Best method? Overhead POV pullback from person on ground to outer space...
Greetings Cow'ers. I've got a gig doing a 10 sec. shot that starts from an overhead crane of a person, which I'm then going to pull back from until my POV is in space and can see the earth, sun, and the moon.
I'm having a difficult time making the transition from my 2D AE sequence to my 3D maya model. Currently, I'm just eyeballing the shape of the US as I try to soft transition from AE to Maya. Is there a more accurate work-flow? How would you approach this job? Here are some of the things that rattle around in my mind as I work to make sure I start with my best foot forward...
--I've got a nice smooth move from the plate through 3 digital globe images, until I can see most of the eastern part of the US. I did it by simply accurately scaling the comps and lining them up, then moving my camera on the z axis from aprox. 200 to something like 2,000,000.
--I've got a decent globe model in maya that I've built, but for memory/performance reasons, I've only been able to apply a texture that won't let me start much closer than seeing about 50% of the US. Should I be isolating the close-up area and applying a higher-res texture to the earth model? (I'm less proficient in Maya than AE).
--I've read a bit about importing maya camera data into AE, which seems like a great solution, but my initial move from the ground up is so large in scale, that I'm wondering if this is going to be a reasonable way to do it. After all, how in the world would I put my plate photography on a globe that is actually scaled accurately in maya? (Don't worry. I'll post this to the maya boards as well)
--Should I try to hide the transition from AE to maya in a high-altitude cloud? It seems like a good idea, but it's really a bit closer to the ground than where I was hoping to put that transition, largely because I'm not sure yet how to get such a hi-res map working well in maya.
--Then there's the curvature issue... Maybe use a lens effect to start to add some distortion to the end of the AE sequence before I hit maya?
--And lastly, should I just be build this whole thing in maya from the ground up?! Is there a way to use AE for the compositing of the different satellite images (digitalglobe.com), then import it into maya? Maybe projecting that movie sequence to an object, and not starting my maya cam move until I get to the end of the AE pre-rendered movie that's been projected on to the maya model?
Anyway, you can see my situation. I'm not lost without hope, but I'd love some input from some of you who have had to pull this kind of fx shot before. Of course, I'd like to "hide" my transitions as little as possible.
Thanks for your input. I'll post a link of the finished shot when I'm finished.
Well, I know you say you want to hide your transitions as little as possible, but I might seriously consider using a cloud layer to hide that transition from 2D to 3D.
That said, you can certainly use various image files of various amounts of detail to texture your globe. Here's a workflow that should be pretty straightforward: map your highest-detail texture onto the sphere to start. Then, in the 2D placement node, turn off repeat and drop the Coverage attributes. You should end up with your texture covering just a small section of the sphere, leaving the rest of it plain gray. Tweak Coverage and Translate Frame attributes to scale and position the image appropriately.
Now, in the 'Color Balance' section of the file texture node, map another file texture to the Default Color attribute. This should be your next-highest-detail photo. Again, prevent it from repeating, and scale and position it appropriately. Continue nesting images inside one another in this way until you have the whole globe textured.
If you want to improve the effect, use a Box Ramp texture instead of just a file texture node. By mapping the higher detail texture to the outside of the box pattern, as well as the Default Color, you can create a smooth transition between images.
I hope that's clear; my browser crashed so I had to retype and I'm not quite as patient the second time around! Let me know if you need more detailed instructions.
Thanks. That's exactly what I'm looking for. Do you think that method in maya is robust enough to handle the extreme scale all the way down to the point where my camera is focused on my actual footage? (overhead shot of a farmer in a field, showing an area of only about 15 feet across)
Or, when we get down to practicality of actually pulling it off, is it easier to use AE for some of the lower altitude shots where we're not yet seeing any distortion from earth curvature?
Also, should I set any sort of scale in Maya that actually tries to replicate the units of the real world (i.e. make my globe actually the size of earth), or is it better to just let it all be relative to eachother in a more miniature setup? I guess what I'm asking is whether or not the camera scale and earth model scale has an impact on the viewed output.
Thanks for the input,
[GregNeumayer] "Do you think that method in maya is robust enough to handle the extreme scale all the way down to the point where my camera is focused on my actual footage?"
I'm honestly not sure; you'll have to do some tests to find out. I suppose there's a chance you might run into the limits of texture placement precision at the extreme end.
[GregNeumayer] "Or, when we get down to practicality of actually pulling it off, is it easier to use AE for some of the lower altitude shots where we're not yet seeing any distortion from earth curvature?"
