abdullah

Left: Normal ASCII shader Right: Shape-aware ASCII shader ASCII characters are chosen based on their shape, so you get clearer edges Vectors are calculated and compared with WebGPU compute shaders It's based on an altered version of Alex Harri's approach. I'll link below

The slop increases exponentially. It quickly becomes a large chunk of the codebase Then you end up in this position: - you either continue like this, but every little change requires that devs trudge through layers and layers of slop, slowing them down significantly and leaving the app riddled with bugs - you suggest a refactor, but absolutely nobody wants to hear that

What happens to your codebase when you stop reviewing your agents' output

@asanwal you could use a live transcript ai model to make the screen flash bright red every time you say "you know" i assume you'll learn the lesson after a few red flashbangs

Just looked at a transcript of a call I had and I said "you know" 110 times in 35 mins What's the most effective way for me to get rid of this?

@JohnPhamous are you doing that thing where you search using a generated hypothetical answer to a query instead of the raw query itself? or like expanding/rewriting the query in some other way before running the search?

can now semantically search for everything ive ever seen no more digging through slack/notion/history - records all text nodes on screen - stores embeddings locally - source app stored as attribute most used workflow is searching tweets ive seen

@gentlycarved @cpendas @vanschneider or are you talking about node based vibe coding tools?

@gentlycarved @cpendas @vanschneider > dropdown of models you're talking about image models right? very curious how you use image models when you're designing something. is it mostly for assets or like for ideation?

I feel like I’m the only one who doesn’t find node based design tools intuitive

@asciidiego They post these sage tweets every hype cycle. When the hype dies they start again with the new Current Thing x.com/naval/status/1…

@asciidiego can't take anything VCs say seriously anymore

Pure software was never investable. No one invested in Windows 95, or Mac OS Aqua. People invest in organizations that make software—whether it’s with people, AIs, compilers, or whatever. People invest in higher order orgs that make stuff that people buy—or have a shot at that.

@lumosterris Also I don't really understand why he uses circles either

@lumosterris That's Alex Harri's blog post not mine! That's why I said it's based on his approach in the original tweet. In his version he uses sampling circles but I used a subgrid of the cell instead. x.com/verse_/status/…

Left: Normal ASCII shader Right: Shape-aware ASCII shader ASCII characters are chosen based on their shape, so you get clearer edges Vectors are calculated and compared with WebGPU compute shaders It's based on an altered version of Alex Harri's approach. I'll link below

Just to be extra clear this is not my blog post. This is written by Alex Harri. I couldn't find him on X so I can't tag. This is why the original tweet says "based on Alex Harri's approach"

The way I build the vectors is slightly different: The screen is divded into cells. Then I divide those cells into a subgrid. The average luminance of each cell in the subgrid is what I use to build the vector I think the math also differs in some places because it all runs in compute shaders alexharri.com/blog/ascii-ren…

@RogerAlsing no, I think this is because I had the edgeContrast var turned up. It alters the cell vectors a little to increase the visual separation between different colors in the image generally makes the image look a little better especially at the meeting points between the cubes faces

@verse_ Was the char pixel fit computed from some other font? e.g. some older FixedSys or similar? "i" vs "yz" and "." vs "__", looks somewhat sus tbh.

@Lngly_ pmndrs leva github.com/pmndrs/leva

@verse_ very cool! what's that debug ui?

@DonnyLi2001 Yeah that's what I did here. I decided to do the glyph vector precomputation in a compute shader too just because I thought it was cooler. But it only runs once on mount, and then re-runs when variables in the control panel change

@verse_ Shape-aware glyph selection looks great. One trick is precomputing glyph edge maps and doing a small KNN in the compute shader so per-frame cost stays low as resolution scales.

@taigrr very cool

@verse_ Nice! I published this a few weeks back: github.com/taigrr/trophy

@dc1b3l @ID_AA_Carmack I can't remember. what did I say that reminded you of this?

@ID_AA_Carmack @verse_ remember you posted about this a while back

The modern age has richly rewarded people with a combination of high intelligence and high agency. Now that many aspects of intelligence are successfully being automated, it seems likely that people with relatively lower intelligence but exceptional agency will come into their own if they are willing to egolessly accept AI advice. Imagine a ruthless criminal that completely trusts everything their always-on AI glasses are telling them, knowing that it is carefully looking out for their best interests and isn’t scheming to betray them.

@baoskee imo with SaaS, the "service" part is the main thing you're selling, and the "software" part is like a necessary evil that you have to deal with to be able to do that I think that's why everyone has been trying to kill the need for code since day 1. but it just won't die

people who think that Salesforce moat is software are retarded have u seen their software? everyone hates it their users hate it. it's because their users are not the buyers Wall St really missed the mark on this one

@doctorbaytas 🔥

@verse_ similar effect in prod: baytas.net

when you have a list of links like this, you can get this nice trailing effect if you only transition colours when the cursor leaves