PDA

View Full Version : What is the best way to do camera mapping in LW?


ArneK
08-27-2005, 02:16 AM
I'm just wondering how you go about making proper camera mapping in Lightwave, as we all know - you can't model in Layout.

How do you match your camera to fit objects perfectly? The only tutorials I've found on camera mapping in Lightwave has been on boxes and doorways etc. Not very difficult to accomplish. How would you approach a more complex matte painting projection with more organic shapes etc?

Here (http://www.3dfluff.com/cameramapping/cameramappingtut.htm) is a great tutorial for C4D. How would you match those mountains in Lightwave..?

Mike RB
08-27-2005, 02:58 AM
one way is to load geometry that matches roughly, then turn it into a softbody, tweak it with the edit tool for cloth, then save transformed to get it back into modeler to finish the job. Sucks but works.

Mike

mav3rick
08-27-2005, 09:33 AM
OMG:) what a workround:)
i think 9 will have camera mapping option.. that s good stuff

EsHrA
08-27-2005, 12:42 PM
i believe Worley has a tutorial on camera mapping using one of his plugz.
http://www.worley.com/tutorials.html

mlon

ArneK
08-27-2005, 12:57 PM
one way is to load geometry that matches roughly, then turn it into a softbody, tweak it with the edit tool for cloth, then save transformed to get it back into modeler to finish the job. Sucks but works.

Mike

Just what I've experienced - camera mapping in LW is a pain in the butt. Once you want to map more organic shapes LW falls short... Hmm...

Mlon, thanks I know about that one. It still only handles box-shapes.

gerardo
08-31-2005, 10:11 AM
CameraMatch can match complex shapes too, but you need to choose well your reference points. If you want to map your photogrphs in an object, Sticky Front Projection is very useful for this, although most of the times LW Front Projection is enough. If you have a complex shape, and you want to model it from photographs (some kind of photogrametry process) is better to work with several photographs, taken from different positions to triangulate the model and use one LW camera by each photograph position (later if you want, you can use weightmaps to blend the textures). To adjust your object vertices maybe you want to try this too:
1. In Modeler copy and paste your mesh in the next layer
2. Kill the polys and convert the points into Points to polys
3. Make Skelegons
4. Cut and paste skelegons in the first layer. Save
5. In Layout convert skelegons into bones
Now you can adjust every vertex in your mesh as you desire :)



Gerardo

ArneK
08-31-2005, 11:40 AM
Gerardo, that's brilliant. Why didn't I think of that. Making a bunch of skelegons should solve some problems. "Skelegon modelling" in Layout :thumbsup:

softdistortion
08-31-2005, 12:37 PM
Wouldn't it be easier to use modeller to create geometry that fits your image, then use that in layout? :shrug:

PaZ
08-31-2005, 05:51 PM
Wouldn't it be easier to use modeller to create geometry that fits your image, then use that in layout? :shrug:

Just use the FREE plugin FI's CameraMap (or FI's Cmap).
you get a window where you can align manually points of a model on a BG image (you can zoom and pan in the window), then you clock "match" key. Done, nothing more than that... it has its limitations but i used it several times, it has no rivals when it comes to static Camera Matching.
Unfortunately, FI's site is down now, but maybe you'll find it with some search on Flay or other sites. I'm not sure i can post it here since redistributing software is not allowed.

Paolo Zambrini

gerardo
09-01-2005, 03:27 AM
ArneK, that trick is useful also to mimic the animation style of PuppetMaster (there are a tutorial in Newtek's website about it) :)

Hey! Paolo, thank you. That plugin is very handy (FI's plugins are amazing!)

Softdistortion, yes, if you know the exact dimensions/proportions/measures of the real object, then the only thing you need to worry about is to match your camera with the BG image and use FrontProjection to map the texture; but when the object is too complex for just to be photographed in the traditional way (a street for example) or is a more simple object but you don't have the appropriate views or measures or even worse, you only have the BG image as reference, is a little tricky to check your background in Layout, go to Modeler, adjust, come back to Layout... and so on. I think in those cases is better to model an approximation in Modeler and tune it in Layout.



