Which Photographs Animate Well and Which Are Not Worth Trying

The success of an animation is decided before you press the button — in the choice of still. The model works equally hard on any source, but the amount it has to invent varies by orders of magnitude. Here are six properties worth assessing in advance.

1. Shot size and edge margin

Waist-up or chest-up shots with visible margin around the figure animate best. Motion always displaces the subject, and if the head sits flush against the frame edge, the first turn pushes it outside — and the model starts inventing what was never there.

A tight portrait filling the whole frame works worse for exactly this reason: there is no free space and any movement hits the border.

2. Background

A simple background — a plain wall, a blurred street, sky — animates cleanly. A detailed background with fine structure (bookshelves, a crowd, patterned wallpaper, nearby foliage) trembles: every frame redraws the small stuff anew and it reads as noise.

The middle ground is a blurred background with large patches of light: it gives depth without detail that can drift.

3. Hand position

Hands in frame are the main risk factor. If palms are visible and especially if fingers are bent, an artefact is likely: the finger count can change mid-clip.

Best options: hands out of frame entirely, at the sides, behind the back, or folded so that fingers do not overlap. A frame where a hand touches the face is a near-guaranteed defect.

4. Sharpness and resolution

A soft photo animates softly: the model cannot add detail that is missing, but it readily adds trembling wherever object edges are vague. A sharp photo gives a stable picture with clean contours.

Resolution matters less than sharpness, but a very small file (a screenshot, an old avatar) is limiting: half the frames go into guessing at facial structure.

5. Light and shadow

Soft even light is the most stable option. Hard shadows on the face are problematic: as the head moves, the shadow should travel according to physics, and the model reproduces that approximately — so the shadow starts living its own life.

Rim light and strong specular highlights on skin often produce brightness flicker between frames.

6. What is happening in the frame

A static pose animates better than a dynamic one. A shot of someone mid-jump, running, or in an active gesture already contains smeared motion, and the model tries to continue a trajectory it cannot correctly reconstruct.

Particular hard cases

• Group photos: each face gets less capacity, and the relative positions of people break down under motion.
• Animal photos: fur and muzzle animate reasonably, but paws and tails often behave oddly.
• Shots with text in frame: letters start drifting by the second second.
• Photos with glasses: lens glare jumps between frames.
• Photos through glass or with reflections: the reflection does not obey the motion of the scene.

A fast check

Answer one question in your head: if the person in the still turned their head slightly, what would appear that is not visible now? If the answer is "almost nothing", the photo will do. If it is "the other ear, part of the neck, a hand, a chunk of background behind the shoulder", the model will invent all of it, and the result will show.