Media tools
Sume media tools by API: trim, filter, captions, frames, inspect, speech-to-text, upscaling, background removal, transitions, and Timeline assembly.
Start with Video editing API limits: Sume media tool caps in one table

Check an AI take says your script: hypit align and unmatched words
hypit align pairs each script token with the transcript words of a generated take, and lists unmatched words. What it measures and what it does not do.

Edit Video in Sume: which models appear and why H3 Max does not
Sume's Edit Video mode lists Auto, Kling 3.0, Wan 3.0 and MiniMax H3 and needs a reference video. H3 Max and Grok Imagine are missing; the API differs.

English to Korean captions: style follows the text, not language
Localising a video to Korean? Sume's caption style follows the wording, and a Latin style on Hangul text is a 400. Pick styles per locale before a batch.

Find the video frame where a phrase is spoken: hypit tile around
hypit tile with around {phrase} resolves a spoken phrase on the transcript and returns labeled frames around it. Parameters, limits and the error codes.

hypit boundaries: cut candidates and the 0.1 threshold
hypit boundaries scores possible cuts from a 12 fps 32x32 difference pass. What rate and threshold do, and how it differs from reference ingest's shot list.

Word-level transcript with confidence scores: hypit transcribe
hypit transcribe returns each word with start, end and a 0-1 score using WhisperX, or Sume STT 1.0 without scores. Engines, checkpoints, billing and errors.

Korean caption styles compared: weight-shift to editorial-emphasis
Sume has six Hangul caption identities plus korean-ad. Compare black-outline, weight-shift, highlight, pill-karaoke, clip-wipe and editorial-emphasis.

Korean subtitle fonts for a video API: 29 faces on Sume
Sume's caption font field names 29 Hangul faces, all SIL OFL 1.1, from Pretendard to Gmarket Sans. Which suit which style, and what fails on a Latin style.

Add a launch date plate to a teaser video with compose overlay
Put a launch-date card on the bottom of a teaser with Timeline compose overlay: width_ratio, margin_ratio, a 300-second ceiling and $0.02 a job.

Launch teaser music on Lyria 3.5: section markers, 30 seconds
Write a 30-second launch-teaser track on Sume Music Router: section markers, one named moment and no duration field. $0.125 per generation.

Keep the sound of AI clips in a Sume timeline: detach to a spine
A Timeline 1.0 render takes one audio spine. To keep audio generated with your clips, detach each clip's track, join the parts, and use that as the spine.

Reference ingest OCR needs_verification: read the crop
Low-confidence on-screen text from reference ingest returns as needs_verification with a native crop. How the 0.85 default works and how to read the manifest.

reference_ingest_semantic_unavailable: what to do instead
semantic: true is refused on reference ingest, and reference_ingest_unavailable means the media runtime lacks the function. The two errors and what to call.

source_too_long_for_reference_ingest: clips over 300 s
Reference ingest rejects a clip over 300 seconds with source_too_long_for_reference_ingest. What the cap covers and the routes Sume's docs point to instead.

Square YouTube Short (1:1) from a vertical master with Timeline
YouTube accepts square or vertical Shorts. Render a 1080x1080 version of a vertical clip with Sume Timeline, using fit blur or contain instead of a hard crop.

TikTok trending video transcripts via API: what summary_mode returns
Can you get a transcript with Sume's trending video search? summary_mode transcript returns metadata plus an unsupported warning. What to do for the words.

Trim an AI clip first, or use source_in on the timeline?
For a shot that is too long, set source_in and duration on the Timeline 1.0 slot. Use video_trim ($0.02 per job) only when another step needs a new MP4 file.

Which AI video models take reference images in Sume's Videos panel?
Auto, Kling 3.0, Wan 3.0 and MiniMax H3 show reference slots in Sume's panel; H3 Max and Grok do not. Plus the API limits for image, video and audio references.

Why did my video settings change when I switched model in Sume?
Sume's Videos panel clamps resolution, aspect ratio, duration and audio to the new model: 720p first, first listed length, first listed ratio. Worked examples.

Which AI video models take an end frame in Sume's video picker?
Wan 3.0, MiniMax H3, H3 Max and Auto take an end frame in Sume's Videos panel; Kling 3.0 and Grok Imagine do not. How to check the API list too.

YouTube caption file for a Short: with timing or without timing?
YouTube's Upload file option asks for With timing or Without timing. Which to pick for a Short, and how Sume's transcript segments and burned-in captions fit.

YouTube Shorts max 1080p: downscale a 4K vertical clip with Sume
YouTube's Shorts help says uploads have a maximum resolution of 1080p. Downscale a 2160x3840 clip to 1080x1920 with video-trim's exact-mode output conform.

YouTube Shorts auto-captions not showing? Burn them in with Sume
YouTube says Shorts captions exist only in certain languages and on mobile devices. When that is not enough, Sume burns captions into the MP4 for $0.20.

YouTube Shorts safe zone: place burned-in captions with anchor_ratio
YouTube's Shorts editor shows white lines and icons where overlays may be hidden. Move Sume's burned-in captions clear of them with placement.anchor_ratio.

Crop a 16:9 video to a centered 9:16 strip: the crop fractions
For a centered 9:16 crop of a 16:9 video, send crop x 0.3418, y 0, width 0.3164, height 1 to Sume video-filter. 1:1 and 4:5 values are in the table.

