zavertnik

This was outside an immigration facility and not a result of the supposed doxing.

Same here. Easily one of the worst parts of AI in my personal life. They don't even know about en dashes :(

I completely agree with your sentiment. Its basically Right to Repair's cousin! Fortunately, there is still time!

Some context for those just learning about this:

Prior discussions: https://hn.algolia.com/?q=stop+killing+games

Official website: https://www.stopkillinggames.com

It's tedious.

Worse, it's distracting.

It's hard not to see this as a continuation of the American corporate interests controlling the media their population consumes.

Do you find the natsec argument to be compelling considering:

TikTok I think has the largest share of American's attention out of all the social media?

Yes. Copyright issues are handled in the way you'd expect. Take down requests outside of copyright are reviewed internally without guarantee of outcome.

https://help.archive.org/help/how-do-i-request-to-remove-som...

This is something I've wanted to do for a while! I wish Samsung still produced their 55" 8K displays-- 8k @ 55" gives you effectively the same PPI as a 27" 4K display. Maybe someday.

Cool project! This post also appears to be the second result on google when searching "vercel app" (as of +7 hours from your post), which I also found interesting.

I wasn't aware of Roccat! sounds like what I'm after, especially without the gamer aesthetics. The Roccat Nyth looks close to what I'm after, but I'm pretty married to my muscle memory with using 4x3 buttons (of which are all in use, in both layers).

Because it is possible with G-Hub, I was just curious if anyone knew what the gap is between SteerMouse's support and G-Hub's support for the seamless G-Shift/shift key experience. I don't know enough about hardware to make a guess.

I'm 100% fine with using a separate OS to config the mouse, since my layout is not app specific. Despite the button layout difference, this will definitely be the mouse I try out next.

Haha, of course!

So when you click a button on the G600 (and most other mice with side buttons), the button fires when you press down, just like the Mouse 1 or Mouse 2 button. On the G600, there is a third click button to the right of the right click, which is called G-Shift. When pressed, all of the side buttons have secondary assignments. Since you have to hold G-Shift to access this other layer, the macros are often referred to as chords in mouse customization software like Steermouse, since it requires two buttons to fire.

To configure this, you need the G-Hub software, which is in a nightmarish state on macOS. To get around this, I use Steermouse. Steermouse lets me get around this, however with one trade off. If a side button has two assignments (one when pressed by itself, another when pressed with the G-Shift), then the button does not actually fire when its pressed, but instead fires when the button is depressed/let go of. I imagine this is just how Steermouse handles buttons which have more than one assignment.

I haven't found a suitable replacement that is as robust as steermouse. Its one of the first apps I install on my mac, but this is one killer feature that I've only found in the G-Hub app.

I have used SteerMouse for years after giving up on the dumpster fire that is Logitech's G-HUB for macOS for my G600, which has 12 side buttons + a the G-Shift button which when held acts a modifier for additional macros.

My issue with SteerMouse is that when creating chord macros, it forces the original macro to only work when depressed, rather than activating upon press.

I haven't been able to find a suitable replacement. Curious if anyone here on HN has worked around this in any way?

The argument is basically that for the average US national, mass surveillance and propaganda by US tech is more of an issue.

Correct me if I'm wrong, but doesn't this track 100% with the bills stated intent: to prevent foreign adversaries from having this kind of influence?

I don't think the US is making an argument that data collection, surveillance, and propaganda are bad and then selectively only choosing to enforce these ideals on China. I think the US is making the argument that a company which belongs to an adversary and operates an a domain that is out of bounds from US regs/enforcement is a NatSec issue.

I'm seeing a lot of comments like this, but I have no idea what your camp is trying to say.

The bill isn't a criticism of TikTok's operation, its a criticism on its ownership and how that ownership + influence creates an exploit that poses a threat to NatSec.

This kind of reactionary framing feels like an attempt to put Facebook and TikTok on a level playing field, but the premise of the bill is about tech ownership and influence by companies of or from a foreign adversary.

Facebook is a US company. What am I missing here?

How does this invalidate the argument Warner is making?

Is China really a foreign adversary?

Yes, they are officially recognized as a foreign adversary. They're the first country listed:

https://www.ecfr.gov/current/title-15/subtitle-A/part-7/subp...

I have had the same results with multiple VPN providers for the better part a month now.

I daily drove a hackintosh for years until I recently pivoted to apple silicon. I was a very enjoyable experience for me. The success and reliability of a hackintosh is really dependent on your hardware configuration. I had lucked out that my desktop tower that I had built years prior just so happened to coincide almost 1:1 with hardware requirements for a golden build. (6700k, 64gb ram, Vega 64, compatible wifi/bluetooth pcie, compatible m.2 controllers, z170 mb which is well known in the hackintosh community, etc.)

