Synthetic data · Conditions
We didn't wait for winter: one railway scene, four conditions (synthetic data)
One railway scene, one camera path, four environment conditions, rendered in Blender instead of filmed. The orange boxes are ground truth written by the renderer, identical in all four conditions.
Chapters
SYNTHETIC DATA
The footage was generated in Blender and is not a field recording. The boxes are ground truth produced at render time, not model predictions. Internal R&D work.
01What the film shows
Winter under a clear sky, winter under overcast, dusk and daylight haze play at once in a 2×2 grid, then each one takes the centre in turn so they can be compared on the same geometry.
Building the scene instead of filming it turns the weather into a parameter: no waiting for the season or the light. The renderer knows where every object is, so it writes the labels while it draws the frame. No manual annotation, and the same object is labelled identically in all four conditions. The hard cases, an obstacle on the rail under fresh snow, at dusk, in haze, are exactly the ones you cannot go out and photograph.
This is an internal Spikedge R&D scene, not customer work. Rendered at 3840×2160, 30 fps, with 12 object classes. The next step is measuring how a detector trained on this data behaves when conditions shift.
