Results 1 to 7 of 7

Thread: ilda test pattern vs. graphics

  1. #1
    Join Date
    May 2007
    Location
    1 hr from everything in SoCal
    Posts
    2,549

    Default ilda test pattern vs. graphics

    I have heard a few different things on this so, I figured I would post a thread. First, does the ilda test pattern tax the scanners pretty hard, or is it just difficult to display well, but doesn't put a lot of strain on the scanners? Second, if you can't scan the ilda test pattern well at 20 degrees at a given speed, does that mean that most graphics scanned at said settings will all look poor?

    Let's say that the scanners produce the test pattern perfect at 8 degrees at 30K. What if we turned the scan angle up to 20 degrees and tweaked the tuning to get rid of over/undershoot, but the circle cannot be made better and is not produced properly, what kind of graphics can we expect?

    Is the test pattern just a tool to show the mean performance of the scanners, or does it directly say "this is how your graphics will look, even if they are not as taxing as the test pattern"?

    Please also let me know if these are valid questions!!
    Last edited by absolom7691; 11-04-2008 at 09:39.
    Those who fail to grasp art are the ones who criticize it.

  2. #2
    Join Date
    Oct 2006
    Location
    Cleveland, Ohio
    Posts
    2,333

    Default

    I think the test pattern is two things: a standard by which you can rate your max scanner speed, and a tuning standard with which ild graphics files should look the same across properly tuned projectors.

    If you re-tune the pattern up at 20 degrees, then the ild files that are meant to be displayed using 8 degree tuned galvos won't look right.

    So, it's not really directly showing how your graphics will look, because you could create graphics that look ok on your theoretical 20 degree tuned scanners. The problem is you wouldn't get better performance really, just 'different' and in doing so your ild files wouldn't be compatible with other peoples 8 degree tuned scanners.

  3. #3
    Join Date
    May 2007
    Location
    1 hr from everything in SoCal
    Posts
    2,549

    Default

    I understand that the performance at a larger scan angle won't stay the same or get better, but will the graphics look "okay"? If the test pattern is one of the hardest things to display and at 20 degrees no longer looks better, then would it not be true that the letter "A", optimized, will look better than the test pattern?

    I guess what would be best is, if I needed a larger scan angle, then 30K would not be advisable and it would be better to step the galvos down to 25K or whatever speed would look best at that angle? I guess I was just trying to figure out that if the ilda pattern didn't look great, my graphics would look poor or if they would look "acceptable".

    Thanks Lava!
    Those who fail to grasp art are the ones who criticize it.

  4. #4
    Join Date
    Jan 2006
    Location
    Charleston, SC
    Posts
    2,147,489,071

    Smile Thoughts on scan angle and tuning...

    Quote Originally Posted by absolom7691 View Post
    First, does the ilda test pattern tax the scanners pretty hard, or is it just difficult to display well, but doesn't put a lot of strain on the scanners?
    That depends on how wide you scan it.

    The center circle is certainly taxing on the scanners (meaning, they're under maximum continuous acceleration), but since the points are in (more or less) a circle, you don't have hard steps in the input that might throw the scanners into some sort of resonance. They can still get very hot, and overheating *can* cause scanners to fail. (Albeit over a longer period of time.)

    However, if you scan the entire pattern too wide, then the corners and other lines which *do* have abrupt transitions *could* cause problems - especially if your tuning is already off enough that resonances could develop. Resonance can cause things to break in a very short period of time.
    Second, if you can't scan the ilda test pattern well at 20 degrees at a given speed, does that mean that most graphics scanned at said settings will all look poor?
    Not enough information to answer. How bad does the ILDA pattern look at 20 degrees? Is the circle still round, but it just doesn't quite reach the square, or is it some malformed elipse that looks like it was drawn by a 2 year old? In the former case, graphics will probably look fine. In the latter, all but the simplest graphics will still look awful.
    Let's say that the scanners produce the test pattern perfect at 8 degrees at 30K. What if we turned the scan angle up to 20 degrees and tweaked the tuning to get rid of over/undershoot, but the circle cannot be made better and is not produced properly, what kind of graphics can we expect?
    Probably not very good, but it depends on the individual scanner.

    One thing you *can* do to demonstrate this is to crank up the scan angle to, say, 13 or 14 degrees and then try the tuning adjustments you just talked about. (Be careful now, because you're pushing things here, but it's worth it to illustrate the issue.) You'll see that it becomes very hard to get rid of overshoot without introducing undershoot, and nearly impossible to eliminate undershoot without causing some other artifact to crop up. Play with it for a while though, and you'll get a feel for it. Just remember that when it's all over with, you're going to have to turn the scan angle down to a reasonable value and then re-tune everything again.
    Is the test pattern just a tool to show the mean performance of the scanners, or does it directly say "this is how your graphics will look, even if they are not as taxing as the test pattern"?
    It's more of a "worst case scenario" than an average performance indicator. If you can scan the ILDA test pattern, anything else you scan should be easy by comparison. That's why you may only get 8 degrees on the test pattern, but you might get 20 degrees out of most shows, and there may be a few that you can run at 30 degrees.

    Also, don't forget the other test patterns, especially the laser media test pattern and the grid pattern.
    Quote Originally Posted by drlava View Post
    I think the test pattern is two things: a standard by which you can rate your max scanner speed, and a tuning standard with which ild graphics files should look the same across properly tuned projectors.
    This is an excellent point that I didn't address... The fact that it also serves as a standard with which to compare systems. Good point, Andrew!
    If you re-tune the pattern up at 20 degrees, then the ild files that are meant to be displayed using 8 degree tuned galvos won't look right.
    I don't agree with this. If you *properly* tune your scanners, and find that you can get 20 degrees out of them, then everything else you scan will look fine as well, so long as you don't try to scan significantly wider than you tuned.

