Hello vvvvolks.
I inherited a projection mapping project gone wrong, they tried to do a 3d workflow with 2d technology. Now the customer is thinking about taking a second approach I’d be glad to hear your opinions.
They have 5 projectors in a cave-like setup with several objects in front of the walls. Unfortunately the projectors are not very solid and tend to move a little bit from time to time. Also the objects tend to move a little bit (we’re talking about millimeters) when they are cleaned.
I need a solution that can do:
3d projection mapping (has 3d model, uv-mapping)
multi-projector playout (boygrouping)
3d softedge feathering, softedge areas are also located somewhere in space
3d brightness correction, areas closer to the projector have to be dimmed
3d black-level adjustment, areas hit only by the light of one projetor have to be brightened to match overlapped areas and actually the black level has to be brightness corrected everywhere
automatically camera calibration of projector parameters with multiple cameras or a ptz cam
automatically calibration of projector brightness, gamma and color unifomity
automated mesh deform to match the real world positions of the objects
There are also several LCDs in the room, so a nice bonus would be to match the LCDs and projection and to match all of this with the ambient light.
I know most of this can be done with disguise but at a price tag where my customer would kick my butt thru the door. All of this can also be done with vvvv (because everything can be done with vvvv) but I guess nobody has done all of this before, right?
If there are any takers here, that’s your opportunity to offer your support. If you know of any other solution please share your vvvvisdom.
and I guess all of this has to be done until yesterday?
unfortunately there’s no complete package about this in vvvv, and there are couple of features on that list which to my knowledge has not publicly done inside vvvv yet.
No, I’ll have to do a proposal and then the customer will decide wether to do something and what to do. The timeframe is not set yet but realistic time will be part of my proposal.
I also think an external tool for the calibration part will be the best solution, but it has to spit out geometry data as well. Typically this is a point-cloud and this point-cloud needs to be used to deform the mesh.
My own processing tool died together with the Barco projectors I wrote it for because it generated the warping data directly for the Barcos. Today there are much better open source libraries availabe. Computer vision has become so popular.
I can do align stuff for your 3D scene that is not perfectly fits the projection. Even setup stuff to create augmented mapping from scratch (calibrate by markers> build scene > aligns uv’s to your scene or video file you have) using only my set of custom build tools and UI for mapping with cost much under 20k euro. Also includes training period for a technician to realign stuff solo.
Cost would be transfer and some salary to cover the expenses, let’s say about 5k.
Also can synchronize stuff across network so it can envolve unlimited amount of pc’s
Production time is about 2-3 clear days (in case of fatal errors in pre production stage, excluding day of arrival (don’t want to work without a nap)).
Also for about 10k it’s possible to write in c# and vvvv fully custom special setup and alignment system that is made solo for that project and have all the property exposed to the end user without any additional stress (custom mapping primitives, content update system and ui) this would take from a week up to a month.
Guarantee is about 12 years of mapping stuff with vvvv…
Haha, microdee has been concerned about the time and you offer to do it in two days without even knowing the complexity. Ok, I admit it’s not that complex, but I think two days are not enough. What do you mean with calibrate by markers? The semi-automatic approach where you move marker lines to vertices? What I need is a full automatic system usually using structured light to generate all camera and projector parameters and the geometry. And that’s only 1/3 of the job. Then I need the point-clouds to be processed and aligned and simplified and so on and make them modify the original mesh, that’s another 1/3. Finally I need all the spacial stitching and edge feathering and calibaration of color, brightnes and so on…
Well augmented approach is to first a setup a 3D calibrated space across projectors and to draw geometry for actual projectors after projectors aligned between each other. Two full days is enough when you have a preproduction phase complete. If you want to do that from scratch it would take about a week of pre production time. And all that only if you have an ready made footage you want to run, if you want it custom you have to provide all the details like time and brief, photos, renders, schemes on what you expect to get on my email antokhio(at)gmail.com and i’ll take a look and give you more detailed info, like a schedule.
About stitching, i can teach you to make it pixel perfect. About color correction and brightness across beamers sort you have to do that by hand bu that’s also not a rocket since so…
I only say I can do it cheaper then a an automated layout system, and more users friendly and customizable, but it would have an GUI elements that you have to tweak.
You can find me on skype or WhatsApp, let’s chat whatsoever… maybe my long mapping practice can do any good to you…
antokhio, I forgot to mention that this is a permanent installation. You cannot do it cheaper than an automated system. The system should be recalibrated every day to compensate for small misalignments and lamp ageing.
