I think people have been experimenting with integrating stable diffusion into text adventure games (like AI Dungeon) for a while now.
Prompt adherence is always an issue particularly in a text adventure game where position matters - e.g. (The crystalline amulet is sitting on top of the old wooden table. A fireplace at the end of the room gently crackles. To the east lies a door, cracked with age. On the west wall you can see a painting of the lords of the manor. etc)
Flux could probably get close - though it clearly doesn't have any examples of an amulet in its training data. :)
The later Infocom games — I’m thinking especially Moriarty’s evocative Wishbringer, Trinity, Beyond Zork — have much improved writing in general; it might be interesting to see what an LLM actually does with those.
A Mind Forever Voyaging has very good writing as well (and less in the way of puzzles). Of course, Hitchhiker's has great writing too.Not sure I ever played Wishbringer or Beyond Zork but Trinity is good.
Eh, as a controls engineer, I get a bit of Gell-Mann amnesia watching his content on drives and motors. There's a complete lack of tuning described in the video, it seems a lot of his performance complaints are due to conservative out-of-the box defaults.
The stuff he builds very cheaply with little more than some wood, Python, and Raspberry Pis is impressive, and he deserves all the kudos for building cool shit and putting it on the Internet (I'm just a consumer and critic, not a creator in comparison). But serial control from an interpreted script instead of CAN/Ethercat messaging from a motion controller or PLC is not the way these products are typically used in industry, and most people don't run the defaults.
There's definitely a niche for hobbyist-grade, student-grade, or lab-grade motion products that name brands like Beckhoff, Omron, Fanuc, Rockwell, and Siemens largely ignore. You could get 4 axes from DMM for the price of 1 from most of those vendors. And while simple serial commands can make interesting things happen with the DMM unit, you'd have to scale a daunting cliff of a learning curve just to get a motion axis initialized in their massive, standardized, proprietary, legacy ecosystems.
Again, no disrespect intended: I've invested thousands of hours into building custom, multi-million-dollar machine tools at my day job and instead of challenging myself, going to the workshop and turning on a camera when I get home, I've vegetated on the couch, entertained, and sometimes educated by his content building a milling machine or lathe out of wood. But this "servos vs. steppers" debate really only applies to low-cost, simple, hobbyist-grade equipment, and isn't such a big topic in the industrial space.
> I've invested thousands of hours into building custom, multi-million-dollar machine tools at my day job
One key is find a hobby that is enough different from your day job you are not burned out of doing it. I can write code at home and sometimes I do - but most of the time I'm burned out after doing that for my day job. However I can still bend the sides of a ukulele, use CAD to design a new switch housing for some manual machine, practice trumpet, or other such tasks that are not related to my day job. I personally am not interested in editing a video (which takes a lot of time to do well) so you won't see me on youtube, that too is something I could do if I was interested in it.
Though with kids often all I have time for is cooking a meal before getting them to bed and then I'm off to bed myself. I wouldn't trade it for the world, but there is limited time and so there are a lot of things I want to do that I don't have time to do.
Please re-watch the video, but THIS TIME PAY ATTENTION. I do talk about the motion parameters, even show the code on the screen briefly. I even show the effects of changing the parameters.
> But this "servos vs. steppers" debate really only applies to low-cost, simple, hobbyist-grade equipment, and isn't such a big topic in the industrial space.
I assume these videos are targeted at hobbyists… I can’t imagine people in your position using him as a source of knowledge.
Yeah. The first (and only, so far) time I came up against a servo controller from AMC (https://www.a-m-c.com/) my initial thought was "why the hell is this so difficult?" Sure, the product manual has everything you need to know about how to program the device. But that manual is also practically unreadable unless you already know how to program the device
Contrast with Teknic where I could get the servo drive up and running in a few hours because of an actually readable product manual and plenty of sample code and a Windows DLL to make everything easy.
There's definitely opportunity at the lower end of the market.
Yeah Matthias is fun to watch because he does a lot of hack stuff with plywood that works better than you'd expect. He does a lot of stuff with a lot of confidence, a lot of it dangerous. He's a smart guy for sure, so for what the setups are, the data is interesting. However, the problem is that people extrapolate beyond the setup. It's rare that the way he does something is a _good_ way to do something. He doesn't demonstrate himself to be an expert in much beyond software and hacking stuff together with wood, and I don't think he claims much beyond that.
I’ve never seen him doing anything that I would call dangerous. Not in the sense that he, someone or something is at risk. He make things that can fail, but it will not reault in an injury.
I’ve seen quite a few. His pantorouter series is very good and I have a lot of respect for doing this sort of thing without CNC. His most recent series building a milling machine is asinine, but fun to watch.
https://en.m.wikipedia.org/wiki/F._D._C._Willard