SPOT THE SLOP
Slop hides in stills and multiplies in motion. Inspect the exact crop you intend to use, name the visible defect, then decide whether that crop is safe to pass downstream.
A visible still-frame defect is already a stop. Motion is not a second opinion — use it only to confirm an edge, reflection or object you genuinely cannot judge while it is still.
THE TELLS
Run these automatically until you stop having to think about them.
Crates, railings, hardware you can almost read. It looks like a thing, and the closer you look the less it resolves into one.
In motion it turns to mush and multiplies — every frame re-guesses what the object was, so the ambiguity compounds instead of settling.
The source deck names four tells; three arrived intact. The fourth is missing rather than merged into these — left blank on purpose, because inventing a plausible one and presenting it as received doctrine is exactly the failure this whole course is about.
EVERY MODEL HAS AN ACCENT
Beyond the universal tells, each generator fails in its own characteristic way. Learn two and you can usually name the culprit from a single frame.
Ruler-and-stock
Everything parallel, everything square, everything evenly lit.
- — Ruler-straight symmetry — fence rails parallel, drives in perfect curves
- — Flat light and colour with no contrast play, so nothing has volume
- — Detailed textures that still read as a 3D render rather than a photograph
- — Hyperbolises every edit — ask for graffiti on one wall and the whole location gets tagged
That last one is the trap. Ask for “more alive and cinematic” and it takes you literally, littering the frame with junk and dirt. The result is beautiful, staged, and completely lifeless.
Sharpened-and-warm
Microcontrast at the ceiling, depth abolished, white balance dragged yellow.
- — Hard halos on every edge from oversharpening
- — No depth at all — everything in focus, no bokeh, no separation between planes
- — White balance pulled warm until the whole frame yellows
- — Dynamic range squeezed to the middle, so shadows and highlights meet halfway and the frame goes limp
Its most damning tell is a single sickly texture laid over the ENTIRE frame — the same film-wrap pattern sitting on kitchen cabinets and on car bodies alike. One material, everywhere, regardless of what the object is.
Named by behaviour rather than by product, because the products change and the failure modes outlive them. When a new model arrives, the question is which accent it has.
CALIBRATE ON GOOD FRAMES
The instruction is literal: stare at good frames until your eye adjusts. You cannot reliably detect a defect without a working sense of what undamaged looks like. But “good” is not a feeling — each clean frame proves something specific, and that is the thing to look for.
- Held textures
- Surf, black rock and wet foliage all reading as different materials under one grey sky. Proves the plate has real surfaces, not one pattern laid over everything.
- A real light gradient
- Hard light running from bright panes to deep shadow through every step between. Proves dynamic range that has not been squeezed to the middle.
- Legible grime
- Rust and graffiti holding their edges while the clouds keep their volume. Proves detail and softness can coexist — the frame is not uniformly sharpened.
- Motion and depth
- Sharp foreground detail falling into real motion blur on the track behind it. Proves separated planes, which is exactly what oversharpening destroys.
- Density that holds
- A busy period crowd where every face, uniform and reflection survives a dense composition. Proves the model did not run out of attention when the frame got complicated.
Notice that each one is the exact inverse of a tell. Held textures against oily surfaces; a real gradient against squeezed range; motion and depth against everything-in-focus. That is not a coincidence — a calibration set is only useful if it disproves something.
Run the scan before you choose a location and again before you approve a final shot. It is cheapest at the start and most expensive after everything downstream has inherited the flaw.
Related: Proof, not promises for deciding something is wrong, and Test it in motion for finding what caused it.