    For example, my new ScanPro50 scanners are running at 30K in my projector. As a result, I can scan the ilda test pattern at 18.5 degrees now. And all my shows look good up to around 35 degrees or so. (A few shows start to look bad at around 30 degrees though.) But everything looks just fine at 4 degrees as well. The angle limit really only applies on the upper end.

    The key is *proper* tuning. If you can get the ILDA test pattern and the LaserMedia test pattern *both* to look perfect at a given scan angle, then you're safe at that angle or below. (Actually, you can probably exceed that angle by a good margin, because the ILDA test pattern is a worst-case scenario that is rarely, if ever, found in normal ILDA frames.)

    But if you tune your scanners at 20 degrees, and the test pattern looks half-assed at that size, then you can't really predict how your graphics will look at *any* given angle. There may be an angle that they look good at, or they might look like hash at all angles. That's because you don't know which artifacts are caused by bad tuning and which ones are caused by pushing the scanners harder than they can handle.
    Quote Originally Posted by absolom7691 View Post
    I understand that the performance at a larger scan angle won't stay the same or get better, but will the graphics look "okay"? If the test pattern is one of the hardest things to display and at 20 degrees no longer looks better, then would it not be true that the letter "A", optimized, will look better than the test pattern?
    Maybe... Maybe not. It's hard to judge what looks "OK" to you. Maybe you don't mind seeing tails on otherwise straight lines. Maybe you don't mind that circles become slightly-squashed elipses. That's why we have the test pattern. It's got a set of standards that go with it, so everyone can look at it and say "Yes, that's right", or "No, that's wrong."
    I guess what would be best is, if I needed a larger scan angle, then 30K would not be advisable and it would be better to step the galvos down to 25K or whatever speed would look best at that angle?
    Absolutely! If you need a wider scan angle, slow your galvos down and re-tune them at that slower speed.

    This is *exactly* why I'm running my Scanpro 50 scanners at 30K. Sure, if I re-tuned them for 50K I would get a little less flicker on some complex images. But I'd loose 2/3 of my scan angle! (I could only get 6.5 degrees out of them at 50K speeds.) So by slowing them down to the standard speed of 30K, I picked up a *huge* increase (up to 18.5 degrees on the test pattern) in scan angle. I'm at the point now where my images are frequently too wide to fit on the wall, and I have to run my scan angle setting in the Pangolin software down around 36% or so to make it fit!

    Adam

  5. #5
    Join Date
    May 2007
    Location
    1 hr from everything in SoCal
    Posts
    2,549

    Default

    Thank you both. Adam, you answered every single question I had with an "easy to understand" answer. Thank you. You also taught me a little more about scanners and how they work besides just an XY operation. One last simple question, and I know this depends on how complex the image it, but how much flicker could one expect running 25K instead of 30K. On an average image, or hell, even the test pattern?
    Those who fail to grasp art are the ones who criticize it.

  6. #6
    Join Date
    Oct 2006
    Location
    Cleveland, Ohio
    Posts
    2,333

    Default

    Quote Originally Posted by buffo View Post
    Let's say that the scanners produce the test pattern perfect at 8 degrees at 30K. What if we turned the scan angle up to 20 degrees and tweaked the tuning to get rid of over/undershoot, but the circle cannot be made better and is not produced properly, what kind of graphics can we expect?
    If you re-tune the pattern up at 20 degrees, then the ild files that are meant to be displayed using 8 degree tuned galvos won't look right.
    I don't agree with this. If you *properly* tune your scanners, and find that you can get 20 degrees out of them, then everything else you scan will look fine as well, so long as you don't try to scan significantly wider than you tuned.

    Adam
    If the circle cannot reach the square in the 20 degree tuned galvos as he stated, then I stand by that some graphics won't look right. I think you would agree, maybe you missed the 'circle cannot be made better and is not produced properly' part. If he slows the scan down from 30k to the point where the circle looks right, then yes the graphics will look right too.

  7. #7
    Join Date
    Jan 2006
    Location
    Charleston, SC
    Posts
    2,147,489,071

    Cool

    Quote Originally Posted by drlava View Post
    If the circle cannot reach the square in the 20 degree tuned galvos as he stated, then I stand by that some graphics won't look right. I think you would agree, maybe you missed the 'circle cannot be made better and is not produced properly' part.
    Yes - I agree with you that if it's not producing the circle properly, then everything else will look bad as well. My point (which I admit was rather obtuse) is that if you are at that point, then you can't say you have "tuned them to 20 degrees". Thus my comment about properly tuned scanners that followed.

    Essentially, if you have "tuned" your scanners, then that means the circle *is* properly displayed. (That's what tuning involves.) And if that is the case, then the angle doesn't have to be 8 degrees, it just has to be within the absolute abilities of the scanners.

    So it really is a problem of definition... If you "tuned" the scanners at 20 degrees, and the circle looks lousy, then the scanners aren't really tuned, are they?

    But if they *have* been properly tuned, then it is possible to have a set of scanners that could display the test pattern at 20 degrees, assuming that the scanners are quite robust. (For the record, the best I've ever seen is 18.5 degrees, out of the ScanPro 50's.)

    Though it would be interesting to see what you could get out of the Cambridge 6215's with the high current amps if you dialed them back to 30K and tuned for max angle... I wonder if you'd get close to 20 degrees? I'm thinking you probably would.
    If he slows the scan down from 30k to the point where the circle looks right, then yes the graphics will look right too.
    Right. I think we're both on the same page here.

    Adam

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •