I just learned about #BlurHash while working on a #mastodon client app. https://blurha.sh/

tl;dr: it's a simple hash to generate a client-side blurry placeholder while the real image is loading.

#blurhash

Now further rabbit hole of looking at various instance's Instance endpoint https://docs.joinmastodon.org/methods/instance/#v2, e.g. this very instance's: https://fosstodon.org/api/v2/instance but also others like @grafana's
https://grafana.social/api/v2/instance and discovering Mastodon Glitch edition https://glitch-soc.github.io/docs/
and a reminder of how images are blurred here for loading using #blurhash https://blurha.sh/ that looks pretty cool and would love to understand the process / see if it can be used well in code I maintain.

TIL: Characters 3-6 of a BlurHash form the average colour of the image (sRGB) – easy to extract if that's all you need, without decoding the whole thing.

Using this for my post media viewer, where the full BlurHash might be too distracting. Could use a saturation boost in most cases though. (Image by @mark)

#BlurHash #SwiftUI #Mastodon

Custom #SwiftUI #Blurhash View and on-device Image Segmentation – I've since found the private subject segmentation API powering Photos.app etc., might revisit this.

Remember my experiments with #blurhash recently?

https://mastodon.social/@dec_hl/109469332447530799

You can (for now) generate a blurhash image of your mastodon account name by mentioning my new bot: @blurbot

Rate limit is one request per user per day for now and it might not be running 24/7...

#CreativeCoding #BotsOfMastodon #GenerativeArt

@dec_hl Your account name as #blurhash

@smallsees blurhash-python made it out of NEW, it is now in sid https://tracker.debian.org/pkg/blurhash-python #blurhash #DebianSid #python

Some random examples of random #blurhash

#CreativeCoding #JavaScript #MSDOS #GenerativeArt

I recently played a little with the #blurhash algorithm. I also integrated a fast C version of the decoder into #DOjS.
Today I created some sort of #blurhash explorer.
You can either enter arbitrary text or you can randomly generate a hash and the result is drawn to the screen.
You can also save the result as PNG to disk...

#CreativeCoding #JavaScript #MSDOS

Mastodon uses https://blurha.sh/ for image previews and CWs. These are basically small strings transferred for each image, which are then converted into a blurred representation of said image.

I was wondering what would happen when you feed the algorithm readable text word-by-word instead of real hash codes. This is what I found out...

https://openprocessing.org/sketch/1761462

#CreativeCoding #p5js #BlurHash #OpenProcessing

I'm trying to understand how to use BlurHash[1] to load the final image. Am I forced to use JavaScript for it?

I've read something about a placeholder. Can I use <img src="image.jpg" placeholder="image-blurhashed.jpg">? Can't try it out locally since the image is loading super fast!

[1]: https://blurha.sh

#blurhash #javascript #html

It would be kinda cool if #blurhash became accepted as an image format in its own right.

@WAHa_06x36@mastodon.social Is the #blurhash format versioned? Has it changed at all|did it used to be different from what Algorithm.md currently specifies? I'm thinking of using it for something and concerned|curious about forward compatibility.

@anglk @etc the internet went down before the image loaded so I only got the #blurhash

@ConnyDuck @charlag I love #blurhash! thanks @WAHa_06x36 for making it!

Just tagged v1.0.0 of my #Blurhash implementation: API stable && documentation is complete - https://github.com/buckket/go-blurhash

Added tests and documentation. Rdy for consumption, I guess. #blurhash

And here is an update of the #python #blurhash module already, with a bug fixed and also some added functionality and an example usage script.

https://github.com/halcy/blurhash-python

https://github.com/halcy/blurhash-python/blob/master/example.py

Version 1.1.0 on pip! 🐍

I think #Mastodon’s new #Blurhash feature (obfuscating nsfw-ish images in a colorful very blurry fashion) is a pretty good idea – and works well in this first iteration (the best setting might be debatable).

Does somebody not like it? Are there any objective arguments against it?

Who else is using it?

Are there other use-cases besides the stated one?