I do think the positives of image AI just aren't really there right now. Outside of maybe DLSS.
I agree, you can use an old desktop, laptop, or if you don't have something I had good luck with the local university surplus store.
A new exciting package that can only be obtained through my touchpad drivers would be mildly concerning
Not necessarily disagreeing with you here, but I will point out that Photoshop can and probably has been used to create non-consentual pornography and scams as well
Intuition
Harris was less than 2% behind, and Clinton was ahead in the popular vote. I don't think gender was not a factor, but I also don't think it was a major one.
I tried searching "civetta scintilla" -beamng, google does what I want but I don't think duckduckgo implements the - operator
(the civetta scintilla is a fake car from the game beamng drive, so most results have both in them)
I usually end up switching to google when I can't find things for this reason
although google does insert a "videos" section at the top of the list which does completely ignore the - operator
lol I found someone on facebook marketplace with screenshots from the game trying to sell it as a real car https://www.facebook.com/marketplace/item/1083690136569890/
Only half of a toaster (in the US at least)
It's a nice easy unit to compare against because all of the ones I've seen draw basically exactly 1000 watts
It's also less than double what my desktop draws (11700k, rtx 3060) and (those aren't particularly demanding components, and I only get 16 threads) (that's basically exactly 6x more power draw per core, although the cores themselves perform differently ofc)
It is slightly silly to have that many cores tho. I guess the main reason to not just use a gpu would be because pcie doesn't have enough bandwidth, or if you need a ton of RAM? For a pure compute application I don't think there are many cases where a GPU isn't the obvious choice when you're going to have almost 400 threads anyways. An A100 has half the tdp and there's no way the epyc chip can even come close in performance (even if you assume the cpu can use avx512 while the gpu can't use it's tensor cores, it having about a third of the memory bandwidth isn't exactly encouraging about the level of peak compute they're expecting)
You could just hop back through time doing this, and no one could ever catch you at the end because you hadn't even committed those crimes yet
Tech bro is the specific subset tho
As much as I don't like the Chinese government for all the censorship and etc, they do really know what they're doing in terms of managing the economy and the infrastructure it relies on, as far as I can tell. Including their education system, it's fairly shocking how many of the top engineering and CS colleges are in China.
Italian style chickpea or lentil soup is always great, you can make them almost the same too (just swap out the legumes)
The Chinese market already has better EVs than Tesla for cheaper afaik. Teslas don't sell well there anymore.
Blahaj.zone is an instance aimed at queer people, but it doesn't have to prevent non-queer people from participating. I would imagine an instance aimed at women to be similar.
If you've seen that blue-ness in the sky you've seen air.
Just scoot forwards a bit?
That's interesting, I can see both circles but my hand does always go to my right eye unless I force myself to look at the other. That makes sense I guess because my right eye is more focused at a distance. I wonder if that switches though when I look at something close up for a while, because my left eye is more focused then.
I'm kind of weird I think in that I can focus further away than something I'm looking at, but I can't focus closer than something I'm looking at. At the same time though I can cross my eyes to converge closer, but can't push them apart to converge further away.
Barely anyone uses state flags, unless its hung right next to a national flag. No-one really cares about them honestly, at least where I'm from. Conservatives are more likely to use the national flag, but I don't think I've ever seen anyone with just the state flag.
I know some people are suspicious of fedora specifically because of its ties with IBM.
double slit rule
What New York might look like with a double slit as your camera aperture.
Original picture:
Double slit kernel:
What an eye might see, for comparison:
Here's a different, big double slit:
rust rule
in the new minecraft april fools snapshot
it makes your gear degrade quicker with damage
deepseek r1 with the prompt "why do I"
With the smaller 14b model (q4_k_m), just letting it complete the text starting with "why do I"
edit: bonus, completely nonsensical (?) starting with "I don't" (what could possibly be causing it to say this?)
