Projection mapping and calibration test - help needed

Hello everybody,
today i made my first try in mapping a box with vvvv, after reading all the threads and documents about the projector node i did all the necessary measures (damn that was hard!!)… i think the measurement i took were not very precise because:
1) how can i know if the projector is perfectly parallel to the projection surface (if it isn’t i will calculate wrong the screen width and the shift x and y!!)??
2) if (as i was doing) the projector is on a surface that is’t parallel to the ground (or it is not perfectly straight) that will cause other measures to come wrong too…

I took all the measures with a laser meter…some photos of the setup below (in this moment i was projecting a grid to view the shift x and shift y)

Now arrives the fun part… i inserted all the parameters in my vvvv patch (using meters as my unit of measure) and i put a box on the origins of the xyz axis with the same dimensions of my real box i wanted to project onto…
I did all the measures as you can see from the picture below and recreated my real scene in vvvv…

also this time how can i know if the projector is perfectly parallel to the xyz axis? i put roll, pitch and yaw to zero because i tryed to maintain the projector “flat” on each axis…but is this really true? how could i verify this?
Another question…howcan i measure the inclination on each axis if i would rotate the projector for example along the y axis?
When i put the renderer full screen i noticed that only the right side of the box was projected correctly, why do you think this happened?

i had to tweak some settings as the translate y, translate x and scale x of the projector to make it fit…and this was just a single box…what would happen if i had a more complicated geometry? i don’t think that some tweaks like this would solve the problem…
Here is an image of the image projected after the tweaks…

I really hope that you can help answering my questions, i think i understood pretty well how the process and the measures work in theory but i think i’m mistaking something…

I’ve attached the simple test patch i did if it may be of some help!

Projector box test (9.9 kB)

According to your pictures you used a spirit level and I´d say that´s the way to check if a projector is level/plumb.

Did you maybe use the (physical) lenseshift of your projector?
That would of course distort your measurements.

Have you activated the depthbuffer of the renderer?

thank you for the answer
First of all i can’t really understand what you mean by phisical lenseshift, are you referring to the measures of the shift x and y pins? if this is right i calculated them phisically by projecting the cross you can see in one of the first images above, i measured the distance between the red spot (that represents the center of the lens, more or less 'cause the laser is on top of the projector and not exactly on the lens, but i did take this in consideration) and the center of the cross (that represented the center of the image). I think i have to do this measure, this crappy projector has the lens that can me moved with a “joystick” and this causes the lenseshift to vary so i don’t think i will find that value on a manual…do you think i should not measure this by hand? i should find that value written on the manual?

yes i think i’ve activated the depthbuffer and set it to D16 for fullscreen and res to 1280x1024 (maximum resolution of my projector is 800x600 but i don’t think this will make the difference if the proportions remain the same…)

yes i used the spirit level to level the projector but this will just tell me if it is rotated along the z and x axis (referring to the above photo with the axis drawn) but how can i measure the rotation on the y axis? this should be very important to know because it tells me if the projector is parallel to the projection surface isn’t it?

one thing I always check when working with a projector is use a test pattern fullscreen to double check that you’re projector is infact projecting everything that it should be. I’ve seen projectors with things like auto resize and miss edges of projections. Also, VGA can be un predictable with the screen jumping around every time you go full screen. Best use DVI.

But who knows what has gone wrong. I’ve done a few setup using this technique and every time it comes down to really doing your best to get everything measured properly but then the real art tweaking the setting to get the model to match up. Through experience you really get a feel for what adjustments to make in order to get ya geometry to line up. Things like really visualising the view of the object from the projector position and thinking, ‘I’d be seeing that surface or not seeing that one, I’m too high’. I did one recently on a building and we had really rough distances, rotations, we had a really great 3D model of the building, we knew the ratio of the lens, size of image on lens, lens depth, and the Y and X shift of the projectors, and just tweaked away and nailed it in about 20mins for each projector. Sometimes all the measuring in the world won’t get it lined up, but it ceraintly a good place to start. Having a perfect 3D model is crucial. But then you do have the point editor for meshes that even let you tweak that. I’ve done that before because the model was out and it worked fine.

Why are you trying to project the box in the middle of the projection area? why right in the corner?

Also, i know it sounds a little silly but try doing this on more complex geometry, i found it helps get a better feel for what the projector node is doing. Try two small boxes like 30cm apart. trying to line up two boxes gives you more to play with that trying to line up edges of a box.

But yes

thanks for the reply,
this night i did the projector calibration again, i tried to be more accurate and precise… i used a pattern as xd_nitro said and infact the projector was showing me 1,2 pixels less than it should be…
There’s no reason for the position of the box in my first test, i wanted to do a quick try and i positioned it there, in the origins of the xyz axis and then i just moved the projector in the side not caring about the fact that the beam did not cover the entire box…
By the way tomorrow i will try another projection, as you said i will put in the same setup two different boxes, i’m not very confident with the point editor for meshes (i had a look at the example on the girl power folder but did not understood much of it…i will look more deeply onto it and i’ll try to get the best out of it :) )

I will update this post tomorrow after my second test setup to show the my results!

I have never actually tried this, but it came to my mind that:

To measure if the projector is parallel to the screen you’re projecting on take a rule and measure the sides of your projection. Opposite sites should have equal length.
If not all of your projection is on the same surface (like in your case above) render two small lines of equal pixel length on each side of the screen and measure them. In fact the lines could be anywhere, they don’t have to be right at the edge - but the longer the distance between them, the greater your accuracy when measuring the length difference will be.

thank you Grandchild for you answer, on the calibration i did yesterday i used the method you said to verify if the projector was parallel to the projection surface!
i did not try to do any other projection yet cause i’ve been practicing a bit on the meshEditor, btw i hope to do it today or tomorrow!