Resize the transparent image to be the size of the target canvas. This will result in a larger transparent canvas.

--->
<cfset ImageResize(
     objWatermark,
     190,
     30
     ) />

<!---

Set text drawing anti-aliasing to be on. We're going to be writing text and then blurring it, so we want to let it be a little fuzzy.

--->
<cfset ImageSetAntialiasing(
     objWatermark,
     "on"
     ) />

<!--- Set the drawing color to be white. --->
<cfset ImageSetDrawingColor(
     objWatermark,
     "##FFFFFF"
     ) />

<!--- Create text attributes. --->
<cfset objAttributes = {
     Font = "Arial Black",
     Size = "20"
     } />

<!--- Draw the water mark text. --->
<cfset ImageDrawText(
     objWatermark,
     "Kinky Solutions",
     11,
     20,
     objAttributes
     ) />

<!--- Blur the watermark image. --->
<cfset ImageBlur(
     objWatermark,
     6
     ) />

<!---

Set text drawing anti-aliasing to be on for the target image. Since the image we are pasting in, we want a slight blurring.

--->
<cfset ImageSetAntialiasing(
     objImage,
     "on"
     ) />

<!---

Paste the manually created watermark image onto the photo. This time, we don't need to turn on any anti aliasing since the watermark has solid borders. The anti-alisasing only helps us when the pasted image has anti-aliased perimeter.

--->
<cfset ImagePaste(
     objImage,
     objWatermark,
     (objImage.GetWidth() - objWatermark.GetWidth()),
     (objImage.GetHeight() - objWatermark.GetHeight())
     ) />

<!--- Write the image with the new watermark to the browser. --->
<cfimage
     action="writetobrowser"
     source="#objImage#"
     />

Notice that the image we are reading in is the Base64 data:

R0lGODlhAQABAIAAAP///////yH5BAEHAAEALAAAAAABAAEAAAICTAEAOw==

This is the encoded 1x1 transparent GIF. We're also including the headers in the Base64 data, but this isn't required (it works if you take it out); I left it in just for demonstration purposes. Because we're using Base64 data rather than an actual GIF image, it lets us create transparent canvases without having a physical image on hand.

Running the above code, we get:
Clearly this is a hack. I'm sure that ColdFusion 8 has a way to do this that's more elegant, but for now, until I can figure it out, this is a rather simple workaround. I hope, however, that you're beginning to see how amazing the ColdFusion 8 image manipulation functionality is.

This article was reprinted with permision from Ben Nadel's blog:
www.bennadel.com/blog/775-Learning-ColdFusion-8-CFImage-Part-III-Watermarks-And-Transparency.htm