Good question. Let's say A represents the original "CMU" word, and B represents the blurry version of "CMU".
One approach is to keep track of an alpha channel (see slide) to identify which pixels are "transparent" and which pixels are "opaque". Let's say the alpha channel is represented by m and has the same dimensions as both A and B. The alpha channel m has a value 1 to represent opaque pixels, and 0 to represent transparent pixels. We can then produce a composite through some simple per-pixel processing: A m + B (1-m).
We won't be touching transparency and composites much in this course. But if you're interested, there's a nice lecture on transparency in Computer Graphics (15462/662).
how is the overlay part done?
Good question. Let's say A represents the original "CMU" word, and B represents the blurry version of "CMU".
One approach is to keep track of an alpha channel (see slide) to identify which pixels are "transparent" and which pixels are "opaque". Let's say the alpha channel is represented by m and has the same dimensions as both A and B. The alpha channel m has a value 1 to represent opaque pixels, and 0 to represent transparent pixels. We can then produce a composite through some simple per-pixel processing: A m + B (1-m).
We won't be touching transparency and composites much in this course. But if you're interested, there's a nice lecture on transparency in Computer Graphics (15462/662).