I created a quick test app to experiment with the performance of the P/Invoke AlphaBlend calls as discussed in this blog post by Chris Lorton. The AlphaBlend native call does not support per-pixel alpha blending so the alpha value is specified for the entire image. The "incr" label in these two videos is the step value for the alpha channel used in the animation. The frame rates were better than expected, right around 17-20fps.
The tool I am using to record the video is called CoolCapture (I'm using the trial version here). It's very cool. It's essentially a fully featured screen capture tool for Windows Mobile 5 devices. You can generate both videos or pictures. I was fairly impressed with the performance as well; capturing these two videos below only resulted in about a ~6 fps reduction in my animation (thus the blending doesn't look as smooth as it actually is). The trial version limits the recording interval to ~10 seconds (though the webpage says 5 seconds).
Note that the capture resolution was 240x320 (the device resolution); however, this was automatically changed to 640x480 when I uploaded the video to YouTube. This is why the videos have that stretched out look.
Showing posts with label Transparency. Show all posts
Showing posts with label Transparency. Show all posts
Wednesday, December 20, 2006
Wednesday, May 24, 2006
Alpha Blending
Since my last post was on drawing transparency color keys in .NET CF, I thought I should explore alpha blending as well. It is possible to do alpha blending .NET CF, but not without some effort (i.e., venturing into p/invoke). Here, at the .NET CF UI and More blog, the author writes:
Briefly, there are two ways to do alpha blending in WM5: the AlphaBlend() function and with the Image COM object in the Imaging API.
He then goes on to explain how to implement these two methods. (direct link). The happy alternative is offered through OpenNetCF (who else) in their SDF 2.0 library. See Alex Feinman's post here.
Transparency
This morning I googled around for "standard transparency colors" in hopes that I would find a color key that is typically used to represent transparency in images. In .NET CF, the only way to make part of an image transparent is through the following code:
ImageAttributes attr = new ImageAttributes();
attr.SetColorKey(Color.White, Color.White);
Rectangle rDest = new Rectangle(xImage, yImage, Image.Width, Image.Height);
g.DrawImage(Image, rDest, 0, 0, Image.Width, Image.Height, GraphicsUnit.Pixel, attr);
In this case, the color "white" would be made transparent in the given Image. Note that although SetColorKey allows the developer to specify a transparency color range, I believe that this range is limited to one color on .NET CF. During my search, I found that the standard chroma key is "bright pink." However, it's also convention to automatically select the chroma key based on the upper-left hand corner pixel of the image, in hopes that this is representative of the "background" pixels. Here's a post by Ed Kaim (MSFT) that really explains this well imo.
"However, the issue you're raising is why Microsoft recommends the usage of ImageAttributes in order to draw transparency, and why the top left pixel is called out as a prime source. First of all, ImageAttributes is the only feasible way to draw transparency in managed code on the .NET Compact Framework. Otherwise you'd have to go pixel by pixel and only draw the nontransparent ones, which would be slow and painful. The reason the top left pixel is offered is because it's often transparent in images with transparency. Sure, it's not *always* transparent, but it's usually transparent, due to its location in the image and the fact that most pictures on buttons don't go all the way to the top left border. However, you can decide that your standard chroma key color is bright pink (as many do) and hardcode your ImageAttributes object to use that color for all images, provided you do the work in the art to make sure the same exact color is used on the transparent spots in your files, whatever format they may be."
ImageAttributes attr = new ImageAttributes();
attr.SetColorKey(Color.White, Color.White);
Rectangle rDest = new Rectangle(xImage, yImage, Image.Width, Image.Height);
g.DrawImage(Image, rDest, 0, 0, Image.Width, Image.Height, GraphicsUnit.Pixel, attr);
In this case, the color "white" would be made transparent in the given Image. Note that although SetColorKey allows the developer to specify a transparency color range, I believe that this range is limited to one color on .NET CF. During my search, I found that the standard chroma key is "bright pink." However, it's also convention to automatically select the chroma key based on the upper-left hand corner pixel of the image, in hopes that this is representative of the "background" pixels. Here's a post by Ed Kaim (MSFT) that really explains this well imo.
"However, the issue you're raising is why Microsoft recommends the usage of ImageAttributes in order to draw transparency, and why the top left pixel is called out as a prime source. First of all, ImageAttributes is the only feasible way to draw transparency in managed code on the .NET Compact Framework. Otherwise you'd have to go pixel by pixel and only draw the nontransparent ones, which would be slow and painful. The reason the top left pixel is offered is because it's often transparent in images with transparency. Sure, it's not *always* transparent, but it's usually transparent, due to its location in the image and the fact that most pictures on buttons don't go all the way to the top left border. However, you can decide that your standard chroma key color is bright pink (as many do) and hardcode your ImageAttributes object to use that color for all images, provided you do the work in the art to make sure the same exact color is used on the transparent spots in your files, whatever format they may be."
Subscribe to:
Posts (Atom)
