I am trying render a stereoscopic animation, but ran into problems with the compositing nodes…
I have two images viewed from slightly different angles (see lower down). I have the following very simple node setup (the actual one is more complex, but this one focuses on the problem):
These are fine, but when I add them with the ADD node and render them I only get one middle image, consisting of two halves of the two images. Even the ADD node’s output looks incorrect. NOTE in the node setup image (top) that the render size is double-width: 1280X480.
For whoever runs into this problem in future: one needs a (reference) picture of the eventual rendered size. So, I rendered a blank (with alpha) image of correct output size (1280X480) - and added that to the input images. That way the input images are scaled to the correct size.
Solution of nodes attached. Compare it to the original:
(IMPORTANT NOTE: The reference image must connect at the TOP bubble)
Thanks PapaSmurf. It is there, but one reads over these things and then don’t catch the full meaning/implication. When you read it the first (few) times, there is just too much info and this is the detail - that’s why I thought I’ll mention the solution too.
Heh, that might be an interesting experiment …hehe :eyebrowlift2: and I just noticed how it fits in with the stereoscopic theme! clap-clap
Of course, all the detailed information is why recipes & “wizards” are so good for newbies or infrequent actions (when properly written). Actually numbering the steps in particular makes it so much easier to understand what is involved with each discrete step, and where decisions need to be made. Makes it easier to write clearly too.
step 1. open blender
step 2. put stuff in blender
step 3. push button.
step 4. enjoy.
Don’t get me wrong PapaSmurf, I’m actually complementing you. There is a lot of info - which is a good thing. One just misses the the details when you are in information overload mode.
Rather keep up the good work - it’s MUCH appreciated!
Hey! I just had to do this for an update to the composite blur example. What I did is to set my format resolution to the desired size. I then positioned the object in the left half, and then took the renderlayer through a Distort-Transform node to shift it, and fed the original and the shifted to an AlphaOver. For your, you would need to use a second linked scene with a different camera slightly to the right for that right half.
In your noodle, AnyMation, you might want to add a Scale before the translate to Absolute scale them to 320 wide so there is no overlap.
