What happens when you take a hero frame from an AI-generated video and push it back into an image edit model?

This weekend I took a still from a recent Seedance 2.5 piece I made —depicting a solo mountaineer in a harsh alpine environment — and ran it through Midjourney's new v8.2 edit model to see how far I could iterate on a single frame without regenerating from scratch.

Tested wardrobe swaps, environment shifts, and style transfers.

Video walkthrough of the process coming in a separate post.

This looks cool

Experimented a bit today with Krea and image generation for a case study I’m putting together around AI EarPods connected to OpenAI.

The focus has been on creating fashion-forward product imagery and art directing a world that feels specific to the identity, rather than just generating nice-looking AI images.

The trickiest part has been product consistency. Especially getting the EarPods to actually sit snug in the ear. If you’ve worked through this process, you probably know the struggle 😅

Simply telling AI to “make it fit more snug or in the ear” doesn’t always work. It loves to reinterpret the product every time.

Still experimenting, but getting closer. If anyone has found a good workflow for keeping products consistent across AI-generated shoots, I’d love to hear it!

TOP!