Looks like I need to autoNAK and try again.
The problem occurs when the *previous* frame is capable of overflowing
damage (eg: bilinear filtering was used), and can't be solved completely
based purely on current state.
On 02/12/15 01:46 PM, Derek Foreman wrote:
> When we zoom or scale an output i
When we zoom or scale an output it's possible for a single buffer
pixel to be stretched over multiple output pixels.
When this happens, if a client is sending exact damage rectangles
we can end up with incomplete updates - the client doesn't know
that the previous frame resulted in some pixel blee
