VSE transform position attribute issue

Can somebody explain what is the logic behind that the VSE transform position attributes are integer data type?
The position X/Y are pixel values as the UI hints:

And that means (AFIK) if we animate those attributes it results a kind of a stepped animation curve:

And the animation is obviously stop/stepped motion accordingly. I think this is definitely not what we want most of the time.

1 Like

Pixel is not divisible, and previously offset values were integer, so they have been kept as integer. Why do you think this is an issue?

1 Like

I understand that pixel is not divisible that is why 2D transformations are filtered/interpolated in CGI (as the Transformation Effect Strip is) . So maybe it is more precise to ask: why the 2D transform of a strip is not filtered?

During preview nearest interpolation is used and bilinear for rendering. This decision has been questioned in ⚙ D12807 VSE: Added filter mode to strip transform.

But you are correct in that using float values can result in smoother looking animations, which I haven’t thought about. I think this would be good idea to change. Thanks for feedback!