In a nutshell, Livepano has been designed to work with 2 MANDATORY inputs: a video and an Autopano project used for stitching the background. Background and video are supposed to be shot with the same lens in the exact same conditions.
That´s what i said before.
In your case, you have neither the stitched panorama, nor the video shot in same conditions. So I really wonder why you want to use Livepano for this project. This is nonsense.
True - but it´s a question whether that is really impossible. Just give it a closer look:
You see: Alex produces a panorama using a 3D application. I don´t know whether you ever did such a thing?
Let´s see the basics: when you generate a virtual qtvr camera for shooting the virtual scene in Cinema4D or Max or whatever, this camera basically behaves exactly
like a real camera on a NPP-head! The only diffgerence is: it doesn´t do single images but renders an equirectangular straight. But we can make cubefaces from it as usual.
Each time you use a camera in a 3D app you set a focal-lens and a film/sensor size. That´s what you also do using a real camera!
So my idea is: doing that you might use a real camera providing the very same sensor size and focal-length - resp. you set the VR-camera to the very same values the real camera has.
The point is: you need to shoot the video using the absolute identical
settings as the virtual camera: lens (doable), size (doable), perspective (doable).
Whether the video is keyed or not i guess doesn´t matter as long as it provides the same outline-size!
My time is somewhat limited theses days - but i´ll give it a try again . .
What we need to pin down: is my thinking correct? Ist the basic thing really the format-SIZE, the LENS and the perspective which need to be identical? Nothing else?
Because: THAT is possible to achieve using a 3D application! And given this is correct it must be possible to do what Alex asked for - and what i experimented with some time ago.
But I can try to suggest 2 options to help you:
- you really really want to use Livepano: try to generate two spherical images, with a 10° shift e.g. Feed Panotour with them, and extract cube faces for both. There you have 12 planar pictures of the same place, with some overlap. Use the 12 pictures as Autopano input to stitch the panorama. Now you should have a panorama stitched with Autopano (project file + rendered image) and a video that you can integrate following my previous explanations.
- ok Livepano is a very cool tool, but you don't really need it there: center your panorama where you want to merge your video. Extract cube faces, keep centered one and merge your video with this. Then in Panotour, create a video hotspot with the merged video that will be easily positionned.
Yann - you made my day (and i guess you made also Alex´day
) for giving such a great input
Thank you for your really valuable words!