Being able to have a modular Mac was really something and I exploited that to tailor my machine to my use case (television/video production). I never had issues with bluetooth or WiFi, nor did I have ever have an issue with Apple's services like iMessage/Facetime.

What sucked about the process was staying current with system updates. Updates within the macOS release went without a hitch, but my hardware was aged out in newer macOS versions which made upgrading a bit too much like surgery, and since this hackintosh was my production device, that wasn't something I wanted to roll the dice on.

Having switched to apple silicon, I do kind of miss that freedom, but I've found that same freedom just by doing things a little less hacky. Instead of a board I can add drives to, I just setup a NAS, instead of using an old PCIE HDMI capture card, I got a more modern USB one, etc.

For a long time, Hackintosh was an opportunity to do things my way, and that experience led me to learning experiences that have improved my day to day that I otherwise may not have learned. It was a freeing experience. Today I still do things my way, but these days my way is more focused on convenience for the things that should "just work" so I put my attention on things that matter, rather than things that shouldn't, such as modifying my EFI before a macOS update to trick macOS into thinking I have the iGPU of a newer chipset because Apple dropped support for Skylake on a new release.

Good times, the headaches were worth it in hindsight.

From my experience, the thing that makes using AI image gen hard to use is nailing specificity. I often find myself having to resort to generating all of the elements I want out of an image separately and then comp them together with photoshop. This isn't a bad workflow, but it is tedious (I often equate it to putting coins in a slot machine, hoping it 'hits').

Generating good images is easy but generating good images with very specific instructions is not. For example, try getting midjourney to generate a shot of a road from the side (ie standing on the shoulder of a road taking a photo of the shoulder on the other side with the road crossing frame from left to right)...you'll find midjourney only wants to generate images of roads coming at the "camera" from the vanishing point. I even tried feeding an example image with the correct framing for midjourney to analyze to help inform what prompts to use, but this still did not result in the expected output. This is obviously not the only framing + subject combination that model(s) struggle with.

For people who use image generation as a tool within a larger project's workflow, this hurdle makes the tool swing back and forth from "game changing technology" to "major time sink".

If this example prompt/output is an honest demonstration of SD3's attention to specificity, especially as it pertains to framing and composition of objects + subjects, then I think its definitely impressive.

For context, I've used SD (via comfyUI), midjourney, and Dalle. All of these models + UIs have shared this issue in varying degrees.

I thought it was "Pea" in the sense that the site is a small .js codebase. Now I'm very curious of the lore...are all 3 of us wrong?

Nice post, OP! I was super impressed with the Apple's vision framework. I used it on a personal project involving the OCRing of tens of thousands of spreadsheet screenshots and ingesting them into a postgres database. I tried other OCR CPU methods (since macOS and Nvidia still don't play nice together) such as Tesseract but found the output to be incorrect too often. The vision framework was not only the highest quality output I had seen, but it also used the least amount of compute. It was fairly unstable, but I can chalk that up to user error w/ my implementation.

I used a combination of RHetTbull's vision.py (for the actual implementation) [1] + ocrmac (for experimentation) [2] and was pleasantly surprised by the performance on my i7 6700k hackintosh.

I wouldn't call myself a programmer but I can generally troubleshoot anything if given enough time, but it did cost time.

[1] https://gist.github.com/RhetTbull/1c34fc07c95733642cffcd1ac5...

[2] https://github.com/straussmaximilian/ocrmac

I found a config that worked for me using TST so I haven’t explored other methods for vertical tabs. What eventually convinced you to try Sidebery and what convinced you not to switch back to TST? I’m curious what I’m missing out on!

I don’t use containers all that much (I just auto clear cookies and history when Firefox is relaunched) but I can’t overstate the workflow upgrade that was my switching to TST a few years ago. That combined with a plethora of mouse macros for moving between tabs, changing their position in the tree hierarchy, and other basic functionality really made the experience of using a web browser more enjoyable and interesting. Lots of customization to be had with userChrome.css + TST css theming. Thankfully there’s a lot of prebuilt configs shared on GitHub to help get you started.

They may be referring to stackline https://github.com/AdamWagner/stackline

I've used it in the past, it was very aesthetically pleasing but did not work consistently enough for me to use day to day.

For me, the gap between 3.5 and 4 is massive. If I'm stuck between using 3.5 and doing the work myself, more often than not, I'm choosing to do it myself. Not to imply 3.5 is unusable, its just my bar for minimum fidelity is closer to 4 than 3.5 with respect to tasks that I'm comfortable offloading onto an AI.

