Motion Blur - Sampling precision

Question

Hello!I'm working on my own motion blur tool, and I've hit a snag with what looks like a precision limitation within AE. In short, the plug-in is able to do up to 2048 samples. It even processes that amount. BUT, there is no visual difference from 512 samples and above, like in the attached image. It's a rectangle that flaps up-and-down 30 times a second, which renders identically between 512 and 2048 samples.WING_TimeRemap_CLOSE_HSMB.jpg I'm using "A_Time" to determine the temporal sampling, which should technically give me a microsecond of precision.A_Time time_val;
PF_FpLong exact_time = (PF_FpLong)in_data->current_time / in_data->time_scale + sample_offset;
const A_long high_scale = 1000000;
time_val.value = (A_long)(exact_time * high_scale);
time_val.scale = high_scale; And my debugger tells me that the desired amount of sub-frame positions are being calculated and rendered, but it seems that they're either discarded in the final render or snaps to their closest valid values. Is this simply an inherent limitation of After Effects, where it quantizes positions above/below a certain value?Thanks in advance- Ø

shachar carmi · Answer

i remebered vaguely that there is a limit to the time resolution in AE at 1/32000, but coupldn't find it in the docs.
so i did a little experiment:
1. create a comp, set it's fps to 999 (any higher value will autmoatically be lowered to 999).
2. put a very small white solid in it.
3. create a position keyframe at frame 0 on the left side of the comp.
4. on frame 1, move the solid to right side of the comp.
5. turn on motion blur for the layer and comp, cranck the comp's motion blur setting to the max.

you'll see indivitual squares representing the motion blur samples. i counted exactly 32 squares.
now:
6. turn the comp's fps down to 500.
7. move the end keyframe to frame 1, becuase now it's halfway between 0 and 1.
you'll see the amount of motion blur samples have doubled.

that means AE caps the time resolution at 1000 fps * 32 samples = 1/32000sec.
(at least that's what i gather from this experiment, and that correlates to what i vaguely remembered)

Sign up

To post, reply, or follow discussions, please sign in with your Adobe ID.

Sign in to Adobe Community

To post, reply, or follow discussions, please sign in with your Adobe ID.

Scanning file for viruses.

This file cannot be downloaded