Preserve Resolution while scaling?

Question

Smart-Object layers are ruining the quality of my work when scaling up.
I create 16-bit style animation where you can see the pixel jagged edges (think Street Fighter 2 look) using Photoshop.  One of the cool tricks I learned was that rather than working at full resolution (1920 X 1080), was that I could work at 1/10th the size and do it in 1/10 the time.
That's because when working at full size, I have to use the grid feature and paint in each grid box to mimic a zoomed in pixel.
But working small, I simply use the pencil tool at 1px, and scale up using Nearest Neighbor, which preserves the hard edges.  A brilliant time saving feature which allows me to focus on drawing the whole picture and not on filling in boxes, which is tedious at best.
The problem occurs when I work in CS4's animation timeline mode when I try to convert Image Size.
Normal layers preserve their hard edges when using the Nearest Neighbor method of re-sampling.  Yet video layers must be converted to smart object layers in order to be resized.  Regardless of whether I convert it to a smart object layer during the resizing process (which it does after a warning pop-up) or before I resize, the smart-object/video-layers appear fuzzy while the normal layers preserve their original hard-edge quality.
Why is that and is there a work around?  I'd hate to have to re-animate everything using the tedious gridding technique.
Thanks!

Answer 1

Thanks for the feedback.  I'll give it a shot and even try scaling each frame up as a normal frame and then copy/pasting into the upscaled version.
And After Effects is great for tweening animation, but not so much for traditional/cartoon animation.
Something like this requires drawing each new frame, and Flash is limited by the "vector look" :
http://vimeo.com/11315189 (made entirely in Photoshop by Kris Anka)
Photoshop should definitely consider enlisting the Flash team to better it's timeline abilities as well as fixing this basic issue of upscaling video layers (by not treating it like a video layer and instead looking at it like normal layers viewed with a time dimension).
Thanks again!