Camera mapping?
Roygee
Posts: 2,247
Every once in a while there is a question about whether Carrara can do camera mapping and the answer is invariably no, no way, far too advanced.
This seemed like some magical method until I watched a video demonstration on doing it in Blender.
Picture this - a still image of an abandoned wreck of a factory - big room, with a pillar in the middle ground obscuring part of the floor and back wall. What the demonstrator did was to position the picture in the camera view, insert a cube and match it to the dimensions of the room. Then deleted the face facing the camera. Then separated the faces so that he had four panels - ground, left wall, right wall, back wall and roof.
Took the picture into Photoshop and made separate pics of the ground, etc. Where the pillar obscured the ground and back walls, he cloned over the pillar. Made a panel for the pillar and a matching pic in Photoshop for that.
Then used a "projection from view" to map the various pictures onto the panels and did a fly-through to demonstrate.
My question to the wise men is how is that different to anything that can be done in Carrara by flat-mapping?
Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
Interesting subject. Was thinking about the same idea a week ago.
Matte Paintings these days are done with Camera Projections either in a dedicated 3D software (Maya, C4D) or in Nuke.
Dylan cole (Matte Painter for The Lord of the Rings) uses Cinema 4D as it has an in-built Camera Projection tool called Projection Man.
This is a technique I have been researching and want to use it for my work.
In Nuke one can place images onto simple 3D geometry and do all the compositing.
in a 3D software one can model more complex geometry to project the image onto.
Camera Mapping is useful if you are animating the camera to create the illusion of moving through a scene.
I believe Shader Ops 2 plugin is useful for Camera Mapping in Carrara. I have not tried it.
http://www.digitalcarversguild.com/plugin.php?ProductId=20
I don't know if these qualify, but in my Black Hole video, I placed a spherical camera on the ship where I wanted the scene perspective to be and rendered with an alpha channel.
I then applied the render to a sphere. The areas of the render without geometry were clear due to the alpha, and allowed me to composite in a starfield so that I could get the desired parallax when I moved the camera. The point in the video where I used this method was all the scenes with the astronauts.
https://www.youtube.com/watch?v=iS-sNje4k0o
The other video also uses a spherical render of a night scene, and two objects. The spaceship and the building it approaches. That's it.
https://www.youtube.com/watch?v=QN6E8ji3oHA
I think you're basically asking if you can convert a 2D image into a 3D scene, correct?
The answer is always yes. :) :)
But of course you are limited to what 2D image you're starting with. You basically take the components of the 2D image and apply them as a texture to a 3D object, usually after applying some sort of un-distortion to account for the perspective in the original 2D image. Also, lighting and shadows in the original are a factor, so often you're very limited in the motion you can have in your 3D scene. So by necessity it is generally a limited procedure. Usually you're better off taking your own photos so you can control light and shadow and perspective for your textures.
Any 3D software on the planet can do this, it's just a matter of how much work you have to do to parse the 2D scene into textures for your 3D objects.
Some software does the most difficult part, what I call the "un-distortion", semi-automatically. But no software can generate new faces that didn't appear in the original 2D image, any better than you can just doing a copy/paste. And of course lighting and shadows can throw a big monkey wrench into your process.
It's a cool idea, but IMO you're much better off getting out your camera and taking a walk and make your own textures.
Yes, I know the process of mapping...let's forget for a moment whether this is good, bad or indifferent, or whether I want to do it or not; the question was how is the process of "camera mapping", which Carrara doesn't have, different to the "usual" method of mapping an image onto a plane, which Carrara does have?
Does camera mapping somehow correct for perspective? Take the picture of the factory, for instance - the side walls, roof and floor have perspective. If it were an really long room, they would appear to converge to a vanishing point. When the fly-through is done and the camera rotates to point directly at a wall, at a point some distance from where the picture was taken, it is as if you were in the building and looking directly at the wall.
In order to get the same effect by using the "usual" method of mapping, I would have to take dozens of pictures directly facing the walls at different points - which would not be possible without using scaffolding.
Yeah, but of course it depends on the software. In some you click the 4 corners, say, of a wall, and the software automatically recognizes the perspective and "un-distorts" the image and applies it to the plane. Heck, even Photoshop has something along those lines with the Vanishing Point feature. You tell it the corners of a wall, for example, and it maintains that perspective in future edits to that area.
Some interesting examples of Camera Mapping.
https://vimeo.com/50203339
I like the comments about the video... :) :)
"WOW, AWESOME !!! INCREDIBLE WORK !!!! SHOW US A TUTORIAL !!! " :) :) :) :)
Yes, the video was beautiful. But what people don't realize is that what is beautiful is mostly the gorgeous photos that are the basis for this. The motion "through" the photos, as well as the rendering, while well done, were really pretty simple and nothing really spectacular. With gorgeous, hi-res photos like that, it's not really what the software can "do" that's important.
