🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

GPT4o image Gen V1

GPT4o image Gen V1

The results for V1 are impressive, but there is still a lot of work to be done for V2. Hopefully, V8 and V2 will both arrive before the end of the year.

If Midjourney V1 had arrived early this year, it would have been a game changer, but everything after Veo3 seems worthless unless you have the same quality.

MDJ V1

Testing MDJ V1 extend feature as I said, this model before Veo3 would have been a game changer, but not today.

MDJ V1

This is the first test with MDJ V1, a 20-second clip, and the result is the same problem as Sora: Diffusion Transformer architecture, not world model. They do not understand physics and have no memory. This is just a waste of time and resources.

MDJ V1

I'm trying the old prompt with Hailuo AI (MiniMax)'s new video model. Overall, there are significant improvements, but there is so much that needs to be done in the coming months.

Veo3 Scene Setup 8 s 4 K live-action cinematic realism, interior of a cluttered 1890s lab lit by crackling coils. Ozone haze. (No subtitles.) Subject Nikola Tesla, 40s, tall, neat moustache, dark waistcoat. Beat Map 0-2 s Camera cranes over humming coils; Tesla lifts a

Veo3 Scene Setup 8 s 4K photoreal desk vlog, candlelit study lined with brass gears and punch cards. (No subtitles.) Subject Ada Lovelace, mid-20s, silk gown, ink-smudged gloves, analytical diagrams behind. Beat Map 0-2 s She slides a stack of coded cards into view. 2-6 s

Really looking forward to OpenAI image Gen V2