poor accessibility: if the recipient needs to copy/paste the text into dictionaries or other learning tools, read it with a speech synthesizer or reflow it to a different width, then this may require OCR which is unreliable;
2. No interactivity: pages where the recipient must enter information or follow links won't work as screenshots.

Creating a narrow-column screenshot

My Web Adjuster can be used with Headless Chrome (or Firefox or PhantomJS) to create an appropriate "narrow-column screenshot" by configuring with a js_interpreter setting (ideally HeadlessChrome, but if you can't install that then you could try the older PhantomJS), and with --js_size=400 --viewsource and append .screenshot to the end of the URL you give it.

On GNU/Linux I also recommend creating a ~/.fonts.conf or ~/.config/fontconfig/fonts.conf file (depending on which one the system uses) to switch off anti-aliasing, as non-antialiased text is usually more readable and makes for better compression in PNG files. Here is a non-antialiasing fonts.conf.

There is a possible loophole in Microsoft's YaHei font license that might allow you to use that font if you want to make the Chinese text look a little nicer at small pixel sizes. Usual disclaimers apply---I am not a lawyer.

Size limits

You may also assemble up to 30 images into a WeChat "Note", to be displayed as a continuous stack of images. These may display blurred until tapped unless you limit each one's vertical dimension to about twice the horizontal (e.g. 400x800, giving you a total vertical capacity of 24,000 which is typically 3-4x what you'd get if doing the same thing in individual images if the page makes moderate use of graphics). The more you exceed this per-image vertical size, the more the WeChat app will blur the image before it's tapped, even though the image's physical display size remains in correct proportion.

If you wish to crop out an irrelevant section of the page---for example a box of links at the bottom, which won't be usable on a screenshot---you can either write a site-specific rule for it in Web Adjuster, or else load the screenshot into The GIMP, select and cut unwanted full-width areas to transparent (making sure not to cut from the text or credits---don't misrepresent or plagiarise!), export to PNG, and use a command like:
python2 -c 'f="test.png"; from PIL import Image as I; i=I.open(f); b=i.tobytes(); c="".join(b[n:n+4] for n in xrange(0,len(b),4) if ord(b[n+3])); I.frombytes("RGBA", (i.width, len(c)/i.width/4), c) .save(f, optimize=True)'
(command works only if you cut full-width areas).

If the page has many pictures but little text, consider using JPG instead. With ImageMagick, use:
convert input.png -define jpeg:extent=300KiB output.jpg
(this option was introduced in version 6.5.8-1 and uses 'binary chop' to find the most suitable JPEG quality setting within 8 tries; if you have a version between 6.9.1-0 and 6.9.2-4 you should also add -quality 100 which is ignored by higher versions). If you are sequencing multiple images into a "Note", you can choose between PNG and JPEG on an image-by-image basis.