It’s been only some weeks since Midjourney launched its newest and most superior AI picture mannequin, V6. My preliminary evaluation of V6 was largely constructive, noting the considerably improved picture high quality and the addition of a brand new textual content function.
Just when many people thought that they had reached a peak, Midjourney stunned us once more.
A couple of days later, Midjourney rolled out an replace to V6 with improved picture high quality, improved textual content rendering, and extra.
In this text, let’s discover the brand new updates to Midjourney V6.
What’s new in V6?
- Upscales at the moment are twice as quick, using half the GPU sources.
- Notable enhancements in aesthetics, coherence, immediate adherence, and total picture high quality.
- Enhanced textual content rendering (bear in mind to surround textual content in “quotations” in your prompts).
- Improved efficiency at excessive
--stylizevalues - Upgraded upscaler for lowered artifacts, addressing points like blur and noise.
- Addition of an AI moderator on the discover web page, guaranteeing safer and cleaner content material.
I shall be evaluating the outcomes generated with the preliminary V6 mannequin and the most recent and up to date one. I will even be evaluating the up to date textual content rendering functionality of the brand new V6.
Let’s get began.
Improved photographs
To gauge the extent of high quality enchancment, I revisited some prompts I had beforehand generated with the older V6 mannequin and reprocessed them within the up to date V6 mannequin.
Prompt: face manufactured from newspaper, in tabloid artwork model — ar 4:5
Honestly, it’s difficult to claim that the brand new mannequin has made strides in picture high quality. Some customers may even favor the output from the older mannequin. What’s your take?
Let’s do one other one.
Prompt: A good looking attractive woman, pale pores and skin and freckles, swimming in clouds, relaxed, meditating in clouds with closed eyes, numb…
