3 ms·
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.
by rms 19y ago
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.
- jey 19y agoMaybe 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?
- bct 19y agoIt 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).
- ivankirigin 19y agoGradients 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.