Why should designers focus on object-oriented editing instead of canvas-based tools?

Photoshop has been around for 30-plus years, and every tool and menu in it is oriented around manipulating images at the pixel level. You draw pixels, box pixels, circle pixels. Why? Because the application doesn't really know what it's looking at. There's no semantic understanding of the subject matter baked in.

That creates a translation problem. When you edit a photo, you think in terms of things: I want her hair to flow to the left a little. So why do I have to translate that into a set of tools, and then into pixels? In an object-centric model, every object is identified and you primarily edit by changing the subject. Move the cat here. Make it a sunny day. Make the cat's eyes green.

This also matches how we deal with the physical world. If I want to move my phone, I grab it and put it over there. I don't draw a little box around it, grab a little hand icon, and drag it to the left. Object-centric editing gets much closer to that natural interaction.

When I first wrote about this shift in 2023, it was mostly early demos and research projects. Now it's real. Reve (a company we kicked off a couple years ago) has a model that thinks in layouts instead of prose: a structured, hierarchical description where every element has a location, size, description, and attributes like color. Think of it this way: layout is to an image what HTML is to a webpage.

That matters for control. Most image models use text internally, and loose input means loose control. Tweak a prompt and watch the whole image change. With addressable elements, you can change the color of a car without regenerating the whole image, and shadows, reflections, and perspective stay coherent because the model understands how the pieces relate.

Of course, pixel tools don't disappear. Both pixel and object manipulation are likely needed for a lot of image editing tasks. The real opportunity is in flipping the hierarchy: object manipulation as the high-order bit, pixel editing as a secondary level of refinement.

So if you're designing creative tools, start from what people actually want to change (the things in the picture) and push the pixels behind the scenes.