ElevenLabs sound effects API: 0.5-30 s, loop, and Sume's route
ElevenLabs POST /v1/sound-generation takes duration_seconds 0.5-30 and a loop flag. Sume lists no sound-effect endpoint; here is the music route and its limits.

FFmpeg drawtext on a hosted API: why Sume filters refuse it, and cues
Sume video-filter refuses drawtext and subtitles because they read files. To burn text onto a clip, send cues to video-captions: text, start and end seconds.

Adjust video brightness, contrast, saturation by API: eq filter
Send eq=brightness=0.06:contrast=1.1:saturation=1.2 as the filtergraph of Sume video-filter. Ranges come from FFmpeg: brightness -1 to 1, saturation 0 to 3.

Runway Enhance Frame Rate: 10 target rates, 300 s, vs Sume
Runway Enhance Frame Rate takes clips up to 300 seconds at 1 credit per 2 seconds. Sume has no such model; its video-trim output.fps accepts 24, 25, 30 or 60.

Put a product shot above a UGC creator clip: compose stack
Timeline compose stacks one Sume-hosted still over one video into a single MP4. Set ratio for the still's share. The public rate is $0.02 per job.

Video frames vs video inspect: max_edge 16 vs 64 and the 768 default
Sume video-frames max_edge is 16 to 2160 and keeps source size if omitted. Video-inspect max_edge is 64 to 2160 and defaults to 768.

inspect_source_has_no_audio: check probe.has_audio before transcribe
transcribe true on a silent clip fails as inspect_source_has_no_audio. Probe first with frames false and read probe.has_audio, then ask for the transcript.

Instagram First Draft cuts pauses: the same rough cut off the phone
Instagram's First Draft trims clips and cuts pauses on iPhone. Sume has no pause-remover, but trim and timeline give you a scripted rough cut.

MiniMax H3 reference formats: HEIC, MOV, WAV, MP3 limits
MiniMax H3 accepts HEIC/HEIF images, MP4/MOV video and WAV/MP3 audio as references. A pre-flight checklist before you submit minimax-h3 on Sume.

TikTok API frame rate 23-60 fps: conform a clip with Video trim
TikTok's Content Posting API accepts 23-60 fps video. Check the clip's rate, then conform it to 24, 25, 30 or 60 fps with Sume's Video trim output field.

TikTok API video resolution 360-4096 px: size an AI clip first
TikTok's Content Posting API wants width and height between 360 and 4096 pixels. Sume's Video trim output allows 256-2160, so keep your values at 360 or more.

TikTok API video codecs: H.264, H.265, VP8, VP9 and Video trim modes
TikTok lists H.264, H.265, VP8 and VP9 for API uploads. Sume's exact trim re-encodes to libx264 yuv420p; keyframe trim copies the source stream unchanged.

TikTok photo post image formats: WebP or JPEG, not PNG
TikTok's photo post API lists WebP and JPEG, 20 MB per image, up to 1080p. If your AI images default to PNG, set Sume's output_format to jpeg or webp first.

Reference audio for Wan 3.0 and MiniMax H3: WAV, MP3, 15 MB
Both Wan 3.0 and MiniMax H3 take WAV or MP3 reference audio up to 15 MB and 15 seconds in total. Where the clip counts differ and what Sume says.

Wan 3.0 reference images: 20 MB, 240 to 8,000 px, 8:1 cap
Alibaba's Wan 3.0 reference images must be 20 MB or less, 240 to 8,000 px per side, ratio 8:1 at most. A check to run before wan-3.0 on Sume.

YouTube takes 256 GB uploads, but Sume media tools stop at 30 minutes
videos.insert accepts files up to 256 GB. Sume trim and inspect read sources up to 1,800 seconds. Where the two limits meet when you cut a long recording.

Before/after side-by-side video: hstack in Sume video filter
Make a before/after comparison clip with ffmpeg hstack through the Sume video filter API. Split the video, grade one side, stack, and dry-run first.

Instagram Feed video ad: 1 s to 60 min, 4 GB, no edit lists
Meta's Feed video ad specs: 1 second to 60 minutes, 4 GB, 1% ratio tolerance, no edit lists. What Sume's trim and timeline caps mean for long ads.

Instagram Reels ad length: up to 15 minutes, 500 px over 30 s
Meta's ads guide lists Instagram Reels ads from 0 seconds to 15 minutes, 4 GB, and a 500 px minimum width at 30 seconds or more. How a Sume cut fits.

Instagram ad text limits: 44 characters on Reels, 125 on Feed
Meta's ads guide lists 44 characters of primary text for Instagram Reels ads and 125 for Feed video ads. Where on-video captions from Sume fit in.

Instagram Reels length: 20 minutes, but 3 for new viewers
Instagram says you can build a Reel up to 20 minutes, but Reels over 3 minutes are not recommended to new audiences. How to cut a 3-minute version with Sume.

Fix a product ad's white balance with colortemperature in Sume
Warm or cool a product clip with ffmpeg colortemperature in Sume video filter: 1000 to 40000 K, default 6500, plus mix and pl. Dry-run it for free.

Music for a Short cut from your own video: Audio Library is off
Edit into a Short bars Audio Library sounds. Build the Short as a file instead: Sume Music 1.0 plus a Timeline soundtrack bed that ducks under the voice.