A blackframe report is a candidate location, not proof of an editing fault. Its amount threshold concerns the proportion of pixels classified as black in a frame; its threshold setting defines the pixel-level cutoff. Neither setting tells you whether the frame is an accidental gap, an intentional transition, or a dark image with meaningful detail. Review several frames around each hit, classify the sequence in context, and record why you will keep or investigate it.
Know what the report does and does not say
FFmpeg documents blackframe as a filter with threshold and amount options. The first sets the pixel-value cutoff used to count black pixels; the second sets the percentage required for a frame to be reported. A hit therefore means that the frame met the configured criterion, not that the picture is uniformly black or that anything is wrong.
A stricter or looser setting changes which frames qualify. It does not supply an editorial judgment. The official FFmpeg filter reference describes the filter and its options. Use the settings as a way to produce a manageable review list, not as a universal definition of a bad frame.
Review each candidate in sequence
- Record the reported timestamp and the settings used. Confirm which video stream you scanned and whether you are reviewing the intended source or final export. Keep timestamps tied to that file; a different edit or transcode may have different timing.
- Open the video at the candidate time in a player or editor that lets you step through frames. Inspect the frames before, during, and after the report. A single still cannot show whether brightness is rising or falling.
- Count consecutive reported frames and note their span. Compare the span with the surrounding image sequence. One isolated hit amid visible motion differs from a run of black-looking frames, but duration alone still does not establish intent.
- Classify the picture: full or near-full black; a gradual fade; a dark shot with visible subjects, outlines, or movement; or a planned transition or pause. Consider the cut, soundtrack, captions, and narrative timing. A deliberate fade may be correct even if multiple frames qualify.
- Choose a disposition: accept as intentional, reject as a false or unhelpful candidate, or investigate as a possible fault. Write a short reason. Do not choose a repair method until the context supports one.
Use a candidate worksheet
The following is a hypothetical example, not a measurement or a claim about a particular video. Suppose a 24 fps export contains a fade to black before a title card and a separate night shot. The listed frame spans and observations below are assumed for illustration.
| Timestamp | Consecutive hits | Frames around hit | Classification and decision | Reason |
|---|---|---|---|---|
| 00:00:12.500 | 6 (0.25 s) | Image darkens, then title appears | Intended fade; accept | Matches the planned title transition |
| 00:01:04.000 | 1 | Subject outline and motion remain visible | Dark shot; accept | Picture information is present; the scene is intentionally dim |
| 00:02:18.750 | 18 (0.75 s) | Black-looking interval between two shots | Investigate | Check the edit and source before deciding whether it is a gap or a planned pause |
For each real candidate, include the file or sequence identity, timestamp, consecutive-frame count or duration, what the preceding and following frames show, classification, disposition, and reason. If you change detector settings, keep the results separate so you can tell which configuration produced each list.
When the classification is uncertain
If a shot looks black on a monitor, do not infer that it contains no image information. Inspect adjacent frames and, where available in your review workflow, compare a representative frame at normal display conditions. FFmpeg's separate blackdetect filter documents black intervals using black-level and minimum-duration criteria. It is another detection mechanism, not an adjudicator; its interval report cannot establish whether a fade or pause is intended either.
If the report misses a visually concerning interval, or flags many expected dark scenes, revisit the chosen pixel cutoff and percentage in light of the footage. Change one setting at a time and compare candidate lists; do not treat a particular setting as validated for all cameras, grading, or delivery formats. If timestamps are hard to locate, verify frame rate, time base, stream selection, and whether the report refers to the same cut you are viewing. The cited filter documentation does not promise a visual review interface or prescribe an editorial threshold.
A possible fault should be confirmed against the edit, source material, or intended sequence. Depending on what that review shows, the next action may be to leave it alone, consult the editor, or correct an actual gap. Do not brighten a deliberately dark scene or remove a transition merely to make the detector report clean.
Common questions
Does one reported frame mean there is a black flash?
Not by itself. It means one frame met the configured criteria. Inspect the frame and its neighbors; decide whether the image is truly unwanted and whether the cut makes it conspicuous.
Is a long run of hits automatically a failed export?
No. A longer interval deserves review, but could be a fade, intentional hold, or scene designed to begin in darkness. Compare it with the intended edit and surrounding audio or titles.
Should I keep raising the threshold until every dark shot is detected?
No. Changing the pixel cutoff or required black-pixel percentage changes the candidate set, not the meaning of each candidate. Select settings for a useful review pass, then classify the results manually.
Sources and next step
This guide interprets public documentation and uses explicitly hypothetical acceptance examples, not application tests. Controls and availability can change; verify the cited manual against your installed version.
For work created in Imagild, carry the input requirements and acceptance checklist into the film workspace. The external software controls discussed above belong to their respective products.