Imagetostl

https://imagetostl.com/

Meshroom

But you need a ton of photos.

The technique is called photogrammetry

It depends a lot on what you have in mind?
Something like a parametric mechanical gizmo will be very different than a sculpture thingy.

Parametric CAD:
I had to create a part that laid flat over decorative moulding. I took one end, traced its cross-section on paper, added an “L” traced with a square as a reference to compensate skew and to have an easy known dimension reference, then took a picture of that.
Opened in inkscape and traced path from bitmap, export to dxf, import into 3d cad software and use as sketch in the rest of the part.

Sculpting CAD:
Or maybe you’re talking about something like a mini and looking to do photogrammetry instead.
Haven’t tried that, but at least you know what it’s called

Like others have said, if you’re just trying to make a raised version of a simple flat design, that’s easy. Ideally convert it to a vector first, and then import it into just about any CAD software. Even TinkerCAD can do that.

If it’s more of a real 3d object, i.e. varies across a few dimensions, first consider how complicated it is. As long as there aren’t too many organic curves, the best route might be to take careful measurements and recreate it from scratch in CAD.

Finally, if none of those options are possible, it’s time for photogrammetry. Mix up a simple solution of isopropyl and zinc oxide (or anything talc-like i guess, I dunno), and use a quality sprayer bottle to speckle your object with it. Try to illuminate the object well from only a single source. If you can, pick up some cheap cross polarizing filters, use one on the light source and one on the camera. Take shots at the same distance from as many angles as you can, covering the whole object, ideally with a matte background. As mentioned, Meshroom is pretty decent at turning such a photoset into a decent object, but it does rely on you having a reasonably powerful PC. Alternatively, OpenScan offers a free cloud based option. (They’re also just very cool people who do lotttsss of other stuff, my makerspace just built one of their “Mini” rigs. Works great btw!!)

What exactly are you asking?

It’s pretty easy, for example, to turn an image into a grayscale .bmp and convert that to an elevation map that roughly creates an embossed/engraved surface.

It’s not easy to take photos of objects and turn them into 3d models of that object. It’s pretty much impossible from a single photo, as well. It’s possible, but usually fairly complicated- it’s called photogrammetry and there’s tools and plugins for blender that can do it; but it’s very hit or miss.

Not easy for complex pictures, but if you have a simple png with a transparent background you can import it into blender or orcaslicer and then add to the z axis.

I used vecto3d, but it was just to visualize something quickly so I have no idea how the STL export is.

You can try one of these AI apps. https://makeit3d.app/explore