"secure screenshot" idea
I was thinking about how hard it is to accurately determine whether a screenshot posted online is real or not. I'm thinking there could be an option in the browser to take a "secure screenshot", which would tag the screenshot with the date, url, and whether the page was modified on your computer. It could then hash both the tag and the image data and automatically upload this hash to some secure server somehow. There would need to be a way to guarantee that only the browser could do this, or at least some way to tell exactly what the source was. I'm not much of a cryptography person, but I would be surprised if it isn't possible to do this. Then, you could check if the screenshot you see is legitimate by seeing if it's hash exists in the list of real hashes.
Does anyone else find it weird that the bottom of a graphics card is always the good looking side?
reference image if you have no idea what I'm talking about:
I know this is a minor nitpick, but it's something that annoys me.
I got this graphics card mostly because it was the best deal on Amazon at the time (gpu shortage), and I also thought it looked decent from the images they had. However, when I actually installed it, all I see is the relatively unattractive looking black metal backplate with some white text. The other side is always the side shown in the promotional images too - not a single one of the pictures in the Amazon listing even shows the side that you'll be seeing 99.9% of the time. Do they think everyone hangs their PCs above them from the ceiling, or has open-air testbenches? Why do they never even bother with the other side? I know they want the fans on the bottom so the cooling is better, but the ai
colors rule
my reasoning: the actual colors we can see -> the wavelengths that we can extrapolate to -> basically extrapolated wavelengths plus an 'unpure-ness' factor -> not even real wavelengths (ok well king blue and maybe lavender if I'm being generous could be)
... rule
Just 3% less votes than Jill Stein, and he dropped out 3 months ago
the 'look, we have global illumination!' box
I've often seen this sort of thing in videos advertising GI in minecraft shaders, and tried it out in blender.
RealSense depth camera in ticket machines
This is at JFK, does anyone know what they are used for? There wasn’t an obvious time when it was taking a picture.
Pixart-Σ just feels so much nicer to use
Prompt: A cyberpunk scifi painting of a floating city in the air above the sea
It uses a new, fancier, 18GB text encoder (t5) to follow the prompt much more closely. It isn't perfect, but its much better than SDXL in my opinion. It does seem to be a bit worse at photorealistic subjects and has a tendency to create 1-pixel vertical lines.
Some other images:
impressionist, a woman sits in the middle of a crowded cyberpunk street, people bustling around, orange and blue glowing signs, warm atmosphere
a bright cinematic photo of a solarpunk city at midday, skyscrapers, steel, glass, vines and fields of vivid tropical plants
SDXL Turbo generated all of these images in 10 seconds
I get around 1 image every quarter of a second on my 3060. The quality isn't up to par with regular SDXL (not even close) but it follows prompts well and is extremely fast. Here are some of the best images in this batch:
Prompt: "impressionist oil painting, watercolor, a crying old southern man eats cheese at sunset in front of a futuristic dystopian cyberpunk city"
Add 3D objects to a AI generated image with Blender
Material:
3D model: Original image:Peaceful drive (definitely nothing bad is going to happen here)
Simple Godot graphics demo
Runs at around 300 FPS on my RTX 3060, WITH realtime global illumination enabled (sdfgi). Sadly still not fast enough for the scene to be playable in VR on my hardware - at least not at native resolution.
Some additional screenshots:
I'm pretty new to Godot, and the tower is a model I made for a different project. This is an updated version of the scene I posted here a couple of months ago. I'm posting this to show that you can achieve some pretty nice graphical fidelity in Godot - and you don't need to be a professional artist or have tons of Godot expertise to do it.
Barad-Dûr and the Eye of Sauron - EEVEE
It uses baked ambient occlusion and has 20,070 triangles, 10,576 vertices, 8,752 faces, and is 45mb in total, including textures. It uses 2 4k sprite sheets for the eye to run in Unity without using the Video Player component.
Just started Godot, made a simple graphics test
I'm using SDFGI and TAA for this scene, the one strange thing I've noticed is that rough metallic surfaces like that sphere can look odd, most of it is rough but there are some sharp lines, not realistic behavior for a surface like that. You can see the reflection of the sun is blurred, most of the scene is blurred, but you still get some sharp lines for some reason. This only seems to happen with roughness values between 0.1 and 0.2, a roughness value of 0 looks fine. I'm getting 210-220 FPS in this scene on a 3060 at 1080p.