The Chirpchart plugin draws a chirp chart test image.
A chirp chart is a sine wave which is increases it frequency towards the edge of the image. It is used to examine how cameras, digitisers, digital processing and monitors respond to different spatial frequencies.
When using a chirp chart, you are looking for two effects - Attenuation - where the sine wave stops looking black and white and starts looking grey - and aliassing or moiré patterns appearing. These look like phantom circles - usually grey - appearing where they shouldn't.
The above chart has aliasing: all the circles should spread out from the bottom left corner. The ones from the other corners and the circles near the centre of the image are phantom circles caused by aliassing.
Always view a chirp chart at 1:1 resolution - you'll see in a moment why.
Create a chirp chart, and set it to be as big as possible. Fill the
screen with it (push the space bar to make the viewer the
only thing in Shake's window). Keep the resolution at 1:1
(push Home, just to be sure). Now stand back. Squint at
it. Does it like equally white all over, or do some of the
circles look greyer? Is there a greyish band on the image?
Maybe the white parts are consistent, but the blacks seem to
be brighter in parts. If you see this, it's because the
monitor cannot display certain frequencies.
Avoid using a monitor which exhibits grey bands for detailed finishing work.
Now look at the colours. Is it white all over, or are their pinky/bluey bits? This is the same problem, the monitor cannot display certain frequencies, but it's abilities change depending on the colour. Again, it's a sign of a bad monitor.
Look at your chirp chart zoomed out to a 1:2 view. You will immediately see phantom circles appearing in the image. Shake uses a fast and crude scaling technique in the viewer. In a real image you'd see jaggies, on a chirp chart you see phantom circles.
Zoom back to 1:1 and add a Zoom node to the Chirp Chart. Set its
scaling to 0.5 The cirlces reappear, but not as badly as
before. Shake is doing a better job of scaling the image.
Try a few different filter types. Notice that some
(like Dirac) have a lot of aliassing, a lot of fake circles,
some (like triangle) attenuate the image, so the stripes in
the top right corner look a lot darker than the ones in the
Replace the Zoom node with a blur node. Increase the amount of blur and watch what happens. Perhaps it's a bit surprising: the image doesn't actually get any more blurry, just the top right hand part of the image goes grey. A blur filter is a low-pass filter - it takes away high frequency detail and leaves low frequency information intact. Since in the chirp chart, all the high frequencies are in the top right hand corner, that part of the image disappears. The blur has also aliassed - those phantom circles are back.
Now try a rotate of 2 degrees isntead of the blur. You'll see a patchwork of greyer values. The rotate operator seems to attenuate, but it doesn't seem to alias. That's good news.
Chirp Charts are useful for seeing what effects are doing to your images - you can detect whether they will loose high frequency detail by looking for where the top right hand corner of the image goes grey, and you can look for where they'll introduce jaggies by looking for phantom circles - the closer the circles appear to the bottom left hand corner of the image, the worse the jaggies will look.
Render and print out a chirp chart. Check closely for aliassing on the print-out. (yes, printers can introduce problems too). Now film it with the camera. Move the chart further and further away from the camera. What happens when parts of the chart become too finely detailed to capture clearly? Do they attenuate and become grey, or do those phantom circles appear - does your camera alias? If it aliasses, what colour are the phantom circles? Most single CCD cameras will alias in different colours at different times, usally the red and blue and circles appear closer to the bottom left corner than the green ones. If your phantom circles are white, that's good news - it means you won't get weird coloured jaggies and other bizarre effects in your finished shot.
This is an extremely useful procedure to go through before purchasing or hiring a camera: it will show you exactly how much detail the camera can catch before it attenuates or aliasses, and it will indicate how this will happen, and whether it will mess up the colours when it does. The Resolving Power of a camera is found by measuring how close the lines can get before attenuation or aliassing become noticeable.
|width||Width of chart|
|height||Height of chart|
These may appear at some point in the future - if you want them now, get in touch: email@example.com