It would be hard to make a startup from the idea. The video everyone linked had enough information for me to make a free version. Open ideas that are features rather than products make bad startups, right?

The question is how much else he could do on his own vs. at adobe.

Are you an expert in image recognition? If not, I think this would be considerably more difficult to implement than it looks.

He might have a patent pending on the idea too, which could make his intellectual property valuable.

Yes, I work in image processing, but there are many of people with the requisite level of knowledge.

You really think you could have duplicated his work from that video? He didn't get into the math and I assume that was the hard part.

Gradients from 2D images are easy. The surprise was that this was the only transform needed to get the information content.

Dynamic programming over a 2D array is a pretty standard optimization -- here used to find the seams.

Extracting a seam is trivial.

The hard part for me would be doing this in JavaScript to make it a nice addition to page layout, and that's only because I haven't learned enough yet.

I suppose he might have a patent. I hate the patent system.

Maybe I'm just being extremely naive, but I think I could implement it based on the description given in the video and some experimentation. The elegance and simplicity of their algorithm as they described it in the video is pretty damn impressive.

My real question is how carefully they had to select the source images for that demo. How well does that algorithm work on any random image?

It seems like it would work a lot better for a picture of a beach than one of my living room. The resized image of the woman and the baby has screwed up anatomy (even after the face is fixed).

hard?

i think the guys developing the YC funded web based photo editor, should have made an offer to him to join them. It would sooo much give a boost to their effort.

Right. An online photo editor is a product. The resize method is a feature that can be easily redone. That's what is hard about making it into a company.

He was a professor; I don't think there was much of a company to acquire.