MJW

Is the brass dial generated, or is it a photographic image? I'm betting that it's from a photo. If it was produced in PDN, I'm incredibly impressed, and would likewise like to know how it was done.

I'm afraid that's probably not possible. Paint.NET only runs under Windows. I assume (though I don't know for certain) that MacBook Pro's have an operating system other than Windows. Also (before you ask), Rick Brewster, the PDN developer, has clearly stated he will not port PDN to other operating systems.

I don't think the code they have is the type that could easily be incorporated into a plugin, though I find it hard to tell what exactly it is. It's easy to say, "if I were a programmer ...," but I think it might be easier for a non-programmer to learn programming than for a programmer to convert the colorization code into a plugin.

If you mean the several-pixels-wide border around the edge, what I would do is either: Add a black background layer to make the border and text more visible. Make the original layer active. Select the Eraser tool with a Brush width of about 7 and a Hardness of 100%. Erase the border, being careful around the text. Delete the black layer. Or, Add the black background. Make the original layer active. Select the Rectangle Select tool Draw a rectangle of about the right size to contain everything except the border. Select the Move Selection tool. Move the selection borders to exactly include everything but the border (with a bit of leeway around the border) . Use Edit>Invert Selection so now the border is selected.. Use Edit>Erase Selection. Delete the black layer. (In place of Invert Selection followed by Erase Selection, you could use Image>Crop to Selection.) I wouldn't use the Magic Wand tool to erase the border, since in my experience that sometimes leaves a few almost-transparent pixels. I also wouldn't bother trying to use Image>Canvas Size to reduce the canvas size just enough to eliminate the border, though that should work. Though the methods I suggest may seem simplistic, often the straight-forward ways are the best.

I wrote a (beta) plugin called Perspective Transformation that does something somewhat similar. I'm not sure it does what you want, though, because there's no perspective specification, as in your example. I really don't understand how to make sense of the control you show. It has a quadrilateral, which seems to represent a perspective-transformed rectangle, and it's got a control to specify the amount of perspective. But the perspective transformation for a given rectangle to a given quadrilateral is unique: changing the amount of perspective changes the quadrilateral. I don't doubt what they're doing makes sense; I just don't know what they're doing.

Congratulations to Pixey on her imaginative and well-executed entry, and to Sasha (whose entry I particularly liked) and Lynxster4.

Thank you, Rick.

When there's an exception in a plugin when running under Visual Studio, instead of breaking like it normally would, it brings up the rather terse PDN crash message window. I'd much rather break so I could look at the value of the fields, etc. I assume the crash handling catches all otherwise-unhandled exceptions. Can that be disabled so I can use the debugger instead? Or once the window appears, is there a way to debug the code that caused the exception?

Two other possibilities are Ed Harvey's Color>Threshold and Rei's Stylize>Ostu Threshold. (I think the control given by Two-Tone Threshold or Threshold is probably preferable to the automatic Ostu Threshold.)

Some of that can be done by writing a plugin or, probably more easily, with pyrochild's ScriptLab, but some cannot. Neither a plugin nor ScriptLab can create a new image or write an image to a file. Recently, toe_head2001 posted a link to an interesting technique that might perhaps allow you to do what you want, though I haven't yet read it carefully enough to fully understand it. File-type plugins can read in images and write to files, so perhaps some sort of very unusual file-type plugin could do it. I would guess that using some method of recording the keystrokes would be much easier.

You don't tell us which alignment plugin, but if it's "Object Align" the image must consist of a nontransparent region (called an object) surrounded by transparency. No plugin will center a selection. That's impossible for a plugin to do, because plugins can't move the selection, or write outside the selection. I would guess that what the alignment plugin did when you ran it is to move the nontransparent region within the selection to the center of the selection. If, as is likely, the entire region inside the selection was nontransparent, that would do nothing. Assuming the region you want to center isn't transparent, and that you expanded the canvas, which would make the added region transparent, all you should need to do is to run the Object Align without making a selection.

I was, some time ago, rather chagrined when I realized that merging two layers doesn't produce a singe layer with the equivalent effect. Unfortunately, there's no way that can be done with most combinations of blending modes.

If I deactivate the black layer, copy the merged image, then paste it into a new image, it looks the same. However, if I add a black background, the result looks different then the original image with the black background. That's expected. The merged image can't retain the blending characteristics of the various layers that make up the original image. That's true even with a single layer. If I have a white-to-transparent gradient layer in Additive mode, the flattened version will have the Normal blending mode, so they will look different against a black background.

