References & Examples

• Canary yellow on the one word carrying the joke, at roughly double size
• Caps on the highlight, sometimes italic, against a lowercase base line
• Captions off the bottom band — left-aligned, off-centre, two blocks at once
• Emoji inline at the end of a line
• Song lyrics scattered on the beat instead of stacked
• One caption drawn as a display wordmark, as a gag

⚠️ The wordmark gag imitates a real clothing brand. Clear that with the client before reusing it — the safe technique is the oversized display caption, not the specific brand.

• Whole subtitle track in caps and italic, not just the highlight
• Yellow alternates through the line, landing on the words that carry the insult
• Emoji floating by the face as reaction beats, not parked in the text
• Swearing masked with asterisks rather than cut
• Bottom-left, inside the live area

• Words scattered around the frame instead of one centred block
• Green and blue accents in the same cut
• Single words swapped to a serif or display face
• A sentence split across two separate positions
• 4:5 feed post — captions still well inside the live area

⚠️ The outlier. It works because the show is fast and chaotic — on a calmer title the scatter reads as sloppy, not playful.

• Devanagari and its Latin transliteration set together, so the lyric reads either way
• Scripts mixed inside a single line — aisi की taisi
• Each word in its own solid colour block, spread across the frame width rather than stacked in a caption
• Words repeated in layers as a visual echo when the vocal repeats
• Type placed to the beat, not to a fixed subtitle position
• Bug switches between S1 and S2 as the edit crosses seasons

⚠️ No stroke and no shadow anywhere in this one. The colour block behind each word does that job instead, and the whole piece rides a single purple grade. That works because the background is designed rather than found — it is not licence to drop the stroke on captions over ordinary footage.

• Captions follow the song lyric, not the dialogue
• Built word by word so the hook lands on the beat
• Small white setup, big yellow italic payoff — the same shape every line
• One lyric split across two positions in the frame
• Lines without a hook get no treatment at all

• No text on screen at all except the bug — music and performance carry it
• Clips from two films in one post, held together by the performer
• Bug switches between LAILA MAJNU and BAD NEWZ as the clip changes — with no subtitles it is the only branding in the piece
• Cuts land on the beat; frames left clean

• One line of dialogue assembled from five different films
• Same two-tier caption every time — small white setup, huge acid-green payoff
• All caps in a wide squared display face, locked bottom-left
• Accent is acid #D8FF00, straight from the secondary palette
• Bug names the right title in every one of the five

• Character name cards in a serif display face between beats
• Torn-newspaper headline as a graphic device, not a caption
• Oversized Devanagari set behind the subject as artwork
• Frames composited inside playing-card and poster graphics
• Glitch and VHS passes on selected shots
• Caps subtitles with masked profanity, yellow and red accents

• Plain white lowercase throughout — no accent colour anywhere in the cut
• Italic on the occasional word, and that is the entire treatment
• No emoji, no caps, no scatter, no display faces
• Dialogue assembled into one continuous emotional line
• Multi-panel splits carry the visual interest instead of the type

⚠️ The most useful entry here for a new editor. Restraint was the creative choice — a sincere film gets no jokes. Reach for this register as readily as the loud one.

• Ransom-note captions — every word on its own torn-paper chip, faces mixed
• Character cards pairing the role name with the actor
• An orange bar across the eyes as a repeated graphic motif
• The title wordmark used as a caption in its own right
• Single title, so the bug never changes

⚠️ Uses a song from the film itself, cleared specially for this campaign. Music permission is per-campaign — never assume a track carries over to the next post.

• Type at frame-filling scale — the words are the composition, not a label
• Letters masked so characters pass in front of and behind them
• Several display faces in one cut, matched to the comic-book source
• Halftone and dot-screen passes on selected shots
• Comic-panel splits and shifting letterbox instead of one fixed frame
• Delivered 4:3, not 9:16

⚠️ This one carries no attribution bug, unlike every other reference on this page. Treat that as specific to this piece, not as licence to drop the bug — check before following it.

• White canvas, not footage — clips are cut out and placed on it
• Black serif type, magazine-style, at small sizes
• Devanagari and Latin mixed inside a single line
• Words set vertically, and single words spaced down the frame
• Asymmetric collage layouts rather than a full-bleed frame
• Several titles, bug changing per clip

⚠️ The one place the stroke-and-shadow rule reads differently: the background is a designed white canvas, not unpredictable footage, so black type carries itself. The rule still applies the moment type sits over a clip.

• Every cut matches the one before on composition and pose
• Look across the four stills — same two-people-facing frame, four films
• No subtitles at all; the theme is carried entirely by the matches
• A dozen-plus titles, Indian and international, in twenty-five seconds
• Bug names the right title on every single cut

• Built on a format that was already trending — speed to market is the point
• Subjects cut out and composited onto black with panel splits
• Type set behind the cut-out so the subject overlaps it
• Sparse captions — most of the run has no text at all
• Around nine titles, bug updating on every clip

⚠️ Trends age fast. Copy the technique, and check the format is still current before building a post on it.