I once heard that Ireland experimented with a government allowance, and tax exemption(?), for fiction writers (and maybe other artists too), to preserve and encourage interest in Irish culture (and cover their writers’ living expenses). Considering the current and predictable advances in AI, other countries may need to follow this example.
2 Likes
A Dall-E3 example from Reddit
Here’s an example of Dall-E3 plus GPT4
3 Likes
Playing around with a basic elf character prompt I found on the Civit-ai website (for the AlbedoBase checkpoint). Guess I made the right call to unsubscribe from Daz3D. However, I can’t find a way to remove the halo / reflection effect from the characters’ eyes, even with A1111’s in-painting feature.
…So close to caveman level prompts conjuring photo-realistic images…
2 Likes
Setting up an image prompt for some Freyja portraits in SDXL, and pressed the generate button about six steps too early. Received an almost ideal passport-size picture (512x512).
A successful fail, I guess.
2 Likes
Out of the 3 I edited today, this one came out basically the same as I have here, except this has been cartoonized.
This was also an option:
2 Likes
Dall-E3 is now available for public use.
Reviving this thread with AI news…
Midjourney’s new Upscaler
Free Dall-E3 updated
3 Likes
Funnily enough, now that I have access to the upscaler, I no longer need it. 
I still use Midjourney but for a different purpose.
I’ve been practicing my drawing skills by drawing the characters I’ve created. Believe it or not, but practice actually helps.
I’m feeling pretty confident about my skills now, though I still need Midjourney to create the concepts for me first. I suck at composition. But with a reference image, cool things are possible.
For example, Midjourney vs my marker drawing.
This is Ian, btw
2 Likes
Those art styles, and the character design, reminded me of a lore video series about Dune I watched recently.
Now I’m thinking character background / history (or story lore) images could be an interesting AI art project.
2 Likes
I basically do the same thing now, but all of the art community’s critics never seem to bring this fact up.
1 Like
I think a lot of people do that, but because of the critics, they don’t admit to it.
2 Likes
I think it’s perfectly legit. You see what I edit on the fails? It’s not what I can do from scratch. You’re doing 1 better.
1 Like
Thread CPR.
Stable Video Diffusion now has a local install (personal computer) option.
Recent SD experiment: Freyja searching the villa for her companions...
…Dov is late for story time in the library (check the villa’s gym).
If I use “Tiled VAE” in SD A1111 with default settings I can conjurer 1024x1024 (and x2.5 upscale) images without my PC expiring. However, the image generation times are measured in eras.
^My computer’s response when I attempt a x3 upscale.
3 Likes
Fearmongering of Ai has not yet stopped. Soon it will be the perfect cover for me to destroy this pitiable world and all the pathetic creatures that inhabit it.
Sorry did I say too much?
Here have this video to forget about my burning resentment.
I’ve seen people have mixed results with ai upscaling, they can be crunchy or smudged
1 Like
That audio abomination is yet another reason why SkyNet AGI can’t come fast enough. Next year, maybe.
Oh, I achieved some success with x3 upscales of 1024x1024 images last night. About fifteen minutes per image. Only problems were double-exposure eyes and no fine / unpixelated details in the materials. My DSLR camera is still safe…for now…Different story when we get 4K / 48 Megapixel resolution in Stable Diffusion XL.
You can tell the image is in focus when you can count the pores on the model’s skin ~ Quote about portrait photography.
1 Like