it has been zero days since I started building another meme generation tool.
based on https://xkcd.com/2501/
=> More informations about this toot | More toots from foone@digipres.club
This image is actually an SVG that I'm intentionally exporting to a tiny JPEG to get lower quality text to match the original.
I'm using the xkcd-script font to try and match the original text as much as possible.
https://github.com/ipython/xkcd-font
=> More informations about this toot | More toots from foone@digipres.club
that was mainly a proof of concept. I think SVG is too limiting to do this dynamically, so I'm probably going to write something that uses embedded webfonts and a element, like the Death Generator already does.
=> More informations about this toot | More toots from foone@digipres.club
I also want to make it dynamically change the image size (like some Death Generators do) where it can just get taller if you type too much text into it.
That'd be fun for some examples of "experts"
=> More informations about this toot | More toots from foone@digipres.club
I'm also gonna have it do something I've been thinking about adding to the death generator:
Automatic alt-text generation!
It'll have a button you can press that copies some text that is the alt text for what you generated, which'll be a templated generic explanation of the comic plus whatever text you entered. That should make it easier to provide alt-text when posting these to places that want alt text
=> More informations about this toot | More toots from foone@digipres.club
someone REALLY needs to develop an ad-hoc standard for embedding alt-text in a PNG/JPEG, though. I'd love to have the site just hand you a file that automatically has alt text when you upload it
=> More informations about this toot | More toots from foone@digipres.club
maybe I'll have to Be The Change I Want To See In The World and develop my own standard and then add support for it in Mastodon
=> More informations about this toot | More toots from foone@digipres.club
but that's more work than I can manage on a day when I can't even sit at a desk, so I had to do all this work from a nearly-flat recliner
=> More informations about this toot | More toots from foone@digipres.club
@foone wildly... and the reverse is true too. Laymen have no clue how deep the knowledge of experts go. 👍
=> More informations about this toot | More toots from VeroniqueB99@mastodon.social
@VeroniqueB99 @foone For instance, mere mortals assume mathematics is mostly about something called "numbers".
=> More informations about this toot | More toots from ids1024@fosstodon.org
@ids1024 @foone 🤣🤣🤣oh I'm not even going there... but it is a valid point for any field really...👍
=> More informations about this toot | More toots from VeroniqueB99@mastodon.social
@ids1024
and economics is wholly about an abstract concept commonly called 'money'
&
most sadly, even contemporary economists believe this.
@VeroniqueB99 @foone
=> More informations about this toot | More toots from cpm@spore.social
@foone looks at the comic this is based on oh...I do actually know the composition of quartz
=> More informations about this toot | More toots from endrift@treehouse.systems
@endrift I wasn't sure if I did either, but I went "it's SiO2, right?" and then looked it up.
so apparently I didn't think I knew it, but I did
=> More informations about this toot | More toots from foone@digipres.club
@foone
There's 2 or so "standard" places to put it in the exif data; I haven't seen a js-only solution to use it right in the web UX, but I bet exiftool plus modern wasm tooling either already exists, or could trivially.
=> More informations about this toot | More toots from silvermoon82@wandering.shop
@foone hunh, PNG spec (circa 1996!) actually includes a way to do content warnings inside the PNG: https://www.w3.org/TR/PNG-Chunks.html#C.tEXt (not that I expect any viewer ever did a damn thing with it)
"Description" is arguably the right thing, and is already standardized! The only sticking point is that, as a thing around since 1996 with no(?) usage, it might be ruined by junk values in the wild.
Also I bet pngcrush throws these out in common usage.
=> More informations about this toot | More toots from LionsPhil@plush.city
@foone Ah, zTXt is preferred, to compress it.
PNG really is just the last image format we ever needed.
=> More informations about this toot | More toots from LionsPhil@plush.city
@foone aren't there standards already for that? Would be an interesting area to dig.
=> More informations about this toot | More toots from agorakit@phpc.social
@agorakit standards exist, yes, but are they supported?
=> More informations about this toot | More toots from foone@digipres.club
@foone PNGs has iTXt chunks, http://www.libpng.org/pub/png/spec/1.2/PNG-Chunks.html#C.iTXt, and it includes a utf-8 description. But i've never really seen tools use them (including my own).
