MJW

Null54's Content Aware Fill may also be helpful. I tried it with Sarkut's example pig picture, and it worked quite well, though some areas would require additional hand touch-ups. My approach was to add a new transparent layer, then paint over the areas I wanted to replace -- in the case of the pig picture, the fence. I then selected the painted region with the Magic Wand, switched layers to the image layer, and applied the Content Aware Fill. Many cases would be simpler, requiring only a selection with the Lasso tool. I did, however, find that replacing the whole fence in one fell swoop appeared to work better than replacing a small section at a time, since it seemed to eliminate the problem of the filled-in section containing parts of the fence from elsewhere in the image.

To copy the image into the lower layer, it may sometimes be advantageous to use Paste From Clipboard in place of cutting and pasting

It depends a lot on the the book cover, but there's a comment in a current thread that provides a link that may be useful. The basic idea is to erase the text and then, in a lower layer, fill in the missing regions with the other parts of the book cover.

Try it now. I fixed the link. Also check the earlier comment in the same thread where I explain how I softened the edge after removing the background.

Perhaps you're referring to my recoloring, done using the HSV Eraser. I explain the steps in this comment.

Great jobs by @lynxster4, @welshblue and @Pixey! I had a very difficult time deciding which entries to vote for. I think welshblue may have ended up competing against himself, with two similar entries. Thanks to Pixey for hosting! BTW, my candle (which I think turned out pretty well -- especially if the image is placed against a dark background) is based on a method stolen adapted from an excellent Photoshop tutorial. I had to make a few modifications to handle some PS features that PDN doesn't have. The color-gradient was produced using pyrochild's very useful Gradient Mapping plugin. Instead of the Outer Glow, I used my Edge Expander plugin, followed by a slight blur, then Kris Vandermotten's Sepia 2. To substitute for the PS masking layers, I first drew the diffuse and specular shading, without regard to the candle-image boundaries, then used BoltBait's Paste Alpha to trim them to the image boundaries. PS opacity values must go from 0 to 100, so that needs to be taken into account when setting alpha. If I were to do it over again, I'd try using the Texture Shader to shade a cylinder, with the candle-color gradient in the clipboard. It would no doubt require some experimentation to get it right. (Also, I should have made the flame oranger; it would have worked better against the white background. That's one of those details I realize too late, after the contest deadline.)

A plugin that could be used in conjunction with this method, to give different variations of the knit pattern, is Offset Alternating Stripes.

I see the problem. If the OptionBased library accepts only enums for the list values, that seems like a rather poor design choice, since it doesn't easily allow for any variable-length lists. There's no option to simply return the integer index of the selection? If the length of the list can be specified independently of the number of enum entries, then the problem can be solved (simply, if not elegantly) by using an enum with the maximum number of entries. I expect it can't, though.

Perhaps I'm missing something, having never used the OptionBased library, but that doesn't really seem like much of a limitation. Can't the integer returned by the control be used to index into an array of fonts? Returning anything but an index from a list control never struck me as that great of an advantage. It may save a few steps, but they're pretty simple steps.

Try renaming the file from .pdn to .png. You may have saved a PNG-formatted file with a PDN extension. That may recover the PNG version. If it does, recovering the original PDN version would require the type of program toe_head2001suggests, since the actual file was overwritten.

Congratulations to @Pixey, @welshblue, and @lynxster4! All the entries are excellent, and effectively capture the look and color that make me think peppers are the most attractive vegetable. The subtle shading on the two top entries is very impressive. Thanks to Pixey for hosting. I love the new OOTF banner, showing past contest entries (and not just because some of them are mine).

Also, in the tool settings for the Color Picker you can specify that it will switch back to the previous tool after you pick a color. So with that option selected, you can be using the brush, hit "K" to pick a color, and it will automatically switch back to the brush after the pick. The setting can be specified in the Options window, so it's the default.

Thanks for your interesting responses, @welshblue, @Pixey, and @lynxster4. In case anyone is curious, my plan was to make a height map of a top view of the pepper, then use the View Skewer to rotate it to about a 45° view to form the top and, in a similar manner, the bottom. The problem was connecting the two ends to form the sides. I tried a number of ideas, for example using Trail, but nothing panned out. I still believe the method could have been made to work, given enough time and experimentation. (Of course, I would have still had to figure out how to make the off-angle stem, but I think I could have come up with something.) I recently came up with some promising ideas regarding the View Skewer, which I hope to someday incorporate into a substantially improved version. Right now, it's a useful plugin that's very difficult and frustrating to use.

