tal @ tal @lemmy.today

Posts

159

Comments

6,664

Joined

2 yr. ago

A couple of my favorites from that thread:

@FlyingSquid@lemmy.world

@ThunderWhiskers@lemmy.world

@YodaDaCoda@sh.itjust.works

@DredUnicorn@lemmy.world

@Thelsim@sh.itjust.works

@anonymoose@lemmy.ca

@itslilith@lemmy.blahaj.zone

@starcat@lemmy.world

@Tlaloc_Temporal@lemmy.ca

@Usernameblankface@lemmy.world (also OP here, also the person who brought the idea up in another thread)

@WandFliesenWodka@lemmy.world

Link to the original version of this thread to which OP is referring, which was quite a hit some time back:

https://lemmy.world/post/7371120

Up to par with your usual work, merde!

@mutilated_sphincter@lemmy.world:

Operating on the assumption that even merde may not want to touch this one, using realmixXL_v15:

Sure! I use ComfyUI locally. Replacing the underscore with a space under the assumption that this is the intended wording and using flux1-dev-fp8:

With stoiqNewrealityFLUXSD35_f1DAlphaTwo:

With realmixXL_v15:

Hah, that is a good anecdote. Thanks for sharing!

I'm curious whether your username is going to result in something akin to the Toynbee tiles.

Aw, I guess not. flux1-dev-fp8.

It was both evil and surprisingly cute.

EDIT: With flux1-dev-fp8:

That's actually a lot more pleasant than I'd have expected!

EDIT: For flux1-dev-fp8:

This sounds like a job for @merde@sh.itjust.works, our resident expert in that sort of thing.

Just "tal" was disappointing, something that looks kinda like a bowl of soup.

Taking OP's advice, and doing "an avatar for tal" on stoiqNewrealityFLUXSD35_f1DAlphaTwo yields:

That model's trained on people, though, so it's going to be biased towards people.

"tal" on flux1-dev-fp8:

considers

Fair enough; I can do one for you. I have a local ComfyUI setup, so prompt data won't go anywhere. Heck, even if I were generating it on a service, it'd be useless for letting said service gain information about you, as I don't know who or where you are.

For "over_clox", using stoiqNewrealityFLUXSD35_f1DAlphaTwo:

The State Department is preparing to order the departure of all nonessential personnel from the U.S. Embassy in Baghdad due to the potential for regional unrest, two U.S. officials said Wednesday.

The Baghdad embassy has already been on limited staffing, and the order will not affect a large number of personnel, but the department also is authorizing the departure of nonessential personnel and family members from Bahrain and Kuwait.

Note that we don't have diplomatic relations with Iran, so in the event of a military conflict with Iran, there wouldn't be anyone to evacuate there, just in nearby countries.

Smartphones are fragile without a case. They should have one, and maybe manufacturers should make that clearer, but a world where removable cases didn't exist would just mean that the case you get is the one that the manufacturer chooses for you and permanently attaches to the smartphone. Less options for you.

Just get a case.

I am also more than willing to carry a slightly thicker device if it means greater durability and easier repairability.

Me too. It's why I have a case.

And I am certain many others would gladly trade their bulky, overpriced cases and bumpers for a sturdier device that inherently provides the protection we now have to purchase separately.

If you want a built-in case, you can get them. There is a whole collection of "ruggedized" smartphones from various manufacturers in China that are large, usually have a hefty battery, and have shielding built into the device.

Look at Doogee for one such manufacturer.

https://www.doogee.com/

Oukitel for another:

https://oukitel.com/

Ulefone for another:

https://www.ulefone.com/

Personally, I think that the built-in case isn't very interesting relative to a removable case, but the large battery might be, depending upon your needs.

EDIT: A number of manufacturers will even make official cases for their phones, if you can tolerate a removable case and just want something endorsed by the manufacturer.

Apple, for example:

https://www.apple.com/shop/iphone/accessories/cases-protection

Or Google:

https://store.google.com/product/pixel_8_phone_case?hl=en-US

It'd theoretically be possible to run a straight GNU/Linux tablet or laptop

"GNU/Linux" is the full way to say what sometimes gets shortened to "Linux" --- a family of operating systems based on the Linux kernel and a lot of software from the GNU project. This explicitly distinguishes it from Android, which also used the Linux kernel.

The former is not, in 2025, typically used to run smartphones. The latter is the most-common smartphone operating system in the world. If you buy a smartphone that isn't an Apple smartphone, it almost certainly runs Android.

with a 5G cell modem for data

5G is the current generation of cell phone radio protocols. Communicating directly via voice over this protocol is not something that I believe is available to GNU/Linux in 2025. However, it can send non-voice data.

, use SIP service

SIP is a protocol for running voice over a data connection to the Internet. If you have an Internet connection, you can use SIP. There are companies, SIP service providers, which will, for a fee, provide a phone number at which one may be called or call others from a computer that can make use of SIP.

and a GNU/Linux dialer,

A dialer is the piece of software that on a smartphone, a user would probably call something like "the phone app".

and then run Waydroid for any specific Android apps that one has to run.

