davidd's A DOLL A DAY 2024

[davidd]

I asked for a “mini Pullip pirate doll” and the result was ridiculously cute, although as a shipmate commented, AI can’t do rigging for sh*t  ... It was interesting but I remain conscious of the complete artistic void of the process.

Ah, but AI generative imaging, even at the hobbyist level, has become extremely sophisticated. Which AI app did you use, StableDiffusion (and which version) or Dall-E, or something else? For your ship rigging, or the extra thumbs, have you employed "negative prompting," "hypernets," and "LORAs"? Have you elored the "models" available on a site called Civitai? Have you worked at creating more specific, detailed image prompts? Have you used a "seed image" as a starting point for the AI? I haven't tried anything more than simple sentence prompts, and that was months ago, when the technology was much less advanced, using a basic "free" online AI image generator. I follow one AI art person (who started out doing doll photography) on Instagram, which is where I'm picking up "the lingo" of AI image generation.

Pretty much the only reason I visit Instagram anymore is to follow @PlusUltraSydney, who has been diving deep in to AI generative art for over a year now, and sharing the learning process along the way.

Here's a recent "nautical" pic (that, admittedly, still has a slightly gimpy hand):

https://www.instagram.com/p/C1w2ZWZr3NV/

There are some nice ships in this one:

https://www.instagram.com/p/CyZ4eeSrZ7i/

The Civitai website -- https://civitai.com/ -- which I have never actually looked at until now, offers a gallery of AI-generated images from members (many of which are not very good, some of which are superb) and a section of "models" -- segments of prompting code that can be used to generate specific looks for various aspects of AI images.

---

6 February - A Doll A Day 2024:

6 February - Ankylosaur Week 2024