Ilda laser projector load balancing? anyone?

[masterpj]

Hey i just got a funny idea and it seems quite possible.
How about a piece of software that could splice an ilda in 2 so you have half an ilda.
which you could output to 2 laser projector. thus decreasing point per projector.
and allowing more complex graphics.

A nice setting would be that you can input the KPPS of both projectors and the software will sepperate acording to the scanning speed.
*giving the best scanner the most*

either it can be spliced in 2 parts by the software or at the blanking parts like shapes *if the other option isn't possible*
like 2 boxes and the 1 box for the first projector and the other for the other laser projector.
however splicing might be better since sometimes you have one big shape.

aligning can be done with 2 half circkles on both projector * one in a D shape and the other one flipped* once you get a nice round cirkle that means they are aligned perfectly

[mikkojay]

I did this recently with 3 projectors instead of 2. I posted a few video links, but nobody had any feedback. It was more of an experiment than anything, but the intent was beam effects, not graphics- so those are 2 completely different animals. I was concentrating on splitting off objects that spill outside the frame, while you are looking at splitting one frame.
What I found was this:
IMHO Unless you have two absolute clone projectors (which I basically have) and perfect alignment, getting a contiguous line without a visible glitch would be quite difficult, although technically possible. You would be better off to take a given frame and break it into a series of lines, then split the line/polyline list up and send half to each projector. The reason is this- you would want to avoid chopping visible lines in half.
Drawing multiple objects from different projectors would be a piece of cake though-
Imagine the game asteroids- Say there should be 10 asteroids on the screen. One projector draws 5, the other draws 5. No flicker, great! The only time this would not work would be when your frame consists of one giant line- in which case you would have no choice but to find a nice halfway point and chop it in half.
-Mike

[masterpj]

Nice, and yeah as i thought the splitting is harder than would be expected.
so it would be nice if there's a software which can split up different objects in ilda frames and gives the biggest part to the fastest scanner.
doing it manually is off course an option.
but it could also be done automatically ;3

[flecom]

its not only possible, its been done... pangolin lets you send different tracks to different projectors so you could have one projector drawing a background while another draws something else at the same time etc

[masterpj]

woa that's awsome.
To bad Pangolin is a bit too expensive for me >.> i use easylace.
would be awsome if anything like it exists

[flecom]

maybe ask the Spaghetti guy? his software is already track based so it would just need to support multiple outputs and routing to said outputs... not an easy task but should be possible at least

[drlava]

Spaghetti and Pango and LDS, etc. can do multi track to multi projectors, this is nothing new but what he is talking about is a bit different - for complex graphics display 1 frame split between 2 projectors.

I posted about this in particular at the start of 2009
http://www.photonlexicon.com/forums/...-in-point-rate

Mike, I know we have discussed this technique, but I must have missed your video links. Which thread?

[masterpj]

hmm yeah i should..
sorry for asking.. but who exactly is the spaghetti guy?

[mikkojay]

Mike, I know we have discussed this technique, but I must have missed your video links. Which thread?

Hey Andrew, I mentioned it in the "Paper or Plastic?" grocery store scanner thread a few weeks ago. I have the vids removed from my youtube channel at the moment since the project took yet another turn, but I am planning on replacing the vids soon with some updated versions that match the latest incarnation of the project. I can put the one back up though and send you a link to it when I get home from work. I'd like to capture a less blurry vid with more clarity so the technical details of what's happening are not lost due to bad cinematography
-Mike

[masterpj]

yes please ^^
a video would be nice

[mikkojay]

hmm yeah i should..
sorry for asking.. but who exactly is the spaghetti guy?

That's Gary Harper, aka JohnYayas on this forum. I'm pretty sure Spaghetti already does multiple projector output. I'm also sure he's quite capable of doing what you are talking about, but the question would be whether or not it is within the scope of what that program's function is. Anything is usually possible, it's just up to a combination of why/time/money ya know?
-Mike

[masterpj]

I think I understand what you mean.
and well i think when done it will be beneficial *i might got a part wrong*
*I read you asking what would be the benefit of making a software ( I might be wrong)*
Well it save's alot of time doing it by hand ^^
i will try to contact him... however Í don't really know yet what to ask XD

[flecom]

Spaghetti and Pango and LDS, etc. can do multi track to multi projectors, this is nothing new but what he is talking about is a bit different - for complex graphics display 1 frame split between 2 projectors.

I posted about this in particular at the start of 2009
http://www.photonlexicon.com/forums/...-in-point-rate

Mike, I know we have discussed this technique, but I must have missed your video links. Which thread?

cool good to know, I wanted to get a copy at SELEM from Gary but I was so busy all weekend I forgot lol

[JohnYayas]

It's interesting that this topic came up because this is something that I have thought about doing for a little while. The purpose wouldn't be for load balancing, though. It would be for doing things like projecting a 360 degree laser image, or easily having a ball bounce from one projector output to another. You can do that sort of thing with layers but not as easy as dividing your ILDA view into sections and drawing on it.

The load balancing idea sounds better in theory than it would probably work unless it was very controlled so that objects were tracked and kept with the same projector. Imagine an asteroids game where asteroid objects float around. If the load balancer was smart, it would keep track of the asteroids so that the same projector always displays the same asteroid. But, a simpler approach would be for the software to track some arbitrary segments. This could lead to objects jumping from one projector to another and would almost definitely result in asteroids that change in color and shape as they go from one to the other. It would be very hard to make this sort of thing work for general ILDA display without these problems.

[mikkojay]

Mike, I know we have discussed this technique, but I must have missed your video links. Which thread?

