|
|
Projection Mapping workflow questions I've tried to get started on a couple projection mapping projects so far but get stuck at various points. I haven't come upon a wealth of resources for mapping (besides regular quad shapes) and how some of the higher end stuff gets made. I was wondering if people could share their workflow on things they have worked on in the past. One of my past (or ongoing projects) involves projecting on a small object that has an object that is easily cobbled together in cinema 4D but relatively complex when projected back out. I've done many manner of measurements to try and correct film offset/field of view, but I can't quite set it exactly right (this becomes even more difficult in QC with a bit less control over a camera…but I wouldn't necessarily be using the model in QC). My current project is to try and map onto another small object that is fairly unusual. It is a buckyball shape made out of posterboard (picture below). Working with hexagons/pentagons has it's own challenges. I've mostly been working with the 1024 polygon mask and or the mesh creator for masks and sticking them in RII's to apply masks to sprites that have my moving content. Once I get past just a few of these and apply videos, everything slows to a crawl (which i assume is using too many RII's (8 or so). Perspective warping also presents its own share of challenges. For example..if I made an movie where the outside of a front facing hexagon gets a traced outline, it would be be nearly impossible to correct the perspective to fit inside the mask or onto the physical surface. Ideally I would like to control each face individually in real time (would eventually require 3 projectors) but I'm not sure if my computer could keep up. It seems like one way to work things is to make masks on the physical object, then render stuff in after effects and such, but this doesn't lend itself to easy experimentation…if I have to spend several hours correcting and re-rendering after a minor projector bump I feel like I'm doing something really wrong. Maybe make a mask of everything in one comp, copy that image into another then do perspective warps and image changes. So I guess my questions lie in the areas of masking onto objects that are not based on a 3d model, and how correct 3d models of an object. Any additional tools or resources used besides QC/Kineme would be good to know too. I'd be happy to post a stripped down comp of my buckyball experimenting if that helps.
|
using an alpha'ed image (or video) and 1024_Perspective it just works.
It totally makes sense that it could work w/ _1024 perspective, like franz posts.
Also, you really do have total control of view in QC - between the gl tools matrix patches, and similar NI stuff... and stuff like _1024, or CI, or GLSL, there are many ways to warp textures or entire scenes. The Noise Industries OpenGL Create Matrix / Set Matrix will warp the entire scene. Toneburst also made a macro, available in one of his "clips" qtz's that creates a matrix output, that could be interfaced with the the gl tools matrix environments or ni matrix stuff.
The way I've approached stuff like this is to use kineme gl structure renderers (or separate renderers) that have UV controls for texture, then to have something going that does an overall warp.
Thanks...for some reason I hadn't tried the _1024 Perspective yet, even though it's in my library. Does that work somehow differently than the built-in perspective transform and memo's 4 corner pin? I feel like those weren't quite moving things the way I expected them to. Is there something other than the blend modes that I should be using?
I'm also wondering about the best way to move elements within the masked area. I just made a simple shape in illustrator and dropped that in and used blend with mask and set that as my image. Using image texturing properties ends up moving both in an unusual fashion, and regular image transform on the non-mask image ends up moving both as well and takes a while to dial it in completely. Maybe I'm just not setting my anchor correctly. Is there a better tool I should be using? I'm also curious about doing something where a larger image would appear across 3 or more of the surfaces...not sure if I should be parsing it up or if theres a more intelligent way to do it.
I'll have to look again at warping an entire scene...I think having angle of view and camera position in terms of degrees and relative units in C4d was a bit more intuitive for me at first..especially when trying to relate it to a projector and an object in the real world. I found I could move sliders around for hours and not get it close to what I needed, but measuring usually helped to get it closer.
Mastering texturing and UV's is one of my next big hurdles...
Still struggling with the proper tool for moving textures within the _1024 perspective tool, but another question as popped up for me.
I've been trying to figure out an elegant solution for switching between using the interaction handles to move the shapes around and how to temporarily store the points (so that they don't immediately go away when I stop the render).
I'm guessing some kind of javascript switch would be the best solution as opposed to having a multiplexer for each of the 8 X and Y's. This would probably be best for storing the structures to a file of the entire set of shapes too, but I'm not quite there yet.
I have attached a comp which may help you with saving and loading structures, you will need Kineme file tools plugin installed. This was designed for 10.5 so interaction is handled differently.
Three interactive _1024 perspective quads which can be edited saved and loaded. Instructions included on a note.
This qtz was originally created by Stammhirn and rejigged with help from Cybero. Hope it is useful.
for moving textures, put a queue patch to your texture and get the first indexed pic.
Oh nice..that is pretty handy looking. I'll try and unpack it and see if I can use some of the pieces in my own comp. Thanks!
I don't think I follow what you mean. My textures won't really be moving besides being video sources. Once I get them in position it should stay static. Also..the way I'm combining the hexagon/pentagon mask and the video source is with a blend with mask.
I've tried putting the image texturing properties before the blend thinking it would just move the video but it moves both, so theres somethinng going on with the render chain that I don't understand yet.
Getting there with _1024 perspective and some cleaned up add-ons...still not sure if im doing it the most efficient way, but with a single source it seems to be running just fine...got good point storage as well
Still trying to figure out the best way to move the image that is under the mask, but having it within an RII seems to keep the mask still while i move the image underneath
Next step for me is to try and figure out the trick with 3d modeling physical objects to exact proportions and tackle texturing
more pictures here: http://www.flickr.com/photos/56993427@N04/
thanks to everyone for your plugins and help!