Having analysed the OP video further, I concluded that the object was about 25 cm (10 in) across and traveled with speed of 4 m/sec. This is in the assumption that the object was near the chimney at the moment it appears above the roof. If it was more distant from camera at the time, its size and speed need to be scaled up proportionally to the distance from the camera.
In addition, the object was spinning at 3.75 revolution per second, or 225 rpm. This is from the OP video frame rate of 25 fps. In the last 20 frames before going out of view, the object completed 3 revolutions.
Below is the images of an alignment of 8 frames selected at regular time intervals (every ninth), covering the full span of the object's visible trajectory, and a lattice of evenly spaced parallel rods viewed at an angle to its plane. The angle value was selected so that the projected sequential positions of the object matched those of the sequential rods.
A good match was achieved for the angle value of about 15°, thus providing us with the direction of the object trajectory relative the direction from the camera to the right of the chimney. Coincidentally, the angular widths of the sequential rods turned out to match those of the object at corresponding positions. As the distance between the adjacent rods in my lattice was 6 times greater than the rod diameter, the distance travelled by the object over 9x7=63 frames (or 2.5 sec) was approximately equal to 6x7=42 object's widths.
From the known camera location and direction, the object's putative trajectories can be plotted and measured on Google Earth. The length of its closest to the camera visible segment (green line) is about 10 meters, giving the object's speed of 10/2.5 = 4m/sec and the width of 10/42 = 0.24 m = 24 cm, similar to the size of a typical frisbee. On a more distant parallel trajectory, the visible segment (cyan) is longer, so the object would have to be bigger and faster. The beginning of the trajectory that was obscured from the camera view by the roof, is shown with yellow line. The object could have been thrown from any point of it and might have a greater initial speed if travelled a longer way.
I previously estimated the object width being about 5 inches (13 cm) by its similarity to the chimney flue cap. It was an underestimate, as that particular cap is probably more than 20 cm wide. This can be deduced from the chimney's width. Its side is 2.5 bricks wide, or about 57 cm.