Hey Guys, I uploaded that video here:
http://www.youtube.com/watch?v=FybC_tRtVVE

It is a little rough, but you should be able to get the idea for what I'm trying to do.
@Gary- a 360 room of a ball bouncing would be WICKED!

-Mike

[drlava]

very nice Mike! Can't wait to see more of your software.

[gashead]

Hey Guys, I uploaded that video here:
http://www.youtube.com/watch?v=FybC_tRtVVE

It is a little rough, but you should be able to get the idea for what I'm trying to do.
@Gary- a 360 room of a ball bouncing would be WICKED!

-Mike

Nice work!

You got me all of a flutter this morning.. I love lasers.

[masterpj]

Hey Guys, I uploaded that video here:
http://www.youtube.com/watch?v=FybC_tRtVVE

It is a little rough, but you should be able to get the idea for what I'm trying to do.
@Gary- a 360 room of a ball bouncing would be WICKED!

-Mike

Now thats awsome ^^

And i fully agree with JohnYayas
that proble I was thinking off too.
it might be a cool idea if it could track the objects and stay with them.. even though this would be hard, it is possible.
Once the new shapes are created it assigns it in the conversion and keeps them at that specific projector.
once it dissapears it leaves it.

This might get unneficcient on a few ilda files *if certain shapes from one projector dissapear*
but it just might give a little push to less flickery graphics..

However doing this sounds like an hard effort.

Anyone know what specific problem causes to deform split ilda files? if you play them simultaneous?

[flecom]

I think instead of letting the software do it why not let the timeline do it? have it so you can assign tracks to projectors and then decide, hey, im going to send these graphics to this projector, and these other ones to another projector...

unless your specifically talking about real time application generated stuff (like the asteroids example)

[mixedgas]

This was common in the late 70s and in the 80s, with one head PER COLOR, techniques. Ie a red head, a yellow head, a green head, and a blue head.

If your galvos have good linearity, ie old General Scanning or Modern Cambridge, this is not a problem. But Linearity costs money, and from what I've seen, few if any of the "cheepie" modern scan sets have it or even test for it. So you would have to UGC them in software for a match with graphics.

There are velocity corrected test patterns such as the "Laser Media Clone" pattern out there.

Nough Said!

Steve

[JohnYayas]

Color matching is an issue, too. Probably not a big deal if you have a set of identical projectors. But, in a LEM setting, it will end up looking like those old big screens with a bunch of TVs set in a wall with one of the TVs on the fritz.

But regardless, even with mismatched laser colors and galvo positioning, it would still be a cool thing to goof around with and you could create some really huge displays. How cool would it be if someone could create a 360 degree laser projected BattleZone game? You would be able to look around and see tanks coming up behind you. That would be so rad.

[masterpj]

Color matching is an issue, too. Probably not a big deal if you have a set of identical projectors. But, in a LEM setting, it will end up looking like those old big screens with a bunch of TVs set in a wall with one of the TVs on the fritz.

But regardless, even with mismatched laser colors and galvo positioning, it would still be a cool thing to goof around with and you could create some really huge displays. How cool would it be if someone could create a 360 degree laser projected BattleZone game? You would be able to look around and see tanks coming up behind you. That would be so rad.

Yes indeed.
and woa that is awsome ;3

Well indeed mismatched colors... good calibration does reduce that effect.
even when it still occurs.... i don't mind a bit off colour mismatch for a gain in scanning speed *or at least more KPPS*

[mpolak]

The load-balancing idea can work extremely well in practice, and below are some examples of a dual 6215H 60K scan-set projector that we used for SIGGRAPH 2007's Electronic Theatre. The laser output beam and color matches between both scanners identically due to having used a Chroma 5 mixed-gas laser and dual PCAOMs as the light source. This makes it nearly impossible to pick apart the separate scan-fields and gave us exceedingly sharp images at the screen over considerable distance.

Unfortunately the photo repository of still images from the event was removed recently, but I do have one of the shots handy as an example:

sg07temptest.png

This is load-balanced between both projectors and the beam characteristics are identical, making it nearly impossible to tell which scanner any part of the image is coming from.

Unfortunately I don't have a copy handy of the video we shot on the event so I'm at the mercy of YouTube videos from audience members. Obviously the laser didn't flicker nearly as badly as it shows up on the video. I was specifically asked to do everything possible in order to allow Star Wars to run at near-full-frame-rate without having to eliminate detail as it was our "hero" game.

Star Wars -- http://www.youtube.com/watch?v=5iodpYzTgYE
Tempest -- http://www.youtube.com/watch?v=hvF3JZcGvBQ

The algorithm I employed was pretty simple in that the vector frame data (from the MAME state machine) was taken and then a search was performed forwards and backwards from the center of the display list to find the nearest blanking jump. The split was then done at that location between the scanners and the other frame was padded as necessary to keep both of the scan-list lengths identical for timing reasons. The result was an "equivalent" 120K scan-rate at the screen and an amazing level of detail with surprisingly little flicker. The 'text' screens are the worst for obvious reasons, but gameplay was buttery-smooth!

(And yes, if you were wondering, the legal rights for public exhibition of these games for this event were specifically cleared with both Atari and LucasArt's legal departments specifically for SIGGRAPH!)

[drlava]

Looks beautiful! That's pretty much exactly the way I was envisioning it could be done as well. Nice job!

[mikkojay]

Hey Matt,
I agree with Andrew, that is a great example of doing it right! I saved off a stream from some War of the Worlds so people can see that even with simple games the point count gets up there. I know what you mean about the text screens On this one, the text frames have point counts > 2500, no way is a Chinese 30k scanner going to swallow that in one bite. BTW, I have ogled that SIGGRAPH photo many times- true classic laser porn!

-Mike