Indigo Renderer Forum

Hi, I figured that one might estimate the optimal number of keyframes for a model based on translation vectors angles and rotation angles.

My primary concern is to optimize the file(s) size when exporting large amount of particles, while preserving the required quality level.

I don't think that it should be too complex to compute, but if you have a precise idea or came around this already you are welcome to let me know. Either what I'll post the solution I'd eventually find.

It should also allow to have a scalar as motion blur quality setting [0;1] on the user end, that would be cool (the cooler the more fileprint ^^).

You could use a custom motion blur samples parameter. That way user can increase it if motion blur turns out too "blocky".
Just an idea.

That's what I got already, 2 global parameters (for cameras and objects)

The vantage of an estimator is per-frame and per-object adaptativity, that would be tedious for users to set manually. The alternative is to migrate those parameters at the object level (as per normal), but that would only be half as nice.

We already know the vector of motion per keyframe. For instance, we could add an intermediary keyframe if the angle between two vectors is above eg 120°, or if the rotation goes over this between two frames.

Without concerning the user anyhow... off course, one would need sub-frame data acquisition. Probably slightly more complex than thought

First thing off, if a keyframe uses the same data as the previous and next ones then you can discard the former.

This can reduce the required writing down to 1 key if the object is still at some frames, depending on how you implemented mb. It is important to write the last identical keyframe though or indigo will interpolate much earlier a motion that should start only after.