The film and isolated image are related objects
The film supplies sequence, sound, editing and narrative; the still isolates one moment for examination. A filmography entry alone does not locate the picture, while an image caption alone does not establish the production. Connect the work reference, timecode and figure caption.
Record original and translated titles where relevant, director, production company or distributor, and release year according to the discipline's style. A television reference may include series title, season, episode title, episode number and responsible creators. Check whether the required system also names medium, platform or URL.
Identify the edition actually viewed
Theatrical, director's, restored, DVD, Blu-ray and streaming editions may differ in duration, sequence, aspect ratio and colour. Name the edition, provider and access date when they affect retrieval. A timecode without edition can point to the wrong moment.
Account for adverts, recaps and “skip intro” behaviour. Players may count time differently. Note the displayed total duration and add a concise scene description. For television, episode plus timecode reduces ambiguity.
Use a timecode as a locator, not interpretation
Give hours, minutes and seconds for one still; use a start and end for an extended shot or sequence. Confirm the value in the cited edition. Frame-level analysis may require frame number and frame rate, in which case record how they were measured.
The locator does not explain significance. A caption can identify characters, setting and shot neutrally; interpretation belongs in the text. If subtitles appear, state their language and source because wording may vary by release.
Citation is not permission
A frame can embody several protected contributions. UK copyright exceptions for criticism, review and quotation depend on conditions including fair dealing and sufficient acknowledgement. An ornamental image does not serve the same purpose as analysis of composition, editing or representation. Application depends on facts and jurisdiction.
A publicity still can carry separate terms and may not be a frame from the film. Check licence and required credit. Recognisable people and sensitive contexts may raise additional concerns. A repository or later publication can require a fresh rights review.
Make every reproduced frame do analytical work
Refer to the figure number and identify the visible feature supporting the argument: lighting, spatial arrangement, gaze, costume, edit or word-image relation. Avoid a gallery of undiscussed scenes. Use no more frames than the analysis requires and explain why those moments were selected.
A caption pattern is: “Figure 4. [Neutral scene description], author's screen capture from Title (Director, year), [edition/provider], 00:42:17; uncropped.” Put the complete work reference in the filmography or bibliography and adapt ordering to the required style.
Keep the source frame separate from the publication image
Preserve the original capture in the protected project record and create a publication derivative. Log crop, scale, brightness, arrows and labels. Alterations must not distort the feature analysed. Two compared frames each receive their own still ID and timecode.
Record failed attempts and edition substitutions. If a streaming title disappears, the edition note should still explain provenance. For an online video, use the video–timecode–screenshot schema, which adds channel, upload and mutable platform details.
Build comparison sequences with an explicit selection rule
Editing, gesture and movement may require more than one still. Decide which stages the sequence must represent before capturing them—for example, establishing frame, cut and reaction. Give every frame its own still ID, edition and timecode, then describe the selection rule in the method. This reduces the risk of choosing only the most dramatic images after forming the conclusion.
Players can expose slightly different frames at the same displayed second. If the argument depends on a single frame, record frame rate, frame number and the tool used to locate it. If the analysis concerns a broader scene, false millisecond precision adds little. Report only the precision that the edition and playback method genuinely support.
Before submission, have another reader retrieve at least one frame from each sequence. They should compare title, edition, subtitle language, timecode, caption and file. Resolve discrepancies in the record while preserving the original capture history.