Playing with content aware fill
Posted on September 15th, 2010 by Alex in PhotoFirstly – I don’t have the time to do an update to the How to section at the moment, please hold onto your horses for the next instalment. Or practice your tetris skills…
Instead a quick ramble and play with photoshop
One of my recent extravagances has been upgrading from using Pixelmator to purchasing CS5 (albeit the cheapest version I could).
I was in town a few days ago and in a bit of a rush, so didn’t set myself up properly for taking a panoramic and instead just snapped away fairly arbitrarily. Using the utterly amazing photomerge feature of photoshop (which has been in there for a few versions now) it managed to stitch them together pretty convincingly:

It’s done such a good job I decided to see how well content aware fill would do to try and make the image more… normal:

Not as well as I’d hoped (though what I was expecting I’m not sure). So I decided to try and give it a bit more of a chance, and segment the work load down into chunks:

The thing that surprised me here was how similar the result was – evidently I’ve got a lot to learn about how the fill system works. The only bit that really looks anything close to “real” is the top right corner (the red outline on the second picture), but when zoomed in that falls apart as it’s all wobbly.
Ultimately I’m asking an amazing amount from photoshop, but it was an interesting experiment. I’ll remember to take more photos next time!

