When I put a downscaled text element into a layer and then upscale the layer, the whole thing looks blurry.
What I mean is that after i downscale a text to 20% it makes some kind of rendered buffer and when I upscale the layer (containing the text) to 500% it takes the rendered image and upscales it, resulting in pretty ugly output.
Is this a motion limitation (I know shake has no such problems) or have i set up something wrong?
[vrinek502]"i downscale a text to 20% it makes some kind of rendered buffer and when I upscale the layer (containing the text) to 500%"
Why would you do that? It's a rather tortured workflow. Just make the text the size that you need it, and set the resolution of the layer to be the size that you need. If you need the text to resize itself over the course of your timeline, then scale the text directly.
[vrinek502]"I know shake has no such problems"
Motion is not Shake. Shake is not Motion. They work differently because they're designed to do different things. I suspect that in the near future the 2 products may be merged, or at least they'll start to borrow features from each other, but not today.
Now in post: Peristroika, a film by Slava Tsukerman
I didn't mean I do layering this way, no. I just made an example out of a simple project. The only reason I would put text in a layer and then scale it, is because other stuff need to be in the same layer too.
To clarify myself: Does Motion use a buffer for each layer? (meaning that every time i use a new layer it loses quality on a scale or half-pixel-movement) Is what is missing from Motion concatenation? (I remember this from Shake)
I thought at first that Motion just takes ALL the parameters from ALL the layers and summarize them to render vector content (like text) instead of going a layer up every step to make a render. I believe that sequential rendering (or how the hell this is called) is a huge disadvantage of a compositing software like Motion (if it applies of course).
It all depends if the layer containing the text is getting precomped. If you were to create a layer, drop some text in and then scale the layer, the text would be rasterized after the layer's transforms (scale). That violates Motion's usual compositing order-of-operations, but it helps you maintain text quality, when possible. If the layer gets precomped, though, then the layer's transforms happen after the text object is rendered, resulting in scaling artifacts.
Things that can cause a layer to get precomped: any blend mode other than 'Combine', filters, masks, enabling Fixed Resolution, etc.