It is going to depend on which of the two provides a higher quality resize option. If imagemagick's convert is higher quality than the scale of your encoder, then it will do better, if not, then it will be the other way around.
The one exception to this will be if there is any resolution dependent alterations made to the content along the way. If for example, you add text to the video, it would be much better to scale the images so that the you are creating new content at the native resolution of the finished video.
Anything you can avoid scaling you should and if the images are larger than the output, the earlier you can scale them down, the faster everything else will run anyway.
So with that in mind, I would generally scale down the images initially, if it was necessary to have the video at native resolution already when encoding, though most decent editing software will make it irrelevant because they will use the full quality image and put any new content at the native resolution and still combine everything to the proper resolution when rendering out at the end.