=> More informations about this toot | More toots from drj@typo.social
@foone there's already a standard for this i think? i know pleroma implements it
=> More informations about this toot | More toots from kopper@brain.d.on-t.work
@foone apparently there's an ImageDescription EXIF tag. I wonder if anyone has ever used that.
=> More informations about this toot | More toots from virtulis@loud.computer
@foone https://www.iptc.org/std/photometadata/specification/IPTC-PhotoMetadata#description
=> More informations about this toot | More toots from glyph@mastodon.social
@foone bizarre that this isn't more broadly supported honestly
=> More informations about this toot | More toots from glyph@mastodon.social
@glyph well, that saves me some time
=> More informations about this toot | More toots from foone@digipres.club
@foone see also https://exiv2.org/tags.html Exif.Image.ImageDescription and Exif.Photo.UserComment
=> More informations about this toot | More toots from glyph@mastodon.social
@foone (Extremely telling about the nature of these specifications, given the relative length of the description of these fields and of Exif.Image.Copyright)
=> More informations about this toot | More toots from glyph@mastodon.social
@foone I've thought about this multiple times, but the thing is, alt texts for one and the same image might be hugely different based upon the context in which it is used.
I mean, sure, embedding something is probably helpful nevertheless, but it's not that simple to consider all requirements.
=> More informations about this toot | More toots from scy@chaos.social
@scy oh yeah.
But for a lot of images it's possible to get a good-enough alt text automatically, because a computer is generating the image from known components.
And I'd design it as a "default alt text", not as a definitive alt text. So you could leave it, edit it down to the important things, or replace it entirely. It'd just try to save some time if possible
=> More informations about this toot | More toots from foone@digipres.club
@foone My sole concern is someone will implement it into a camera app with some object identifier or AI bullshit, and other applications that might do things like redacting would not recognize and strip the metadata, causing leaks of information once something is posted and the original unredacted info is in the alt text.
=> More informations about this toot | More toots from dillon@cloudisland.nz
@dillon @foone if your process forwards unanalysed metadata it’s a bit of a stretch to call it “redacting.”
=> More informations about this toot | More toots from rvalue@infosec.exchange
@rvalue @foone Sure, but people often "redact" things in ineffective ways all the time, like covering text or faces using stickers, or they just use an application that's bad at it's job. The point is that metadata can continue to persist information about an image even when that information is no longer visible in the image itself, which can be dangerous.
=> More informations about this toot | More toots from dillon@cloudisland.nz
@foone Apple's Preview now offers to put alt text in. Grep grepme in attached.
=> More informations about this toot | More toots from adamshostack@infosec.exchange
@adamshostack I'm not seeing anything like that on my end, unless I'm using the wrong tools.
Maybe it got stripped. md5sum of the file is 66ada51dccc4dc29db2846f95a970f2b
=> More informations about this toot | More toots from griibor@mas.to
@griibor My mac somehow no longer has md5, but it's not a literal grep. you need to use exiftool (or something)
'exiftool ~/Desktop/Untitled\ 2.png|grep grep'
=> More informations about this toot | More toots from adamshostack@infosec.exchange
@adamshostack SHA1 is b60b22d2f7734f1c36e5e5fa4d587aaed5ad3a95
I see nothing like that come out of exiftool/mediainfo/eog either.
Here's exiftool's output: https://paste.debian.net/hidden/947ef172/
I assume the macos version of exiftool extracts some info that mine does not, for some reason.
=> More informations about this toot | More toots from griibor@mas.to
@griibor I'm on ExifTool Version Number : 12.84 which includes a Artwork Content Description field.
=> More informations about this toot | More toots from adamshostack@infosec.exchange
@griibor Also, shasum reports a different value ee567a393 ...
=> More informations about this toot | More toots from adamshostack@infosec.exchange
@griibor https://drive.google.com/file/d/1_ZE_OJNMA6gKw2vQpbM28AkW7CjFguYf/view?usp=sharing
text/gemini
This content has been proxied by September (3851b).