Preparing Product Photos for AI Product Videos

Why reference images matter

An AI video model has never seen your product. Without a reference, it invents a plausible bottle, box or shoe. With good reference images, it can hold on to your real shape, color and packaging far more closely — although fine text and complex logos can still drift. Better inputs narrow that gap.

What makes a good reference image

• Resolution: at least 1024 px on the short side; sharper is better.
• Lighting: even, soft light with no harsh reflections hiding the label.
• Background: clean and uncluttered so the product edges are obvious.
• Angle: include the hero angle you want in the video; add one or two alternates.
• Legibility: the logo and label should be readable in the photo itself.
• Accuracy: photograph the current packaging, not a mockup.

How many images to use

One excellent image beats four mediocre ones. For image-to-video, the first image becomes the opening frame. Some engines accept extra reference images for identity; Sardav uses them when the selected quality tier supports it.

Write fidelity into the prompt

State the rule explicitly: match the reference exactly; same shape, proportions, colors, packaging, logo placement and label text; do not add new text. Sardav adds this line automatically to every product scene.

Choose shots that flatter fidelity

• Slow orbits and push-ins keep the product readable.
• Macro passes over materials show quality without stressing the label.
• Avoid fast spins, heavy occlusion by hands, or extreme perspective on hero frames.

Review before you scale

Render one scene first and check it closely at full size. If the label drifts, simplify the move, shorten the duration or try a different quality tier before generating the rest of the storyboard.