Edit Photos With a Single Phrase *Replicate + OpenAI*

Just made a demo of interior editor.

Choose a room, write an instruction, generate what you need. Simple.

OpenAI’s gpt-4o-mini for text classification (multi language + prompt boost) & Replicate Model with Image to Image tuned on interiors.

Demo:
12dUntitled (2)

btw you can notice buttons are aligned with choice component within a container, normally they have different height.

I achieved this by using rich text with <div style="width: 100%; height: 2px;"></div>

It’s a good use case. I wish Replicate API would be able to return things faster.

what do you mean by faster? platform inference or how it works with glide?

How it works with Glide. At the moment it still does take a good while to return an image, but I understand it’s more complex to run with image/video models.