I have tried unsuccessfully to get Content Aware Fill in CS5 to work as advertised in the dozens of tutorials I have watched. I have used many of the techniques, but unless the background is a simple one, CAF seems to drag in stuff from all over (even if I protect it in a mask) the background. Is there some special trick I am not aware of?
CAF is like a clone tool, if you get too close to objects it will think this is part of the image to be considered in Content Aware Fill. As Trevor said you want to isolate the part of the image you want in a seperate layer then use CAF.
I saw a preview of CS6 from a Photoshop 'expert' recently, where he explained that the new Content Aware Patch tool meant we can now sample the edge of an image without the bleed the OP refers to. But we have always been able to do that, provided we drag our sampled pixels parallel to the edge we sampled them from. I know we now have more flexibility, and can drag the sampled area from an edge to the center of the image, but it irked me that he was being simplistic, and filtering reality in order to make his point.
Bu@@er! I'm a Photoshop pedant. Somebody shoot me.
Hmmm.... That won't work because a lot of you guys are worse than me!!!
Did anyone follow this link I hid in a post earlier today? I failed with knobs on. :-(
It works marvelously well, if you prep for its use properly.
It never ceases to amaze me. Just yesterday I took a microphone out from in front of a person standing and speaking, then all I had to worry about was its shadow!
Why don't you post the image you're trying to use the tool on and we can offer some suggestions on how to get the best results.
Not exactly sure what is meant by copying the areas to a new layer, but I tried it...didn't work.
This is the procedure I use which I developed by taking notes from tutorials I have watched:
Select object (the cups and pot at top) and save to new channel. In channels tab, Cntrl-click to load selection. Click on RGB channel. Choose Selection/Modify/Expand and expand selection by about 20 pixels. Edit/Fill with Content Aware. Deselect. Fix as necessary.
Heres the image (see below) I'm working with (I want to remove the cups...in preparation for Puppet Warp procedure which I will use to bend them):
CAF is bringing in the white rectangular objects jutting out on the left side, bricks on the right corner, the sign on the bottom, etc., etc.
Thanks for the help.
Well the sad reality is that sometimes the computer tools just won't work properly. I tried CAF and on the 10 or so sample images I selected, only 1 of them worked even marginally well as advertised, and I still had to fix up some small bits here and there. I even intentionally chose images I thought CAF would work well with, patterned backgrounds or solid colors and such.
I work in Prepress and I've seen it all from client submissions and requests. The manual clone tool is something I believe we all need to have skills with (that and the pen tool for clipping paths). If some other people have good results, that's good for them; but I am still of the opinion "magic" like this never works consistently for everyone. At least not like we're led to believe (hmmm advertising lying to us... what a concept).
In my opinion, you're asking too much of CAF for this image. It's a little complex and there is not enough "background" area for the algorithm to pull from. Probably more worth your time to "suck it up" and clone it yourself.
You can direct Content Aware Fill to only choose from certain pixels using a mask... Russell Brown did a great video on that, if only I could find the link...
I just did a search and came up with this: http://av.adobe.com/russellbrown/CS5ContentAwarePatterns_SM.mov
I believe I've seen this same video on the Adobe TV site, which will be delivered via Flash instead of being a Quicktime movie.
Hello, have you watched this video? http://tv.adobe.com/watch/adobe-evangelists-terry-white/what-to-do-whe n-content-aware-fill-or-content-aware-scale-doesnt-work/
Yes, I have...tried it several times, but I think there’s just too much “stuff” around for CAF to handle. I think I’m going to have to give up on CAF for complex situations like this one. Thanks for your help anyway.
By the way, do you know how to get help on using this Forum? I can’t seem to get the Reply button to appear in my responses. The only way I can reply is by replying to these e-mails, which is OK, I guess. Also, don’t know how to start a new question. When I type in a question in the box, I don’t get switched to a question form if I click on Submit Question.
Thanks again for your help.
There is far too much going on in that image for CAF to have a chance. Do you have other pictures of the same building you could 'borrow' wall texture from? If you don't have other images, Google 'Photoshop stone texture' and select each panek of the wall in turn, and composite that texture to cover the cups. Place an empty layer over the stone, and fill it with wall colour sampled from the wall, and set its blend mode to Color. The untextured light brown colour can be filled with solid color, and noise added to make it look real. Finally, look for a poster similar to the one behind the cups, and composite that onto the wall.
It's a lot more work than CAF, but if you need the shot, then it is completely doable, and should not take long.
Noel Carboni wrote:
There has been a forum problem where the Reply button wasn't available, but it's supposedly fixed now.
There was a news item about a guy with serious forum withdrawl symptoms in Florida!
That doesn't work as advertized. Make your selection save it with a name as a Channel. When Content Aware Scale is now selected, I get a bounding box around the selection. Enable Protect and nothing changes. The bounding box is still there and is the only part of the image that will respond and not even as Content Aware. It just moves around. I've tried it over and over with the tutorial open so I could follow and the only difference I can see is he is on a Mac.
What I want to do is stretch so here is the steps;
Make a new Layer the size of the finished image. Move the image into that. You have a white space ready to be filled. Select the protected area. Save as Channel with Name. Go to ContentAware Scale. A Bounding Box immediately appears around the selection. That is the only adjustable part, not the entire image.
Ok, here's the problem.
White has the selection saved in his folder for demoing the process. I noticed that he never showed making the selection and saving it. If you do that, the selection is still outlined and as such, ContentAware or any other Transform Tool will act on the selection. So, an extra step is needed. After making the selection and saving it as a Channel, you have to delete the Selection (Ctrl-D in PC) then proceed. It then works properly.
So, he left out a step.
At any rate, while it worked technically, visually, no. I would also have to select the reflection and when you are finished, artifacts are left behind, and the long glacier field above the building is elongated to the point that anyone who knows Mt. Rainier from that spot will go: WHUT!!