I want to use the Node Editor or Video Sequence Editor to bring in two sequences of images and render them side by side. The images are sized so that they should fit one on the left and one on the right, but using the offsets doesn’t seem to bring the desired results. Any clues on how to accomplish this?
Thanks.
If you know the size of your images you can use the compositor as follows:
My images were 500px*500px here so I resized one to double its size on the x axis, and the applied a 250px offset to the left for the first image and to the right for the other
This is how I usually do it but I wouldn’t be surprised if there are also other more elegant or straightforward ways.
On the left you can see the source strip, the adjustment layer (to save time by simply copying the properties), and the final transform effect. Same for the second source strip.
These are placed there to illustrate what the meta strips on the right are made of. The final step is to stack one meta strip on the top of another and apply an alpha over effect.
image_sideBySide_example.blend (387 KB)
In the Compositor (node Editor) you can use an oversized image as a “canvas” onto which you can apply whatever components are needed for your split-screen or multi-screen effect. If you specify the oversized canvas dimensions as your rendering dimensions, you can position other stuff on it using the Distort>Translate node. This way you can keep your component imagery at its original size. I usually use a pure black frame for the “canvas” but it can be anything you want, even a piece of decorative “frame” art if it works for your concept. Most often, it’s the first element in the node pipeline, since everything sits “on top” of it.
I’ve never had any problem doing this with the VSE, and I think the much smoother handling of sequences/movies makes it the better choice at this point.
Just set the render resolution to 2X the image width, add both strips, enable offset for both and enter the correct value for the one you want to move to the side.
And make sure the top strip uses Alpha Over blending mode.
I think choice of the Video Editor vs. Compositor is mainly a matter of which best suits the specific situation. For simple split-screen, the VSE works very well, but it has many fewer capabilities for image combinations and manipulations than the Compositor, so for more complex FX situations, I’d go with the latter. In any case it’s go to know how to work with either.
The OP explicitly referred to the VSE or the compositor and this is why I proposed a VSE-based solution. I agree with chip that the compositor offers more control.
Ideally, though, my first choice would be neither the compositor nor the sequencer. If I’m not super lazy (which I usually am), I’d simply use the 3d viewport!
Import images as planes (or onto planes) and animate the planes, move them in and out of the frame, toggle plane visibility of each on/off, create outlines, add other background elements such as gradients or moving textures, add captions and/or titles, add other graphic elements etc. For optimal control, some of the elements could be brought into the compositor for additional tweaking. Of course, this last step is totally dependent on the time available to complete the project.
FOSS-based motion graphics! 
If I’m actually translating planar graphics I always do it in the 3D window because the Compositor (not sure about the VSE) has a single-pixel increment limitation that can make subtle moves very choppy. By using planes in front of a camera, the resolution of the applied image and the resolution of the final output don’t affect the smoothness of the motion. But it also means using much larger image sources, so they remain sharp when rendered after being processed as textures.
Using multiple Scenes + the Compositor also lets you use two different camera types in the same composite rendering, so you can combine rostrum-style graphics with 3D perspective scenes (think multiplane backgrounds and such).
