This is my exact setup as well and dear lord gemma is absolutely batshit insane. I'm trying to get a self-reflection and confidence loop going now, but it does feel like it's not the local resources, it's the limits of the training. Dedicated coding or dedicated real-world task models would be a good optimisation.
The roadblock to this is you seem to have to build it yourself. I've noted that none of the current cloud models are very good at building a replacement for themselves, and there's significant work that needs to be done to make a local LLM reliable in any way. I haven't found a single standalone package that makes setting them up easy. Sure, I can run Hermes Agent and a model, but getting the self-reflection loop in and all of the other stuff the need to actually be good? I'm still at it, trying to get anything to work reliably and factually.
Runescape has a similar thing going on. What everyone's not understanding is not that improvements are bad, lots of the player base also want to see games evolve. It's just that a significant proportion of the audience also like The Old Thing and they're often more engaged as an audience in the first place.
You hear this story over and over about every kind of software.
There are two audiences every successful developer needs to cater to: the "I wish this did X, I want new features" side and the "I liked it the way it was" crowd and they're distinctly different groups.
For a long time I produced a popular technical art asset for video games and even I realised I needed to include every single version of the tool with every single installation. If a developer has to go and find out how to get "the right version" at all then I'm 90% likely about to lose a user.
Focusing on the "we did this right just keep going like that" and really understanding WHAT you did right and WHAT people like is really important. It's really hard to be impartial when you made the world rather than consumed it, but always take the win.
It's very familiar. My father was abusive towards me and so I tried to disturb him as little as possible. He was extremely supportive of my younger brother, however, and we both turned out very differently. He turns to my parents constantly for help (and receives it) and the few times I have I've only been brushed off, and so I've always done everything myself, even if it meant spending long periods living in poverty with no footholds upward.
Years later I'm nearly 50 and have transitioned to female. My now divorced mother decided she likes having a daughter and so suddenly all of this support has materialised and with presumably less time to live the life I missed out on I'm making a conscious choice to start being a little dependent on it and shortcut some things that would have taken longer otherwise.
It's hard to describe but it feels like I'm living a more normal, less marginalised life now. I definitely have more friends, where previously I chose to avoid having any because I felt like having friends meant sharing burdens that were mine alone to carry.
I also see a lot of very similar behaviour in the wider trans community where most people's axis of behaviour revolves around some facet of not having support. We don't all handle it in a healthy way, but I now do my best to help my community find stability and adjust their expectations to having better outcomes.
Different people engage in different ways. I think OP is telling us what got them over that initial habit adoption hurdle. Sometimes spending money on something makes you use it.
My example case is I paid for a year of gym membership up front so I could feel the total I'd spent more tangibly. That got me to go regularly, even if I was just showing up to recovery stretching at first. Now I enjoy going and I look forward to it, and I didn't need a PT to motivate me this time.
It looks like it has a 3d draping view, but it didn't seem very good yet. Check out Marvellous Designer, that's what anyone doing digital-first uses.
Drapey clothing is probably the easiest to freehand without a pattern though. It's accurate fits that need more measuring, planning, temporary stitching, test garments, etc.
It's not an accurate recollection at all. In 1990 a couple of us 12 year olds snuck into the university library to use the web to look at the Marathon website. It took 5 minutes to load some trivially-sized gifs and a tiny amount of HTML. They had a pretty decent connection for the day.
Web pages took a minute to load, now we're optimising them for instant response.
It's not that different from how some creative Mac games were doing 3d lighting on 2d textures prior to 3d accelerated hardware being available. The neat part here is that it runs on a Gameboy Colour.
Check out Rustler Grand Theft Horse on the Epic Games Store, it's the same top-down format, very GTA-like but set in medieval times, yet has all of the modern banter. It's so great.
A couple of months ago I had to write a CDL-based triangulator to solve a use case where ear-clipping doesn't support the shapes we had.
We had no AI policy at the time so I had to read up on CDL and implement it by hand. The concept is straight-forward and I also targeted regularity as acceptance criteria for the mesher, but making it optimal was hard.
I ended up having to park it after the ticket ran out of time, but now we have an AI policy this was the first problem I gave it. What it put out was similar but better structured and more informed.
I worry a little that AI will stunt our problem-solving in 20-30 years, we still need new algorithms, even when ML is capable of producing a model that can do the same thing. But right now it's much better at the things we've already done than we are.
Does anyone know if you can disable or lower the volume of its speaker for everyone who might ping it? I use one on my cat's collar and I'm already concerned about it being harmful to her hearing should I ever put it into lost more.
I think the answer is that by the time AI can replace every function you do, it's also replaced everyone else and the world will either already have or will need to change radically.
I personally hope that the future becomes a UBI consumer-as-a-job thing, minus too much of the destructive impact that current consumerism has on the world.
The minimum is four RGB-only cameras (if you want RGB data) but adding lidar really helps.
The standard pipeline can infer a huge amount of data, and there are a few AI tools now for hallucinating missing geometry and backfaces based on context recognition, which can then be converted back into a splat for fast, smooth rendering.
I don't like being devil's advocate on this because I am strongly against the invasion of privacy at that point in the investigation, but without that data, they'd just take a bit longer to have identified the members of the insurrection. There's varying degrees of data you can glean from cellular networks as well, right down to "it was definitely this person, the phone logs show a FaceID unlock at X time" and that action can be inferred by network logs, all information that carriers have retained for over two decades.
What it does become is a data point in an evidential submission that can strengthen a case that could otherwise be argued back as a bit flaky. It's similar to DNA evidence in that it's not actually 100% reliable nor is the data handled forensically at every stage of collection, but it's treated as if it is.
I think it's weighted too heavily in evidence and should not be used as a fine-toothed comb to sweep for "evidence" when it can be so easily tainted or faked. At the same time, I'd love to see the current members of the pushback against ICE using this data fallacy against future prosecutions. "Yeah, I was at home, look" and actually it's just a replay of a touch or face ID login running from a packaged emulator, or whatever signature activities meet the evidential requirement.