Waydroid is a piece of software to run Android apps on a GNU/Linux system.

Idle power usage is gonna be a lot higher than on a phone, though.

Phone hardware and software has had a lot of work put into optimizing it for very low power usage. A larger device, like a laptop or tablet, will probably also have a larger battery, but it will consume more power as well.

And a lot of Android apps are made with a touch interface

Smartphones, due to physical space constraints in one's pocket, typically have an entire side be a touchscreen. They do not have a keyboard. In general, software optimized for this works somewhat differently from software optimized for use with a keyboard and mouse.

Most GNU/Linux software is written with the intent that it be used on a system that almost certainly has a mouse and keyboard available. Most Android software is written with the intent that it be used on a system with a touchscreen available.

This means that even if one can run GNU/Linux software on a phone, much of the (large) collection of GNU/Linux software available will not be designed with an interface ideal for use on a phone.

and small screen in mind and are aware of things in a cell environment, like "only update X when on WiFi". Not really common for GNU/Linux software to do that.

Smartphones have two widely-used mechanisms of accessing the Internet --- connecting to the often slower cell network, or to a much-shorter range, but faster, WiFi network. Many people connect their smartphone to a WiFi network at some times and a cell network at others. Because this is so common, a lot of Android software has behavior designed to support this and act more-appropriately, like having an option to only transfer lots of data when on a WiFi netwprk. This is not the case for most GNU/Linux software.

I mean, you can restrict stuff with only a few customers by controlling trade tightly.

But for common parts --- to use a past example that I saw in a report, voltage regulators --- they're everywhere. You can't do much about that unless you're willing to economically partition the world. Even then, other countries could make them.

The last time I used a commercial VPS, I'm pretty sure it used VNC to provide console access.

The VNC software I linked to above appears to support TLS. If TLS isn't sufficient transport security, then most Internet-using software is going to be in trouble.

I'm not sure what you mean by subjective.

I haven't looked at the VNC protocol for a while, but I don't think that it imposes any terrible inefficiencies. A couple of decades back, I needed to implement something quick-and-dirty similar to VNC, and went with rendering window contents and handling dragging of windows locally, which I don't believe that VNC can do (or didn't then) but IIRC VNC has a tile cache, which, if intelligently used, should avoid most traffic. Dunno if it can deal well with efficiently rendering visual effects.

Right. What I'm saying is that the benefit that VRR provides falls off as monitor refresh rate increases. From your link:

If a game on console doesn't deliver new frame on time, two things can happen.

Console can wait for a new TV frame, delaying display time about 16.7 ms (VSYNC). Which leads to an effect called stuttering and uneven frame pacing...

If you have a 60 Hz display, the maximum amount of time that software can wait until a rendered frame goes to a static refresh rate screen is 1/60th of a second.

But if you have a 240 Hz display, the maximum amount of time that software can wait until a rendered frame is sent to a static refresh rate screen is 1/240th of a second.

OLED monitors have no meaningful framerate physical constraints from the LED elements on refresh rate; that traditionally comes from the LCD elements (well, I mean, you could have higher rates, but the LCD elements can only respond so quickly). If the controller and the display protocol can handle it, an OLED monitor can basically display at whatever rate you want. So OLED monitors out there tend to support pretty good refresh rates.

Looking at Amazon, my first page of OLED monitor results has all capable of 240Hz or 480Hz, except for one at 140 Hz.

That doesn't mean that there is zero latency, but it's getting pretty small.

Doesn't mean that there isn't value to VRR, just that it declines as the refresh rate rises.

Reason I bring it up is because I'd been looking at OLED monitors recently myself, and the VRR brightness issues with current OLED display controllers was one of the main concerns that I had (well, that and burn-in potential) and I'd decided that if I were going to get an OLED monitor before the display controller situation changes WRT VRR, I'd just run at a high static refresh rate.

Setting a high refresh rate is somewhat of a given, but won’t negate anything which VRR helps with - screen tearing.

I mean, I'd just turn on vsync; that's what it's for. VRR is to let you push out a frame at the instant that it finishes rendering. The benefit of that declines as the monitor refresh rate rises, since there's less delay until the next frame goes to the monitor.

If you’re always playing with VSync on and getting constant frame rates, that’s not an issue

looks blank

Constant framerates? You're saying that you get tearing with vsync on if whatever program you're using can't handle rendering at whatever the monitor's refresh rate is? I mean, it shouldn't.

Running a static refresh rate with vsync will add a tiny bit of latency until the image shows up on the screen relative to VRR, but that's a function of the refresh rate; that falls off as the refresh rate rises.

VNC is dead.

How so?

There are a number of software packages in Debian that implement VNC. To grab one random example, the last commit to their git repo was last month.