//=time() ?>
Intellectual properties don't seem to work as well with #StableDiffusion as they do with #Craiyon, so getting a higher fidelity version of Komi-san shaking hands with Xi Jinping (People on MAL were too willing to take every image as being a real screenshot) ain't too easy, sadly.
Usando #craiyon, en cambio, observamos cómo el programa sí que sabe captar aspectos del modelo de Durero. Sobre todo a la hora de experimentar con imágenes no tanto de la estampa como del taco empleado para realizarla.
Aquí van algunas pruebas usando diferentes iteraciones:
Nightcafe, the fancy AI that does high rez images, and limits how many images you can make, is less coherent than Craiyon, which has no limits.
AIに描かせた「スマブラに参戦するデオキシス」
けっこう身体がデオキシス
» Craiyon, formerly DALL-E mini https://t.co/2FZvVuMa11
Starting a thread about AI-generated imagery. My experiments, cool stuff I find, tips & tricks, general thoughts, etc. Right now, I have access to the #MidjourneyAI beta and the free tools like #craiyon, but if I get into the #Dalle2 beta (or any others), I'll include that, too.