Capture planner

A splat is trained from many photos of the same thing from all sides. Plan the shoot here: how many photos, or which frames of a video to keep.

Side view of a capture: three rings of camera positions around an object, one low, one level and one high. Each dot is one photo. high level low
Drawn by us. Each dot is one photo. Walk each ring at an even spacing with the subject in the middle of every shot, then change height and walk the next one; on the high ring, look down.

Photos

Video

Video is quicker to shoot, but every frame is a little softer than a photo. Walk slowly and keep the camera steady.

The ffmpeg line is a standard frame export; we have not run it on your footage. Look through the frames and delete blurry ones before training.

Camera settings checklist

  • Manual mode, so every shot has the same exposure. Lock aperture, shutter speed and ISO.
  • Aperture f/8 to f/11, so the whole subject is sharp.
  • Shutter 1/125 s or faster when handheld, to avoid blur.
  • ISO 100 to 400, as low as the light allows.
  • Overlap: about 70 to 80% between neighboring shots, 60 to 70% between rings.
  • Video: 4K at 60 frames per second or more, with exposure locked.

These are the published PlayCanvas capture guidelines, not our own tests. Shiny, see-through and moving things (glass, water, people) are hard for any splat trainer.

Next steps

See which training tools run on your computer, then pick a file format to share the result. Tools that train splats are in the directory.

Source

Checked 21 November 2026.