A graph of a sigmoid, s-shaped curve, labeled Camera_03 110-175 fov
A graph of three very flat lines, labeled Camera_03 110-175 pos
A graph of four diverging lines, labeled Camera_03 110-175 rot
And we got animation data!
A Terminal screenshot showing various model-node names, like "OB_SprayB_RotacionSpray" and "Luz Foco_Central" followed by a list of animated properties for that object (Position, Rotation, Color, Intensity, FOV)
Some progress on #Scrapland #reverse-engineering:
- Figured out animation track to node mapping
- Figured out the data structure for materials on disk (github.com/ReScrap/leve...)
“try water fox or ice weasel” try chill penguin. install flame mammoth. you just download bubble crab. literally just use boomer kuwanger. sting chameleon has a search bar, use it to find toxic seahorse.
I'm currently working on a ranking system for @ffglitch.bsky.social (not live yet, will probs only post to Mastodon as Bsky compresses the videos into oblivion)
I'm planning on using a (small) neural network and got some questions: www.reddit.com/r/learnmachi... (Reddit link because it's ~3k chars)
I don't think I ever shared this here: suricrasia.online/unfiction/ba...
"vibe coding" needs a direct competitor. I propose "woe coding", which involves locking a fresh CS student in a room with a ThinkPad older than me until they write a web application in ANSI C89 using ed(1)
So, I'm working on v2 of my ffglitch bot and need some help training the scoring function, for that I have made a small webapp (Beware of flashing colors, some of the glitched videos are pretty intense) talos.possum-gila.ts.net no tracking or anything
#glitch #glitchart please boost for wider reach!
So, I'm working on v2 of my ffglitch bot and need some help training the scoring function, for that I have made a small webapp (Beware of flashing colors, some of the glitched videos are pretty intense) talos.possum-gila.ts.net no tracking or anything
#glitch #glitchart please boost for wider reach!
Currently, writing normal software like a normal person (Pixel Shader 1.1 "assembly" parser/symbolic executor so i can convert it to Blender Material Nodes :) )
Suggestions are welcome, i'd want something with a permissive license (CC), videos of at least 30 seconds (preferably longer), ideally with an API that preferably spits out JSON
The old bot pulled from YouTube, but since yt-dlp needs frequent updates that's not really a reliable option, another option I was considering was Wikimedia Commons, but there's some unsightly stuff like surgery videos, etc. in there that I wouldn't want the bot to post.
So, I'm planning on reviving my ffmpeg #glitchart bot I made for mastodon, but having it post to Bluesky instead.
I'm now facing the problem of sourcing videos to "seed" the bot with (it'll eventually pull its own past posts and feed those through the glitch machinery™)
A very messed up looking 3D-model
gltf_validator output: Errors: /buffers/0: Actual data byte length (48768) is less than the declared buffer byte length (227584). Warnings: /meshes/0/primitives/0: Number of vertices or indices (4064) is not compatible with used drawing mode ('TRIANGLES').
Trying to add gLTF export to my #Scrapland level viewer, and it made a perfectly cromulent glb file... i think i'm holding it wrong
Apparently it's written in Go and Go binaries tend to be large, but 360MB .text section and 256MB .rdata sections seems a bit ridiculous
Also recently got into #Warframe fun game
"Big" Projects (github repos might be out of date):
- long-range route plotter for #EliteDangerous github.com/earthnuker/a... (basically spansh.co.uk but offline)
- Reverse engineering American McGee's #Scrapland to build modding tools (github.com/ReScrap)
And like 20 other side projects
Are #introductions a thing here?
I'm Daniel, known online as earthnuker, i do software development, reverse engineering, hacking (in the "a hacker is someone who tries to find a way to make toast with a coffee maker" way), and occasionally making 3D-Art