Gerardo

softdistortion
09-01-2005, 02:26 PM
Softdistortion, yes, if you know the exact dimensions/proportions/measures of the real object, then the only thing you need to worry about is to match your camera with the BG image and use FrontProjection to map the texture; but when the object is too complex for just to be photographed in the traditional way (a street for example) or is a more simple object but you don't have the appropriate views or measures or even worse, you only have the BG image as reference, is a little tricky to check your background in Layout, go to Modeler, adjust, come back to Layout... and so on. I think in those cases is better to model an approximation in Modeler and tune it in Layout.

I get what you are saying Gerardo, but why wouldn't I still go into modeller and just UV whatever image I need for a background ? then I can size and position it either in modeler or Layout. I don't see the advantage of doing complex camera positioning and using bones and all the rest of the suggestions.
So in your example of a street that is too complex to photograph in a traditional method...how would it be usable as a projection if you can't UV it on a plane and cut in a bit of detail in modeler? ...or with the example of a more simple object but you don't have the appropriate views or measures... if you can't rebuild in Modeller how is it easier to project properly in Layout? You still have to basically lay it onto a surface right?

So far I've always had decent photos to work with, so maybe I just don't get it yet because of being spoiled. :D

gerardo
09-02-2005, 02:18 AM
I get what you are saying Gerardo, but why wouldn't I still go into modeller and just UV whatever image I need for a background ? then I can size and position it either in modeler or Layout. I don't see the advantage of doing complex camera positioning and using bones and all the rest of the suggestions.

The purpose of camera mapping is to map a photograph (taken from a specific camera angle, FOV, position and elevation) as a texture. So that this works, first we need an exact 3D representation of the real object to match it with the same real camera conditions. If you only have this photograph as reference to model your 3D object, UVmap this photo into a plane in Modeler isn't an accurate reference because we don't have a camera in Modeler to match angles, positions, lenses, etc. So what we do, is to make the object in Modeler with the proper shape but inexact in proportions (since an UVmap and Modeler perspective are insufficient to model an accurate shape in this case), then in Layout we match our 3D camera with our BGimage and we can adjust the object shape to its real proportions.

So in your example of a street that is too complex to photograph in a traditional method...how would it be usable as a projection if you can't UV it on a plane and cut in a bit of detail in modeler?

I said "traditional way" referring to the way of taking pictures to model in Modeler (top side front). The street's example is more complex than that (and I don't think the method proposed above is the best way to solve it), but basically the idea is the same, model the street shapes in Modeler and adjust the shapes in Layout using cameras that match the photographs views, commonly these views aren't front or side and less top view, but it would be better if these photographs triangulates the zone we are working.

...or with the example of a more simple object but you don't have the appropriate views or measures... if you can't rebuild in Modeller how is it easier to project properly in Layout? You still have to basically lay it onto a surface right?

Yes, as you suggest, we need to model the proper shape (but inaccurate) in Modeler, the Layout process is only to adjust the mesh vertexes if we haven't had the possibility to have the appropriate views or measures. Remember also we can match our 3D camera view with simpler items in the photograph or even with just one known space.

So far I've always had decent photos to work with, so maybe I just don't get it yet because of being spoiled. :D

Sure, if you have decent photos (front, side, top view), Modeler is more than enough :)



Gerardo

walfridson
09-02-2005, 01:23 PM
I use "nulls to points". Only works when tracking a shot though

softdistortion
09-02-2005, 04:50 PM
Thanks Gerardo! That makes sense although until I actually have to do it, I guess it's still a bit foggy to visualize it 100%. I've looked at some tuts on this subject but they seem to be nothing more than a complex way of doing what I am already used to.

toxyn
09-04-2005, 01:22 AM
If you have a least 2 pictures, you can use Realviz ImageModeler to create a 3D scene: http://www.realviz.com/products/im/index.php

borisgoreta2000
09-05-2005, 05:47 PM
I tweak the objects in Layout using EditFX. Add SoftFX dynamics to your ovject, set effect size of operator1 in operator tab to 0% so that SoftFX wont affect your object and click calculate for frames 0 to 1. Then go to EditFX tab and there you can model your object in Layout. Study options under EditFX and you will be able to model in the same manner as with the magnet tool in Modeler. Once you are done hit save transformed object and that's it.

ArneK
09-05-2005, 05:58 PM
Lotsa good suggestions guys. Thanx :)

CGTalk Moderation
09-05-2005, 05:58 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.