XY Pad Calibration

Post Reply
bssmusic
Posts: 35
Joined: Fri Sep 11, 2015 6:48 am
Location: Gensingen, DE
Real Name: Peter Baumgartner
Contact:

Hi Massimo,
one question here: Is ther really no possibility to calibrate the XY Panel? Or am I too dumb to find it?
What I have in mind is to vectorise the positions of moving heads or scanners so that i can "center" the beams to one point so that I can follow let's say the singer by joystick. The idea is to search the 4 corners (and the center?) of the stage with each beam and put this in a kind of matrix.
Until now I have manually created scenes for fixed positions, but this is time consuming and annoying.
Any ideas? Plans?

Peter
janosvitok
Posts: 1266
Joined: Mon Apr 13, 2015 7:05 am
Location: Bratislava, Slovakia
Real Name: Jano Svitok
Contact:

The idea is here for long time... when will the time be here I have no idea :D

Seriously though: the math behind this is quite simple with some edge cases (like are 4 points enough or do we need also the center point?)
Problem is that there are so many good ideas and so little time :)
OddSocks
Posts: 152
Joined: Tue Apr 14, 2015 11:33 am
Real Name: Tim Cullingworth

Could you use "Channel modifiers" on the x / y channels to do this?
pengumaniac
Posts: 75
Joined: Wed Jul 08, 2015 8:59 am
Location: FRH, Germany
Real Name: Manfred Flintstone

a) ma scancommander page 10ff http://rages.ch/download/scengl.pdf
store the 4 corners of the stage should be enough to get a fitted movement to bigger/smaller stages
Tilt 0 -> Border, Tilt -50...+50 (%) moves from outside to inside, aware: if central axis isn't same, the ranges of -50% differ from +50%

I didn*t try that in between 12 years usage, too laborious as we serve normal no tours with almost same set ups ...


b)"Right" mounting is benefical
to know a 90° move (btw: degree integer is enough to display in the pad, floats up to 4 decimals distract me more than useful) ,
we have to know beside the range, the "center rel. to base" of the fixture

e.g. our "good old" Futurelight MH-660s (rPan off/rTilt off):
Mounted it's display at side (upside down: display-> stageright),
Pan 128DMX gives a parallel pivot relative to stage front,
thereto 90° towards public at Tilt ~45DMX like this:
IMG_20150814_151155.jpg
Pan upscaling = turn cw on topview, the left ones move also to stagecenter

one of my ancestors had always his "fun" in programming the right Pan - he had same position at 33%,
while he was used to mount display -> backdrop (because he wasn't able to switch the display-LEDs off...)


Another, but bad example isPHW-250 (to complete register: Pan 630° / Tilt 265°):
"Design u.b.d.", really F...ing if you can't reverse Pan by controller:
rPan=Off -> pivot ~45° to its Base on Pan 128DMX & rPan=On has thereto a different angle!
(moreover Dimmer is inverse and only dimming at Shutter 32-63 DMX)
-> here an offset in Pan + some (~18?) DMX steps could save your head...

but time - Jano, Novinky 2013 says it all... :)
Gentoo | profile=desktop | LXDE | QLC+ qt4-build / openSUSE 13.2-x86_64 | KDE | QLC+ qt5-rpmbuild
The best way to predict the future is to invent it. (Alan Curtis Kay) - I'd like to emerge -avuND world, but there are no news in sync for stable updates
ninjasam
Posts: 8
Joined: Fri Jun 12, 2015 12:43 pm
Location: Lille, France
Real Name: Samuel Boudet

Hi,

I had the same idea and it seemed me a very important feature to replace follow spot.

I have thought about the mathematical problem to find the location and original orientation of a moving-head fixture using the pan/tilt and XY coordinates of at least 3 corners. This was not an easy problem (not linear) and there is no formal solution (would require to find the intersection point of 3 spindle tori). But after a week of headache I found a fast algorithm which always converge. Once we have the location and orientation of the fixture the XY<->pan/tilt conversion is a simple formula.

I am converting it in C++ and I am thinking to add it on QLC+, but it can be complicated for me since I do not know the organization of the QLC+ source code. So, Massimo or someone else could you give me a hand to make the interface?

I would imagine an interface like that:
- Have to be able to import an image of the room viewed from top, with the scene drawn on it (done with paint as example). This image would be put as background image of XY PAD of QLC+.
- on the fixture setting, we need to be able to manually set X Y Z location of the fixture as well as the 3 angles which defines its orientation for pan =0 tilt=0. Most of time, those coordinates are not easy to know, so there should be a calibrate button. Then a window should show the image of the scene. The user has to click on each 3 corners of the scene (or the room) and each time set the pan/tilt of the fixture. Then my algorithm can determine automatically the coordinates of the fixture.
- When there are several fixtures, they often have the same Z and the same orientation. We only have to determine X,Y which can be done by calibrating on only 1 corner.
- On the opposite, if there are more calibrating points, the fixture orientation/location can be more precise, but I do not think it is really required.
- On the virtual console XY pad. When we click on a point of the image of the room, knowing the information of the fixture, an automated conversion to pan/tilt of each feature can be done, thus enabling multiple fixtures to spot at a same point. Any EFX applied on this XY pad will also be corrected. Ideally, a controllable checkbox can be added to put the spot of the various fixtures parallel instead of converging. Then the XY point selected will be the middle of the multiple fixtures.

So I think I will have make a first version of the algorithm tomorrow, but I would be very grateful if some can give me a hand for the interface.
Thank you
User avatar
mcallegari
Posts: 4461
Joined: Sun Apr 12, 2015 9:09 am
Location: Italy
Real Name: Massimo Callegari
Contact:

This idea has been discussed a few times before and my answer is always the same: moving heads spatial tracking is possible only in a 3D environment, which is not there in QLC+ 4 and will never be.

However, it's indeed a feature for QLC+ 5, but the 3D part is one of the hardest topic to implement and it requires time. So I'm afraid you just need to be patient and everything will eventually go into place.
ninjasam
Posts: 8
Joined: Fri Jun 12, 2015 12:43 pm
Location: Lille, France
Real Name: Samuel Boudet

Thanks for your reply,

I do not know exactly what will be the 3D environnement but I think a simple overload of the 2D is enough which can be done with some minor change. The aim is only to spot a floor in 2D so indeed there will be the 3D coordinates of spotlight stored but it can be seen only as a calibration (just need to target 3 points on the floor with each moving head fixtures).

Maybe I can try to develop a first interface to convince you. I hope I 'll manage to :).
filipek.o
Posts: 57
Joined: Thu Nov 19, 2015 12:52 pm
Real Name: Filip

Hey is there any update with this problem ?:)
Post Reply