Is the background a single color that contrasts with the foreground (or at least the edges of the foreground), or is it complex? If it's a single contrasting color, the process will probably be fairly simple; if the background is complex, the process to remove it will probably also be complex. The degree of difficulty almost always depends on how much difference there is between the background and foreground near the edge where they meet. You're much more likely to get a useful answer if you describe in more detail what you're trying to do.

I'm really sorry I didn't submit an entry for the Bell Pepper theme. It's a subject I've wanted to try for a while. Unfortunately, the approach I used for my first attempt didn't work out, and I've been so busy the last couple of weeks that I didn't have time to come up with another method. Fortunatly, there are quite a few excellent entries to choose from. I doubt I could have produced something as good.

There's welshblue's chain link tutorial. It doesn't use Trail, but the step where the rectangle selection is stretched could also be accomplished with Trail. (And perhaps someone in a follow-up comment suggests that.).

Probably he easiest solution is to: Select the File>Open menu in PDN. Navigate to the folder containing the file. Right click on the file, and rename it to the .pdn extension. You should then be able to open it. You can then save it as a JPG file by changing both the name and the file type

@nabsltd, Crop to Selection does what it's supposed to do: it leaves only the selected region in the final image. You just don't like what it does. A while ago, I wrote a plugin that may help, called Unselected Rectangle Keeper. For your example, you'd select the background with the Magic Wand, then run Unselected Rectangle Keeper to delete everything outside the bounding rectangle,. You then can select the transparent region, invert the selection, and crop.

Using Paste into New Image doesn't delete the other image. They're both there in PDN, and you can switch between them. At the top of the PDN window, there will be thumbnails for all the open images. Click on the one you want. In your case, you'll have two: the newly created image, and the image you were working on. If you delete the new image, you'll be right back to the single image you were working on.

@Francesc please see a recent, related thread. In my opinion, the Magic Wand is usually the wrong way to go. Use, instead, a plugin that converts white pixels to transparent pixels. It will handle character edges much better.

I can see how plugins might be able to use this to do some interesting things. A pugin could, for example, expand or smooth the current selection, and generate a file for the user to load. I'm guessing plugins will still be restricted from modifying the selection, themselves, though it might be nice if they could. I realize disk space is cheap these days, but I was wondering if an aliased selection, such as generated by the Magic wand, would result in a rather large file, since the selection boundary must track a jaggy path around every pixel.

@windydick I suggest you just bite the bullet and install a plugin. It's not that hard to do, and once you have the plugin, eliminating the white background will be easier, and much better, than using the Magic Wand. You can install BoltBait's plugin pack, which includes Switch Gray to Alpha, among many other useful plugins. The plugin pack has an installer, which will do most of the work for you. Installing the plugins should be as easy as, or easier than, installing Paint.NET. Running the plugin is as easy as running a built-in effect. You won't even know the difference.

Another method, which I tried, and found works really well for this case, is to use pyrochild's Grid Warp. Set the number of horizontal grids to 1, and the number of vertical to grids to something that fits the height of the text, while leaving it in a singe grid cell. Then just move the left and right nodes around. You could try using more horizontal grids to give more control, though it also make it a little harder to keep the text spacing and orientation correct.

You could try Paste Warp+. For this case, you'd want to set the Horizontal Distortion to 0.0. Type the text with the Text tool. Best to type the text on a transparent background. In another layer, create the region you want to make the text follow. I suggest using the Line/Curve tool to draw the outside outline. Make sure it's a closed curve -- no breaks. Use a Rectangle Select to select the text (or perhaps, select the text background with the Magic Wand at a very low tolerance, then invert the selection). Select the interior of the "swoop" region with the Magic Wand. (Probably in a new layer) Run Paste Warp+. As mentioned, set the Horizontal Distortion to 0. It helps to work at a fairly large scale, so that the results are smoother. Using Antiaiiasing in Paste Warp+ may also help, though part of the problem is that Macic Wand selections aren't antialiased.