Anton Mikhailov

Computer Programmer.
Worked on PSVR & PS Move, Dreams and Substance 3D Modeler. ALife enjoyer 🌱
Crafting by night.
Views my own.

[Anton Mikhailov @antovsky]

I've made a live calculator to easily play with computations, formulas and plots.

https://t.co/03U9uORoar

Builds for Windows / Linux / Mac in thread.

[Anton Mikhailov @antovsky]

@SanderSkjegstad @lyncpk Thank you 🥲

[Anton Mikhailov @antovsky]

@opfleyr @DanielcHooper @nicbarkeragain Forget TigerBeetle and servers for a moment, we're talking about blender.
Are you saying "no good solution exists, period"? Sounds like the whole strategy of dynamic is to "hope" that you don't run out and "detect pressure" (I am not sure what this means in practice though..)

[Anton Mikhailov @antovsky]

@DanielcHooper @nicbarkeragain @prathyvsh Yes, I have the same zine :) (I ought to frame mine too!) I hope they are doing well ✊

[Anton Mikhailov @antovsky]

@opfleyr @DanielcHooper @nicbarkeragain In the dynamic scenario, what happens if the user wants 150 meshes and 150 bones but doesn't have the memory for it? What is your strategy for the user experience when a limit is hit?

[Anton Mikhailov @antovsky]

Brett is hugely inspired by Papert for sure, but something about it being interactive on the computer takes it to a different level imo (though ironically he's moved "back" to more physical media with dynamicland 😅)
I very much love Mindstorms, but I also wonder what we're missing, given that those approaches haven't taken off 🥲

[Anton Mikhailov @antovsky]

@nicbarkeragain @prathyvsh Brett is the original "holy shit I didn't realize education material can be so high quality" guy, so in a sense you're picking up that torch subconciosly even if you haven't seen it ✊

[Anton Mikhailov @antovsky]

@nicbarkeragain 's latest memory video is the first of the series that will dive into this topic, so stay tuned :)

But at the high level, its worth asking what happens with *dynamic* memory approaches as you hit real systems limits, and how they degrade/fail and whether that's any better...

So my counter question is: "How do you do it with dynamic memory, assuming you want to handle the edge cases of your blender scene getting gigantic (larger than ram or vram)? Whats the degradation and failure look like to the user?"

I think NASA/TigerStyle are the logical conclusion of actually wrestling with that question honestly and demanding a high quality answer, rather than passing the buck to the OS or the user (other than perhaps asking them to set a budget for your app based on how much they want you to handle etc, which btw blender does with its undo stack and a few other features...)

[Anton Mikhailov @antovsky]

@kolemannix Much more eloquently stated ✊ (but very happy I'm at least in the ballpark 😅)

[Anton Mikhailov @antovsky]

@Anton_Dankov @nicbarkeragain @thedevbirb That's a good question! At the moment CPU only because we rely on VirtualAlloc, but we're mulling over GPU allocators in the background 🤔

[Anton Mikhailov @antovsky]

Static Allocation Enjoyers Unite ✊

[Jason Rohrer @jasonrohrer]

Hot code reloading, rewinding, time jumping, etc. in Chessamphetamine on the @wookash_podcast
https://t.co/VcIClaaRrO

[Anton Mikhailov @antovsky]

@ikrimae Amazing, thanks for the link 🙏

[Anton Mikhailov @antovsky]

This kind of use of LLM tech really excites me, because it's well matched for what that architecture is designed to do. Also, because we as humans don't have the first intuition about how animals might think or communicate, this seems like a fantastic tool to probe that space!

[Ole Lehmann @itsolelehmann]

i think AI is about to uncover an entire hidden world of animal communication

and what it's already found is insane...

> AI found evidence that elephants call each other by name. when researchers played one its name-like call, it approached the speaker faster and called back more, revealing a form of personal naming once thought almost uniquely human

> AI analyzed 53,993 marmoset calls and found tiny monkeys may use names too. entire families use similar labels for the same monkey, suggesting they learn names and dialects from each other

> AI learned to tell what Egyptian fruit bats were arguing about: food, mating, sleeping spots or personal space. it could even partly identify who was shouting at whom inside a colony of thousands

> AI discovered a “phonetic alphabet” in sperm whale clicks, with rhythm, tempo and tiny timing changes combining into at least 143 recurring patterns. their communication system is nearly 10x more expressive than scientists previously believed, and we still don't know what most of it means

> an AI trained on 1.5 million zebra finch calls held real-time vocal exchanges with living birds, generating calls as they spoke and getting them to respond with the same timing and flexibility they use with other birds

> Google trained an AI on decades of dolphin recordings to predict what sound comes next, then paired it with an underwater device that gives objects their own synthetic whistles. the goal is for a wild dolphin to copy a whistle to ask a human for a specific object, creating a tiny shared vocabulary between species

> AI combined microphones worn by wild crows with nest cameras and uncovered quiet calls that may announce when a crow is arriving at the nest and help entire families coordinate caring for their chicks

> a robot bee performed the waggle dance real bees use to share the direction and distance of food, and the bees changed their flight paths based on its instructions. researchers effectively sent a destination into a hive in bee language

[Anton Mikhailov @antovsky]

I'm in.

I too, have no love for z index.

Adobe Modeler's UI has an explicit 3rd dimension for stacking (similar to Clay's linear order, but in Z), because it's in VR so that's necessary. I had to do a separate thing for popups/dropdowns on top of that though so yeah special kind of hell indeed 🫡

[Anton Mikhailov @antovsky]

@NicholasPe34164 @DoctorGester That was by design.
(Ironically when I asked it first to add dark mode, it made only on the article section 🫠)

[Anton Mikhailov @antovsky]

@DoctorGester Sorry, it's coded by LLM (seems like it would be hard to experience on mobile anyway though)

[Anton Mikhailov @antovsky]

@kinzel_john You be the judge! :)

[Anton Mikhailov @antovsky]

Some time ago, I was curious what would happen if I tried to be totally neutral and entirely devoid of opinion or expertise, when asking an LLM to generate code. It resulted in an amusing sort of fiction (wait for the twist):
https://t.co/RE6UUQsezw

[Jonathan Blow @Jonathan_Blow]

When operating LLMs in a conversational mode, it's clear that they are deeply subject to suggestion, and spend a lot of energy just following the implications built into the questions you ask.

"Estimate this quantity." It does so, gives you a long explanation, and it seems convincing.

"Wait, don't you think this number should be higher, because of reason X?" It will say yes of course, I didn't properly consider X, here is a higher answer.

"Wait, don't you think this number should be lower, because of reason Y?" Yes of course. It will give you a lower answer.

People use these for programming.

[Anton Mikhailov @antovsky]

@DoctorGester I'll pass the feedback on to my developer! 🙏

[Anton Mikhailov @antovsky]

@Hathaway_JA Maybe one day...

[Anton Mikhailov @antovsky]

@GremlinIndustry