Video QC: What Thirty Seconds Cannot Show

One page verdict

A thirty second clip settles less than most buyers expect. It can show motion, sound and continuity, none of which a still photograph carries, and it can show that the item exists in somebody's hands. It cannot show thread count, engraving depth, pore spacing or edge paint thickness, because none of those survive the compression that video encoding applies. The verdict on this page is therefore short: use a clip to confirm identity and function, never to judge finish. That single line is the whole page, and everything below explains where the boundary sits.

Compression is the mechanical reason. Video codecs discard detail in areas of low motion and low contrast, which is exactly where a stitch line and a grain pattern live. A still frame from the same session carries several times the information of a frame extracted from the clip. Where a seller sends video in place of photographs, ask for the stills as well, since the clip does not replace them at any resolution. Ask for three named stills rather than a longer clip, and say which areas they should cover.

Motion hides defects too. A hand moving across a bag at ordinary speed sweeps past a lifted edge within a few frames, and those frames are the most heavily compressed in the clip. Slowing the movement helps, and stopping on the area matters more: a three second hold with the item still against a plain background carries more usable evidence than thirty seconds of continuous handling.

A clip earns its place in three situations. Confirming that a parcel contains the item described. Demonstrating that a mechanism moves, such as a zip, a clasp or a bezel. And showing scale, when a hand or a known object appears beside the item. Three uses, and each one concerns what the item is rather than how well it was made. Outside those three, a photograph is the better request, and it usually takes the seller less time to produce than a clip does.

Five second limits

Five specific things cannot be judged from any clip a seller can record. Stitch density, because compression merges adjacent stitches into a single dark band. Engraving depth, because depth needs light arriving across a surface rather than along it. Material grain, for the same reason as stitches. Plating colour, because auto white balance drifts between frames inside one clip. And completeness, because a clip shows only what the seller chose to point at. Each of those five is a question a photograph answers in one frame with a reference object in shot.

A sixth limit is subtler. A clip cannot be measured. Nothing inside a video frame keeps a reliable scale once the camera moves, and cameras move in every useful clip. A still frame with a bank card in shot can be measured years later; a clip of the same item cannot be measured at all. That difference alone decides which medium to request whenever a dispute is imaginable. Scale is what turns an observation into a measurement, and video strips scale out by design.

Audio is the one partial exception. A zip that binds, a clasp that clicks twice before seating, a rattle inside a case: each is audible and each is invisible in a photograph. Sound does not depend on resolution, so the compression that destroys fine detail leaves it usable. Recorded audio is the channel where a short clip beats every still image, provided the microphone is left on and the room is quiet.

The limits carry a practical consequence for what a reader writes down. A clip supports statements about function and identity, and those belong in the record with the clip date beside them. Anything about finish stays open until a still frame arrives. Mixing the two kinds of statement in one note is what makes a later dispute unwinnable, because half of the note cannot be evidenced at all. Two columns in a notebook, kept apart from the first message onward, prevent the mixture.

Three frequent questions

The first question readers ask is whether a clip counts as evidence in a claim. It does, in the same way a photograph does: as a record of what was shown on a particular date. Its weakness is content rather than standing. A clip that never stops on the disputed area records that the seller handled the item and records nothing about the defect, so a claim resting on it will be answered with a request for a still frame. The clip keeps its value as proof that the item existed and moved on a given day, which is a narrower claim than most readers want it to support.

The second question is whether to accept a clip instead of the photographs requested. The desk's answer is to accept the clip and repeat the request. A seller who has handled the item once can usually be persuaded to hold it still for three frames, and the ask is small. Treating a clip as compliance closes the question while leaving the evidence gap exactly where it was.

The third question is how long a clip should be. Length is the wrong variable and holds are the right one. Three holds of three seconds each, on the areas that matter, beat ninety seconds of sweeping movement. Ask for the holds by name: the heel seam for eight seconds, the hardware close, the box label square on. Named holds are easier to satisfy than a general request for a better video, and they make a second attempt comparable with the first.

Underneath all three questions sits one rule. A clip is a live check, and live checks are consumed in the moment they happen. Stills are records, and records can be re-read, compared and sent on to somebody else. A reader who wants the option of arguing later should ask for the medium that outlives the conversation, which is the still frame with something of known size beside it.

Browse the partner catalogue

Extended walkthrough pointer

The rest of the reading method sits in the angle set this cluster publishes, and the sequence there applies to video as much as to stills. Identity first, then structure, then surface. A clip answers the first group well, contributes nothing to the second, and actively misleads on the third whenever its white balance drifts mid-recording. Reading the three groups in that order is what keeps a clip in its proper place.

For a live call, prepare three requests before the call begins and put them in writing. Ask for a slow hold on the area of concern, a neutral reference in the frame, and a still photograph sent afterwards. Sellers usually agree to all three when the asks are small and specific, and the still that arrives afterwards is the part that gets filed. Requests invented during a call are requests made under time pressure, which is why they belong in writing first. A written ask also survives a change of seller or a handover to a colleague.

Where a call is the only option, keep a written log while it runs. Time, what was shown, what was said, and what was not shown. The last column changes decisions, because a seller who demonstrates a zip and skips the other side of the item has pointed at where to look next without saying anything about the side that was skipped. Writing during the call is the only way that observation survives it.

Points that need measuring belong on the judging pages, and questions beginning with whether something is normal belong on the question entries. This dispatch stops where measurement starts, because the tools it points at do arithmetic a clip cannot support. A reader who takes one habit from this page should take the habit of asking for stills with a reference object, in writing, before the parcel moves. Everything else in this cluster is a request that costs a seller a minute and saves a reader an argument.