Well, as I said, it would probably make things a lot easier if you mask the transition with a cloud layer (blanking the screen for a moment as the camera passes through virtual clouds). The problem with trying to mix AE for the smaller scale and Maya for the larger scale will be preventing a noticeable jump in the smoothness of the motion of the camera; you want it pulling back in one smooth motion, right? You'd probably end up having to eyeball the first frame of the Maya camera's motion to match it to the last frame of the AE camera's motion... again, it's your call about whether or not the hassle outweighs the benefit.
[GregNeumayer] "I guess what I'm asking is whether or not the camera scale and earth model scale has an impact on the viewed output."
I don't think so, but I know this question has been asked in the past--check the archives and see if there's anything useful.
Thanks for the input on using the default color node. I've stacked 4 varying resolutions of my location, my top one being the closest up one to my location. I'd like to soften the edges, but I'm not sure how to implement the "box ramp" method you mentioned. I'm familiar with using a ramp, but your suggestion seems to suggest that all my satellite images would need to line up on centerpoint in order to use them as inputs for the box gradient. They don't line up that way. Each has it's own custom coverage and translation to make it fit correctly.
I had figured that I'd be able to easily employ some sort of black-white ramp as an alpha knockout for the outer edge of each overlapped image, but I can't seem to understand how I'd plug it in. Can I use a ramp to add transparency to the individual file textures? Would it somehow be part of the alpha gain and offset (which seem to have absolutely no effect on my image when adjusted).
The "easy" way would seem like I could just add transparency as an alpha in photoshop, then import with it, but I've never had much luck with imported alphas in Maya. (6.0 on Mac) (Is that what my seemingly dormant alpha gain and offset are for?)
Thanks for your help. Any solution that allows me to soften the transition of the file texture without re-building all my textures would work.
I've posted an image of my hypershade at:
Just to update, I've found a solution that appears to be working using the layer shader. I duplicated my 4 texture placement nodes, since they are all unique and custom, then replaced each of their inputs with a simple box texture jpg as the file. Then each layer in the shader has it's own pair of color w/ placement and transparency w/ placement nodes.
Here's another screen shot of the hypershader.
I'd appreciate any input if there's a better way. I'm a neophyte on using the hypershader effeciently.
Well, I guess I spoke too soon. For some reason, I can't seem to plug in a simple greyscale jpg into the transparency slot of a layer in the layer shader. Still no luck feathering the edges of those textures.
Hi, Greg--sorry to take so long getting back to you.
Your fourfold hierarchy of file images looks good to me. The box ramp strategy I was talking about will let you feather those edges with a similar setup. Here's how to go about using it:
Instead of a file texture at each stage, use a ramp texture. Make it a box ramp, blending between one colour at the outer edge to another colour just a short distance inside. Now, map the narrowest image to the central colour, and map another ramp to both the outer colour and the default colour. Repeat for as many stages as you have images. Each ramp will have its own placement node so you can move them wherever you like independently.
Here's an example:
See how each image blends smoothly into the next, though I am free to move them off-centre? The narrowest image, in this case the bottom one, is only connected to the central colour of the innermost ramp. The other images are connected to both the outer colour (this shows up as '.colorEntryList.color' as you can see) and the default colour of the ramps.
If you have put a lot of work into positioning your image textures just so, you can create your new shading network alongside the old, and just connect the old placement nodes to the new textures--that way you don't have to redo the work.
Hope that helps. Let me know if you have further questions about how to set it up.
Thanks a ton. I thought it was a lost cause at 1 am last night. After trying your idea, getting it completely upside down in the connections, drafting the question to you, getting an idea about why all I see is default colors with thin strips of my file, trying it again and deleting my paragraph to you, then checking my backup software to make sure it's really backing up my files every night, I think I have it! Whew.
Thanks so much for your help. Sometimes the best thing is someone reassuring you that there IS a solution. I'll post a sample when I'm all done with the shot. I still have some minor compositing to do with the first 100,000 feet in After Effects and a bit of camera finessing, but hey, I've got something to show the client tomorrow, which was the important part.
I don't think this project should touch Maya. The CC Sphere effect in After Effects creates very nice spheres from flat art, with camera and lighting controls. As you probably know already, NASA has some great photographic material of the globe and they provide in very high res. They also have cloud layers you can put over the earth and animate (with alpha channels already in it.) I did this in a recent project (matching up the layers like you are doing) and it worked out well.
New York-based Editor-Compositor