Tristさんのイラストまとめ

#AiArt latent space explorations like 🚀

Here's
{allegorical, hyperrealistic} {ascii art, dadaism} painting of the human condition

#midjourneyv5 #aiartcommunity #midjourney

Also fun:

using "/describe" to get a prompt & then running it with "--niji 5"

#midjourney #nijijourneyv5

Left: --niji 5 --stylize 100 (default)
Right: --niji 5 --stylize 500

#nijijourneyv5 #midjourney

Things get really interesting when we tweak the "stylize" values on the new #nijijourneyv5.

Left: --niji 5 --stylize 0
Right: --niji 5 --stylize 50

#Midjourney #niji

🚨 #Midjourney is testing a new version of #Niji, a model specifically tuned on anime aesthetics.

Early results are fantastic! It's finally V5 time for anime! 🤯

☝️ To start exploring, add "--niji 5" to your prompts.

🤩 Additionally, Niji 5 now supports "--stylize"… https://t.co/9gXm1d7Qbv

You can now extract a prompt from ANY picture by using "/describe" + uploading an image.

#Midjourney then gives you 4 variations of a prompt it thinks represents the uploaded image.

Sometimes pretty close to the original ...
sometimes not so much. But hey: free prompt! 😅

told my 10yo daughter she could try #gpt4 if she'd use it to code ... minutes later she came back with an svg version of monkey island 😂❤️ dad+90s gamer heart, both proud

#midjourney5 with strong tendency to photorealism 📷

same prompt, left: V4, right: V5

Some #Midjourney V5 findings 🔥
Finally! Full-blown cinematic aspect ratios!

2.4:1 widescreen cinema standard ✅
2.76:1 ultra panavision 70 ✅

you can go up to 14:1 (!) but: letterboxing seems to set in roughly at ultra widescreen (3.6:1)

#midjourney5 🧵

I generated the images for the game with #midjourney
Prompted for basic 1:1 images, using the tried and tested combination of "white background" and "--no background" to prep transparent PNGs