GPT Image 2.5 came out today.
I find it does best working with reference images, rather than creating content from scratch.
I gave it ONE image. It gave me back an entire professional photo shoot.
Photo 1 → Midjourney (the original)
Photos 2 & 3 → GPT Image 2.5
“Midjourney v 8 is disappointing”
Respectfully you are wrong lol.
If you want hyper realistic outputs nano banana does it perfectly.
Midjourney truly has a level of artistry that is unique from any other model.
Same exact prompt: top is v 8, bottom is v7
@okandship This is all 1 prompt lol it’s so long because I haven’t used mid journey since v5 & this is how I used to do it! It seems things have changed since then. 😂
@okandship never enough, star flare as punctuation on holiness, voidcore goddess, black void as the only appropriate throne
holographic saint dissolving into static, velvet void angelcore, analog noise as divine texture, chro
@okandship black, light refusing to cohere into anything less than beautiful, holographic foil saint, rave angel without the rave just the transcendence, noise pop sacred, analog warmth meeting digital dissolution, the divine feminine rendered in every color simultaneously because one was
@okandship image trying to exist against entropy and winning, static as reverence not failure, phosphor dot goddess, television dream sacred, broadcast from another dimension, stardust rendered in RGB noise, celestial signal bleed, disco sacred feminine, prismatic devotion against absolute