I don't know of a way. I think you ought to just bite the bullet and use a plugin.

I suggest that you don't use the Magic Wand, but instead use a plugin like BoltBait's Paste Alpha. Copy the circle into the clipboard, then paste the alpha into a layer in which you've generated a gradient. (Paste Alpha gives a choice between pasting the alpha or pasting the gray-scale values to be used as alpha. In most cases, it's easier to generate a gray-scale version, since there are more tools and plugins to modify the color than to modify the opacity.)

I've never seen a case where the flattened image didn't look like the original image. Perhaps you could post each separate layer of the three-layer + black-background example, along with the flattened image. Make sure you tell us the order and the layer properties of each layer. That way we can try it ourselves. The simpler the image, the better. You don't need to post a large image; any region that has a difference should be sufficient.

What are you saying happens? If you flatten the image, does it not look the same as the image before flattening? If you save the flattened image, does the saved image not look like the flattened image? If it doesn't work, one of those two things must not be true. Which one is it?

If you just want a flattened version that looks the same, I suggest copying the merged image with Ctrl+Shift+C and pasting it into a new image with "Edit>Paste into New Image" or Ctrl+Alt+V. I prefer that to flattening and undoing, though it's mostly a matter of taste. The advantage, if any, is that the original file name remains associated with the multi-layer image.

I'm confused by something. If I define a PluginSupportInfoClass the following way, the effect's Tooltip has all the correct information: Define a base class as: public class PluginSupportInfoBase : IPluginSupportInfo { public string Author { get { return ((AssemblyCopyrightAttribute)base.GetType().Assembly.GetCustomAttributes(typeof(AssemblyCopyrightAttribute), false)[0]).Copyright; } } public string Copyright { get { return ((AssemblyDescriptionAttribute)base.GetType().Assembly.GetCustomAttributes(typeof(AssemblyDescriptionAttribute), false)[0]).Description; } } public string DisplayName { get { return ((AssemblyProductAttribute)base.GetType().Assembly.GetCustomAttributes(typeof(AssemblyProductAttribute), false)[0]).Product; } } public Version Version { get { return base.GetType().Assembly.GetName().Version; } } public Uri WebsiteUri { get { return new Uri("http://forums.getpaint.net/index.php?/topic/BaseClass/"); } } } Then define a derived class as: public class PluginSupportInfo : PluginSupportInfoBase, IPluginSupportInfo { } If I add a do-nothing, no-argument constructor call to the derived class, as follows, the Tooltip no longer contains the plugin information. public class PluginSupportInfo : PluginSupportInfoBase, IPluginSupportInfo { PluginSupportInfo() : base() { } } For anyone with a firm understating of constructors (i.e., not me), it's probably obvious, but I don't see why the first works and the second doesn't. (I intend to pass arguments in the base constructor eventually, but I wanted to test the simplest case first.)

No. I think you misunderstand what you're seeing. When you paste a selection that has a transparent region into another layer, the transparency is also pasted. So when you move the selection around, the transparency replaces any opaque pixels in the layer it was pasted into. For instance, suppose you have a completely transparent layer and a completely opaque layer. If you make a circular selection while the transparent layer is active, copy it, then paste it into the opaque layer, there will be a transparent round hole in the opaque layer which you can move around. The transparency has been pasted into the opaque layer. It isn't blended with the pixels in the opaque layer. If you want it blended, you need to paste into a new layer. I suggest you: Open a new image. Add a new layer (don't duplicate the background layer). Select a region in the new, transparent layer with one of the selection tools. Paste it into the opaque layer. I think you'll see what I mean. EDIT: I guess I see what you mean, and the answer is, "Sort of, yes." The transparency doesn't work within the same layer because when you paste into a layer, the selection replaces what's in the layer, it isn't blended with it. That's the way it works, and (in my opinion) the way it should work. As I said, to get the result you want, paste the selection into a new layer. You can re-position it, and when you get it where you want it, you can merge the layer down.

It does seem a bit strange, but it makes sense. The selection pasted into a layer replaces the region in the layer, so the corresponding area in the new layer is made transparent. The solution is to paste into a new layer.

I shall use all my power and influence!

That seems very, very, very silly to me. Green ought to be green! As a programmer, I would always assume Red, Blue, and Green represent the full-strength primary colors. That's really annoying!

Instead of what I would expect, (B: 0, G: 255, R: 0). Red and Blue are both what I would expect. This is in PDN 4.0.12.

Thanks, BoltBait. I'll leave them as they are then. I'd never bothered looking at he stuff before, and it seemed slightly weird.