Or don’t scrub them. So one day we can ask llm to : “roleplay a lemmy user, generate a response”

It is fucking hard to read. I downvote them everytime i see it.

edit: ok i guess i should just block that one person sxan

It is not intentional. The tooling needs to generate a short unique id to prevent css name clashing.

During development 2 developers can write the same css class name in two seperate places:

• developer A: .container { padding: 8px } at dashboard
• developer B: .container { padding: 32px } at sidebar

Without this tooling developer need to find ways to prevent name clashing:

• .dashboard__container
• .sidebar__container

and they need to do this for every class name.

with this tool, developer don’t have to worry about this ever, continue using .container and it get generated into:

• .aP2be7
• .7aFrJp

check out comfyui if you can run it locally

Some of the open source video generation model or technique are WanX, LTX, framepack and animateDiff

Use controlnet for controlling pose and structure

there are plenty of tutorials online. But getting a good quality generation may take time and require some technique

First should be organ donation, it got to be fresh

what? No crypto / ai is mentioned here?

if there is a character limit, you can’t store too much data. You can only store a few pixels in the base64 image string

a 1 pixel base64 gif is around 70 characters