I suppose, hypothetically, a person could intend to make a product that is as generic and indistinguishable from other, similar products as possible. Though, I do think there’s an interesting philosophical debate to be had over whether that product would constitute art, especially if they didn’t...
Again, DLSS5 doesn’t have intent, and the artist can’t predict the results. It’s like hitting “randomize” on your character creator and saying “that’s how I meant for the character to be.” No you didn’t. You might like how the character ended up, but you didn’t make any actual choices. It can’t...
It makes the color tone of all of the images cooler. That’s the AI slop effect; regression to the mean. On average, the model’s training data must have a slight bias towards cool lighting.
Theoretically they want the result to look as close as possible to the real human being depicted, and this tech seems to actually make it look less like him. Photorealism doesn’t mean much when the photo isn’t of the subject you meant to photograph. In fact, it isn’t of any subject. It’s of some...
Ironic, because the reason MEA’s facial animations looked like crap on release is that a lot of them were done by an automated system and not quality-checked by humans due to the rushed release. They have mostly been fixed now. But the fundamental problem with them is the same as the one with...
Then that developer doesn’t understand how DLSS5 works. It doesn’t have intent. An artist might theoretically like the results better than their own work. But that’s not the same as it being consistent with their artistic intent.
But at least the creators of those mods made actual artistic decisions. They didn’t like the way the characters looked, so they specifically chose to alter their textures in specific, intentional ways. People may agree or disagree with those mod creators’ aesthetic preferences and choose to...
Because they aren’t choices. No one made any choices here, they just turned on the DLSS5 and an AI made it look more like what it expects the image should look like based on its training data (which is basically every image of a blonde woman on a city street at night on the internet)...
It’s entirely possible! I just think there is a strong tendency to treat “the algorithm” as some kind of vengeful god that grants viewership when properly appeased and takes it away when angered. Nobody really, thoroughly understands how the algorithms work, and while changes in their priorities...
D&D YouTube has been struggling lately. A lot of creators blame it on changes to the algorithms, but personally I suspect that between the OGLpocalypse and the fracturing of the 5e fanbase in it’s aftermath, there’s just less demand for D&D YouTube content than there was between 2020 and 2024.
I have mixed feelings about the quantity of character abilities/features/powers/spells. On one hand, yeah, characters from 3e onward just have so many, it’s often hard for players to remember their own capabilities, which is unnecessary. On the other hand, I as a player would always rather get a...