Differences in ILDA test patterns, scanner noise

[steven850]

Hi,

I am using HE laserscan and included in the software are a couple of tets pictures. One of them is the ILDA 12K/30K test pattern. I also have an ILDA test pattern that I downloaded in the tuning guide thread from buffo.
There seem to be some differences between these two test patterns and I cannot figure out which one I need to use to tune the scanners with.

The one from Buffo displays perfectly as is, even at 50K. I can even Increase the scan angle at 30K to full angle and the pattern still displays perfectly. Displayed here at 30K with 8 Degrees.


With the built in test pattern, I get a Good sized circle at about 15-20k, at 30K the circle is really small and distorted. The scanners also make a clicking noise when displaying the built in ILDA pattern. (See vids) They only make this noise on this pattern tough, I have other patterns that are considerably larger, and the point count is higher and these all display flawlessly without noise.

So what is the difference between the downloaded ILDA file test pattern and the built in one?
Why Can I display one at 50K and the other at only 20K?
What is causing the scanners to make this noise? Why only on this pattern?

Here is a link to a few 5 second videos displaying the different test patterns. all patterns running 30K 8 degrees.

https://dl.dropbox.com/u/88689007/Laser/MOV001.mp4

https://dl.dropbox.com/u/88689007/Laser/MOV002.mp4 Downloaded ILDA file

https://dl.dropbox.com/u/88689007/Laser/MOV003.mp4 Included ILDA test in HE laserscan

Any Opinions? Dont want to start tuning with the wrong pattern.

EDIT: Here is a vid of the lasers built in test pattern running. When ever the pattern makes it to the left side it starts to make a squeeling noise, then the pattern is centered or to the right everything is fine. Bearings in the scanner? Do I need to be worried about this?
https://dl.dropbox.com/u/88689007/Laser/MOV004.mp4

EDIT: Laser is a LaserWorld PRO-1000RGB 50K scanners, I believe everyone on here tunes these at 30K. 8-10 Deg

[steven850]

The Downloaded .ILDA file from buffo is also larger then the inlcuded test file, yet it displays better. Here are some screenshots of the test patterns in the software.


.ILDA from buffo


Included Test Pattern

[Things]

Sounds to me like the one buffo provided has a few more points added. Oddly that circle should be inside the box, not outside of it. Might have to wait for buffo to chime in here about his file

[buffo]

Things: That's not the pattern I provided in my thread.

Steven: I think you have your pictures labeled backwards. The ILDA test pattern I linked to in my original scanner tuning thread is the multi-color version. You appear to also have a monochrome (white) version of the pattern. I am assuming that this monochrome version is the one that came with your software. And it's very clear that the monochrome version of the pattern is broken. That is what is causing the differences in the two images. Let me explain....

You posted a picture of the monochrome test pattern being displayed, and you mentioned that it looked perfect (when in fact, it wasn't - see below). You also mentioned that you could scan it all the way to full size. This is your first hint that something is wrong. You should not be able to scan the 12/30 K pattern at full angle - ever. There are features in the pattern (namely the circle) that are specifically designed to be BEYOND THE CAPABILITIES OF ANY SCAN SYSTEM. If you could scan it at full angle, you'd have a set of scanners that are at least an order of magnitude faster than the best scanners on the market today.

The second hint is the fact that the circle in that first picture (again, the monochrome pattern) lies completely outside the center square. This is incorrect. It is supposed to be fully contained INSIDE the square. Here's why: That circle is actually made up of 12 points that form a dodecahedron (12 sided regular polygon) that is well outside the center square. (You can see this in the ILDA frame editor view.) This is to ensure that the points are so far apart that the scanners have no hope of ever reaching them, thus ensuring that the scanners are under constant, maximum acceleration while drawing that part of the pattern. The scanners actually make 3 trips around that circle, and they are ballistic the entire time. Now, the ILDA tuning standard specifies that the circle should be just touching the walls of the center square, and the circle should be round and completely even (no re-trace visible). This corresponds to approximately 3dB of distortion of those 12 points, which is the agreed upon standard.

If you have the circle outside the square, you are nowhere near 3 dB of distortion. Sure, it might look good, but it doesn't tell you anything about how your scanners will perform when they are truly ballistic. The key to remember is that the circle portion of the ILDA test pattern is not about getting the scanners to faithfully represent exactly where all the points are, but rather to get the scanners to distort in a predictable, repeatable manner when they are pushed beyond their limits.

Bottom line: the monochrome test pattern you have is incorrect. Don't use it. Instead, use the multi-color one that I posed in my thread. (The one where the center circle is drawn in blue.) And right away, you'll notice that the center circle looks terrible at large scan angles. It's pulling apart, it's squashed into an elipse, and it resembles a spiral in a way. This is typical of a set of scanners that are trying to scan the ILDA test pattern too wide. Back off on the scan angle until the circle looks stable, and then try tuning from there. (Note that it's better if you are at 5 degrees or less when you start this.)

Adam

EDIT: It just occurred to me that HE Laserscan might have a problem reading the ILDA format 4 and format 5 frame files. If that's the case, this might be why my pattern is showing up as monochrome when it's really in color.
Unfortunately, I'm also worried now that HE Laserscan might be trying to optimise the frames before they are sent to the scanners. If that's the case, it might make tuning nearly impossible.

[mixedgas]

Ack! One of my worst fears has been that some-one would run the pattern through a frame optimizer/editor and distribute it.
Looks like its happened.

Steve

[steven850]

A quick update for those following the thread.

The images are not labeled backwards. The downloaded .ILD file displays in mono color.

Have been talking to buffo per PM as well as the creator of HE Laserscan.

The .ILD I downloaded from the thread as well as from the pangolin site is the correct file. The file is unaltered and un-optimized.

HE Laserscan has an optimization system built in to protect the scanners. The optimization can be disabled but only if the point count and distances are low enough.

The built in test patterns are all run WITHOUT optimization. So for tuning only the built in patterns can be used.

The mono color on the .ILD files is because there is no color table included in the .ILD file, the file assumes a pangolin color table, I had to go into the settings on laserscan and select "Load Pangoling color table" the file now displays in color, but with optimization still enabled.

The scanner noise is "loud" but normal.

Will start tuning with the built in pattern and update later.

Steven