Hi!
I have a problem with the Skybox configuration.
I ‘ve followed your guides and everything was fine. But now I am trying to do the same thing with a camera with movement (vertical dolly).
The error is in the Skybox Creator config, when I change the Camera from “Cinema 4D Camera” to “Comp Camera” this one doesn’t moves anymore.
What am I doing wrong? Is there any way to do the same process with a moving camera?
I use a demo version here, freshly downloaded from Mettle. Ae is CC2017. I hope the water marks are not in the way.
The camera was animated with the Parent Null in r18. All steps are done as suggested in the tutorial “… Ae with Skybox”.
To my understanding, any VR cube needs to have the camera in the absolute center point, otherwise it would be distorted. (For any later advanced changes, keep the C4D camera data available.)
The parenting of the Sky Box camera is key. The “Transform Data” in Ae will show no change, as it is parented. The image should change, if the context is not moving with the camera.
If this is not solving your problem, perhaps check with the support of Mettle directly, so far my impression is, that they are more than motivated to make things work.
It’s more or less working now. Yesterday I attached the camera to a spline so the the position was not keyed (red). Now I’ve keyed the values of the position of the camera (based on the 0 to 100% spline position).
The problem now is that the lines are well drawed at the begining but with the movement of the camera becomes erased (I attach the pictures), and looks terrible, like with a white noise. Any idea of why is this happening?
It seems that works well in Cinema 4D. If I do a preview in C4D I cannot see the white noise I posted some hours ago. Can this be a modification made by After Effects & Skybox Creator? Yesterday’s problem is solved, camera movement works fine, but the final tiff sequence looks noisy and with some kind of flickering.
C4D
The C4D file was set to Anti Alias> Geometry. It needs a setting much better than that. Best 8x8; 16x16 perhaps even. Eventually choose another Filter even, e.g., Sinc.
AeCC
The equirectangular projection stretches the poles a lot. Extremely said, nearly no pixel becomes stretched to the total width of the file, so the poles will look always weird. If projected on a sphere, things will look better.
Ae Screen resolution/Sample Factor, needs to be on Full for a proper review (I guess everyone knows that of course, but I have to mention it, as other parts here will be mentioned, as this is a forum for all levels)
The CINEWARE render has to be set up to the best possible, Render Final.
Is the first time I create stuff for VR and I am a little bit confused.
What do you want to mean by ‘Projecting on a sphere’? If I export the output as a Cube-Map, how can I visualize it in VR then? For now I just used the 360.Video.Metadata.Tool with an Equirectangular format vid.
Since you showed me an image that was based on an equirectangular projection, I went along with the terminology. A equirectangular, or a 360º*180º image, once projected on a spherical (Texture Tag> Projection: Spherical) will represent the view from the camera position in the center of that object. I write here object, as long as the viewer stays in the center of that object, it doesn’t matter if that object is a sphere or something else, e.g. a cube; Which brings us to your second question.
You ask about Cube Map. Well, in the moment you use a cube and the equirectangular map (projected spherical) you have six flat projections as an result. This can be baked out, given a proper UV(W) map. The only difference is “at the moment” given by sorting these six flat planes inside an image.
I solved part of the problem. As you said changing the render parameters + Encoding parameters the white noise has disappeared. But now (problems seems to never end xD) when I do a preview with my VR glasses (Samsung Gear VR) I can see some kind of blured pinch on the base and the upper side of the 360 video. The strange thing is that before I click on play (in my VR device) the environment looks perfect. But when the video starts it appears this kind of deformations.
The last problem that you describe is out of my scope of experience, I do not own a Samsung Gear VR device. However, it sounds like an often seen problem with equirectagulars. A phenomenon that isn’t available with cube mapping, but they show sometimes different problems (edges). You said above that you need to use cube maps, and since I’m not familiar with Samsungs tech, perhaps if a cube map is the input, the algorithm internally might use one. Again, I might be completely off here with my guess. The best way to figure this out might be here: https://samsungvr.com
I checked some of their footage on a normal screen, and the poles seems to be fine.
Problems never seems to end. Well, after a while this is called experience and helps you to be faster. You know that already, but just to remind you that everything has a positive side.
I would call this a compression problem. Since the structure of the compression method in mp4. It looks like that the max data rate was too low and the it didn’t had a key frame rate that was able to keep up. In other words, the turn down of the camera required too much data at once. When it hits the next full written frame, the I frame, it “pops-out” clear. Some algorithm try to find moving patterns even, and predict the next B or P frame information, then it gets even wilder. You might have seen it on streaming media, or DVDs, the background seems to move differently.
Any chance we could get an addendum to this series?
Now that Mettle is fully incorporated in to AR with CC18 the UI, process, etc has changed enough that I’m having a hard time re-engineering how this should work.
Starting at the point where we go to AE it stops making sense.
Certainly a good idea. If you have additional specific targets, which were not addressed in the series so far, please use the Tutorial Suggestions Forum
to get your ideas into the mix.
-