jimv wrote:
This is confusing to me. How does an irregular, micro pattern (visible at 400% magnification) become the regular, macro checkerboard pattern (visible at 100% magnification) after application of AutoDev. Are you saying AutoDev generates a regular artificial pattern that has no relation to the underlying (irregular) pattern it detects.
Correct! (but you're incorrect that the micro pattern is irregular). You're looking at "intentionally unsuppressed aliasing effects"
Aliasing can cause micro patterns to be visible as macro patterns (
see Moiré patterns for example)
First off, it's important to note that the 'pattern' that AutoDev
shows (not introduces!) is merely cosmetic and is not the result of stretching (or any other operation for that matter). You will see the same macro pattern in any other module in StarTools - it's not inherent to the functionality of the module. Again, at 100% zoom it will go away. Save the image, and it won't be visible (unless viewed within StarTools or another program that uses intentional nearest neighbour scaling). It's a visualisation perk of StarTools;
The way you're seeing AutoDev rendering your data to the screen is purely because of the way the scaling algorithm (for the purpose of fitting the whole image on the screen) works; it is intentionally coded in such a way that visualisation brings out the patterns by way of aliasing artifacts/patterns (which is what you're looking at), if such patterns are present on a mico-level. It also keeps perceptual noise constant at any zoom factor, rather than smoothing it out when zoomed out.
It's a feature, not a bug. It helps to 'unhide' issues with your data that you might have otherwise missed because you were looking at the data on another scale/zoom level.
If you're interested in how this visualisation feature works, here is an explanation; for a
Moiré pattern to become visibile, it requires that two patterns are superimposed over eachother. StarTools uses this fact to its advantage by artificially providing the first pattern. If your data - at any scale, zoomed in or not - contains the other pattern required to make up the Moire pattern, the Moiré pattern becomes visible. If there is no pattern in your data, the conditions for a Moiré pattern to become visible are not met and no Moiré pattern will be visible.
Again, look at the data at 100% zoom and you will not see the aliasing pattern (but you should see the artifacts causing the pattern in the first place). Save the image, and it still won't be visible (unless viewed within StarTools or another program that uses intentional nearest neighbour scaling).
Hope this helps.
As for what is causing the excessive mazing/zippering artefacts, I'm not sure. What I
can tell you, however is that even if those mazing/zippering artifacts were present in the individual frames, they should've long been stacked out by the stacker. It's almost like we're looking at a single frame here. Are you sure you're not just left with a single frame after rejection?