I think people tend to assume he generated those renders from scratch or something, which certainly isn't the case. You're basically looking at moving 2D photos. But I can see how a video like that would get everyone excited about doing camera mapping... "OOoooo, I want a cool animation like THAT !! " :) :) :)
Thanks for bringing this up Roy, looks very interesting.
I would say its about the whole process of bringing an idea from a thought to visual reality through the use of Photography, 3D software and 2D (Photoshop).
Learning the different software and the process of applying the techniques takes years to learn.
Both artistic talent and technical ability come into play to create something beautiful.
ShaderOps2 (DCG) has a camera mapping ability, though I'm not sure if it functions the same way as what was mentioned in Blender.
EDIT, Woops, I see scottidog2 already mentioned ShaderOps2 in the first reply.
This is really a great technique for bringing yet another dimension to our obsession with fakery, trickery, smoke and mirrors that is 3D graphics :)
Glad I investigated and thank you all for enlightening me.
Not anywhere near the level of the Vimeo link, that's some nice work, but here's a sample of camera mapping in Carrara.
Carrara camera mapping
1. It started with just a simple shot of the raisins on the counter.
2. I matched the Carrara camera focal length to the focal length of the lens I used.
3. I projected it against a plane and vertex modeler box.
4. I started the box with the rough dimensions of the raisin box, but then tweaked it a bit in the modeler to align with the image.
5. I removed the lighting from the scene and dropped the projected image into the glow channel (lighting is baked in).
6. Second camera moves around the box.
Regards,
Camera Mapping for Game of Thrones. Very detailed video.
https://vimeo.com/92368236
Or you could have gone online and downloaded one of 3.7 gazillion images of a SunMaid raisin box and applied that to your cube... :) :)
http://www.collectingcandy.com/wordpress/wp-content/uploads/2012/08/CC_Sun-Maid-Natural-California-Raisins-box-2007.jpg
I'm sure what you did is just a proof of concept, but I'm just trying to put things in a bit of perspective. Camera mapping, like I said, is a very limiting and limited procedure, and often there are much better alternatives.
But obviously it does have its uses, especially in professional applications. It's a lot cheaper to hire a photographer to take some photos than an entire film crew....if you can get away with it :) :) :)
I especially like the part where he talks about manually modelling the guy's head for projection mapping. He also says he could have gone on Turbosquid to download a free head model.... :) :)
Now, if anyone asks if you can do camera mapping in Carrara, here's a great example of how Carrara has a huge benefit over some other apps...
GEEZ GUY, JUST GET AN M4, MORPH IT UNTIL IT'S CLOSE TO YOUR SUBJECT, AND SAVE AS AN OBJ !!!! :) :) :)
That would save him a lot of time, rather than modelling it himself....
JoeMamma2000, thats exactly what I was thinking.
Did you use your plugin at all for this, or just as described?
Joe - those guys could have taken a helicopter ride to the top of el Capitan, but they chose to climb it. Sometimes it's great to do something just for the hell of it :)
True, but I was just making folks aware that a helicopter ride is an option, in case they weren't aware. And to put things in perspective for people who might not have considered the real limitations and benefits of the various alternatives. Just some more information. Which is good, right? :) :)
There are advantages to Camera Mapping. Because you are projecting the textures onto geometry, you can then use alphas to play around with transparency.
You can set material properties such as reflection.
Another plus is you can do lighting and particles as they interect with this geometry.
The possibilities are exciting.
What I was thinking is it could be used to make a scene with something that no longer existed except in photographs - such as historic buildings and such.
Joe - you are right - options are always good:)
Did you use your plugin at all for this, or just as described?
Used the plug-in for
3. I projected it against a plane and vertex modeler box.
Regards,
Hya Eric,
I haven't got my head around this one yet.
Could you use thuis to project (eg) a checker shader onto an object in a scene and the shader would obey the uv coords?
I definetly need to explore ShaderOPs 2 a lot further than I've done already. :)
Yes, I have plugins that, though I have read the destructions more than I would like, I have no idea what they do, why I should use them, or how to get them to work. I played with the trial of this today. It works wonderfully. It's a projector in the old fashioned sense of the word. It wasn't quite what mi wanted, but still terrific.
I had no idea this was called Camera mapping but it's basically what masking in Hitfilm does or shadowcatcher using objects
you would just need to split your photo up into background and foreground in something like Gimp and use the clone brush on the bits you removed, its really more of a 2D compositing thing using layers.
You got it. It projects a UV space across the camera's field of view. In it's more traditional use, that would paint a photo across your 3d scene, but you could project sort of like a light and a gel.
Regards,
Thanks Wendy and Eric for the prompt replies. Yes, I understand now! Cheers :)