What are you running goliath-120b on? Is it costly to run all day every day? How long does it take to complete an output? I've thought about building a multi GPU node for local LLMs but I always decide against it on the premise that the tech is so new I figure in the next 3-4 years we'll see specialized hardware combined with efficiency improvements that would make my node obsolete.

The tollbooth? I would imagine so unless until the compute cost comes down and the hardware becomes more accessible/integrated at the consumer level.

If you mean my preference for subscription over ads, that is guaranteed. I'm fine with an ad model for consuming content (like watching YouTube) but never with content generation (like using Photoshop).

Plus, I really like these technologies and want to see them go further and I'm more than happy to pay for my product when the deal is good, which AI costs currently are relative to the hardware cost. Having to pay for these services + having big tech compete with each other for the best cutting edge release = a lot of money, time, and focus in that area to win the consumers on the merits of their products, whether that consumer is an enterprise customer or not.

I don't see this kind of competition in any most other marketplaces for content generation tools, that's partially by virtue of AI being new tech but also because the race for dominating the AI marketplace has only just begun.

I'm not a professional in VFX, but I work in television and do a lot of VFX/3D work on the side. The quality isn't amazing, but it looks like this could be the start of a Midjourney-tier VFX/3D LLM, which would be awesome. For me, this would help bridge the gap between having to use/find premade assets and building what I want.

For context, building from scratch in a 3D pipeline requires you to wear a lot of different hats (modeling, materials, lighting, framing, animating, ect). It costs a lot of time to get to not only learn these hats but also use them together. The individual complexity of those skill sets makes it difficult to experiment and play around, which is how people learn with software.

The shortcut is using premade assets or addons. For instance, being able to use the Source game assets in Source Filmmaker combined with SFM using a familiar game engine makes it easy to build an intuition with the workflow. This makes Source Filmmaker accessible and its why theres so much content out there made with it. So if you have gaps in your skillset or need to save time, you'll buy/use premade assets. This comes at a cost of control, but that's always been the tradeoff between building what you want and building with what you have.

Just like GPT and DALL-E built a bridge between building what you want and building with what you have, a high fidelity GPT for the 3D pipeline would make that world so much more accessible and would bring the kind of attention NLE video editing got in the post-Youtube world. If I could describe in text and/or generate an image of a scene I want and have a GPT create the objects, model them, generate textures, and place them in the scene, I could suddenly just open blender, describe a scene, and just experimenting with shooting in it, as if I was playing in a sandbox FPS game.

I'm not sure if MeshGPT is the ChatGPT of the 3D pipeline, but I do think this is kind of content generation is the conduit for the DALL-E of video that so many people are terrified and/or excited for.

Building on top of any of these platforms provided by trillion dollar companies is a sucker's game.

Until local models reach the fidelity and speed that these megacorps offer, what choice does anyone actually have with respect to AI? I was under the impression that even if you get over the initial cost of hardware to achieve speed, the fidelity of your outputs would still be of a lower overall quality relative to GPT/Claude/Bard(maybe?). I could be 100% wrong though.

I'll take a tollbooth over something passive like ad injection every single day of the week.

I think a lot of the special sauce with the Half Life games is their "Show don't tell" approach to game design.

You load up the game, watch an intro (for games after HL1) and boom you're in the game. Nothing is going to appear on your screen beyond your HUD except for one time tips when acquiring a new weapon and chapter titles when progressing to a new level. Everything happens in real time and everything always happens from the POV of the player. There are no cut scenes, there are no cinematic transitions, and the player's POV is never intercut with cameras/external perspectives.

It makes the games feel grounded and more cinematic, IMO. Its a game design language that really puts the onus on the player to witness the story and game for themselves. When I play a Half-Life or Portal game, I feel like I'm discovering the story as I play, while other games often feel like I'm being presented or shown a story as I play. The execution is more nuanced than that, but the game's simplicity makes it difficult to focus on anything other than the story and world within the game.

When you look at the stories of each Half-Life game, each game tends to follow the same structure with the same kind of plot points. But for me, it never feels repetitive because the experience of witnessing the story is different for each game. Of course Half-Life isn't the only game that does this and this of course isn't the only reason why Half-Life has been enshrined as one of the GOATs.

The underlying content of the game, its aesthetic, the textures and models, sound design and score, voice acting, ect. all are great in Half-Life, but I wouldn't say that Half-Life is the best at any of these things. I also wouldn't say the game's narrative is the best of the best either. Despite this it remains my all time favorite game franchise, possibly/probably because I grew up with it but also because the game does a fantastic job of using that underlying content to give little clues. Everything feels like it has a purpose, whether it is apparent in the moment or not.

But I am just a storyteller that happens to be a barely competent programmer with absolutely no experience in game development, so all of this is just my opinion based on my own experience!