Wow, the ignorance in this topic drives me nuts!
FlySolo, and the OP, all the images you are "zooming" are highly interpolated! You should educate yourself about image scaling and interpolation
before you continue...
Let me make it easy for you to understand...
In image forensics the best way to scale an image (digital zoom) is to use "nearest-neighbor interpolation" also known as "pixel resize". Basically,
nearest-neighbor interpolation replaces every pixel with 4 pixels of the same color. It doesn't change any of the pixels that were originally created
by the camera. However, pixel resize does tend to make images appear pixelated because all you are doing is making the original pixels larger, and
you are not adding or removing any detail, it is the best when studying an image because you can study the exact pixels from the camera.
ALL the images YOU and the OP have scaled (digital zoomed) used other forms of interpolation such as bilinear, bicubic, or lanczos resampling, or
other custom interpolations. Those types of interpolations can NOT and should NOT be used for image forensics because it ADDS PIXELS THAT NEVER
EXISTED. The final result of that type of interpolation is a COMPLETELY COMPUTER GENERATED IMAGE, and is NOT REAL. It SHOULD NOT be used for
studying because you are studying something that was computer generated.
Those types of interpolation were designed to smooth images for aesthetic purposes only. To make your family vacation images look smoother when you
want to scale them up a little.
Using those types of interpolation on images with preexisting compression artifacts on them will make lots of straight edges, and weird anomalies that
don't actually exist anywhere other than that digital image.
With that said, the only real way to extract more detail from less detailed images ( like used by professional forensics labs to read a license plate
number from a crappy security camera ) is to do "IMAGE AVERAGING" of MULTIPLE less detailed images. This technique is only possible if you have
multiple images or multiple frames from a video. The color of each pixel of every image/frame is averaged together to reduce noise and increase
clarity. There is NO OTHER WAY.
So, with that said... There is no "special software" that can resize a single image and get more detail from less detail. So the OP's video is
Almost all image editors today like Photoshop, PaintShopPro, Gimp, etc.. allow you to choose the type if scaling / resizing algorithm to use when
rescaling / resizing an image. The best and only way for image forensics is to use "pixel resize" also known as "nearest-neighbor interpolation".
Every other method of interpolation actually adds pixels that never existed and the final result is completely computer generated, and will create
computer generated anomalies which you may think are physical objects or characteristics.
PS, Well done Phage.
edit on 1-9-2012 by senselessness because: (no reason given)