Loading video...

Video Failed to Load

Go Home

We have released an updated version of Gen-4 References that significantly improves object consistency and prompt adherence. Available now for all users. (1/3)

69,427 views • 1 year ago •via X (Twitter)

10 Comments

Runway's profile picture
Runway1 year ago

On Dreambench++, a standard image personalization benchmark, the new version of Gen-4 References significantly outperforms the previous version as well as other diffusion-based baselines. (2/3)

Runway's profile picture
Runway1 year ago

Try Gen-4 References for free at (3/3)

Ram's profile picture
Ram1 year ago

Any update on video gen improvements?

stimograph 👾's profile picture
stimograph 👾1 year ago

How does this compare to Omni reference in Midjourney?

Sebastien Jefferies's profile picture
Sebastien Jefferies1 year ago

Curious how much closer we can get to prompt = output perfection excited for this update 🔥

Dizzaract's profile picture
Dizzaract1 year ago

A meaningful leap for creators working with dynamic characters and evolving worlds. Stronger consistency and prompt control are sure to unlock faster iteration, sharper storytelling, and richer visual continuity.

Blend America 🇺🇲's profile picture
Blend America 🇺🇲1 year ago

👏🏻👏🏻👏🏻

TRVTH's profile picture
TRVTH1 year ago

How are the hands doing?

gusta.AI's profile picture
gusta.AI1 year ago

I’ve already tested replacing a watch on the wrist with a real product, and it doesn’t match the actual watch. This is the biggest challenge in generative AI today — object consistency.

Ferdinand Terme's profile picture
Ferdinand Terme1 year ago

Is it available through API? If so, what's the cost per imaege please?

Related Videos