
Cloudinary's l_video layer becomes a temporal join when you add fl_splice. Without the splice instruction, it is a video layer over the base rather than a clip that follows it. Define the intended sequence, match the visual dimensions and inspect the join before delivering the whole assembly.
Key Takeaways
- Overlay placement and temporal concatenation are different operations.
- The documented visual splice requires matching width and height.
- Appending, prepending and trimming need deliberate parameter scope.
Write the intended sequence first
Keep both source clips and record which should play first. Specify whether the second clip should appear simultaneously, follow the first or precede it. A successful video response does not establish that the chosen operation matches that sequence.
The Cloudinary concatenation guide explicitly says l_video with fl_splice concatenates the uploaded video instead of treating it as an overlay. Appending to the base video's end is the documented default.
For a hypothetical portfolio film, a portrait introduction should finish before a product close-up begins. Seeing both pictures at once would fail that requirement even if each source is correct.
Match the visual inputs deliberately
Cloudinary requires assets joined by splicing or a custom transition to have the same width and height. Its guidance allows size transformations to establish that match.
Choose the intended canvas and inspect how each source fits it. Do not silently crop a face merely to satisfy equal dimensions. Keep each original and review the transformed candidates separately before evaluating the sequence.
These are Cloudinary's documented conditions, not a claim that every concatenation tool has the same requirements. The stream-compatibility guide concerns FFmpeg's direct joining mechanism and should not be substituted for this operation.
Keep prepend and trim scope explicit
To place the added clip at the beginning, the guide uses so_0. When fl_layer_apply is also used, it distinguishes that special prepend meaning from a start-offset transformation applied within the layer.
For a trimmed added segment, review which component holds the offset. The official guide places a prepend so_0 in the fl_layer_apply component when that closing flag participates. Do not move parameters casually after signing or approving a delivery chain.
For private or authenticated inputs, also use the documented delivery-type form, such as l_video:private or l_video:authenticated, with the appropriate asset identifier. Identifying that type is not authorization to expose client media.
Accept the seam and the complete sequence
Inspect a short interval before and after the join for intended order, sound, repeated frames and missing action. Then review the full result's beginning, ending and duration against the source plan. The export-range guide helps distinguish an intentionally short output from a missing ending.
Record the approved assembly with an approved-file handoff record. Accept when the actual sequence and visual fit match the brief, not merely when a URL decodes. No concatenation, authorization or playback test is claimed here.
Related tasks
- Why Does Joining Two Valid MP4 Clips Break Audio at the Seam?
- The timeline is complete but the export is only ten seconds: check the export range
Sources and scope
- Cloudinary: Video concatenation: l_video with fl_splice concatenates rather than overlays; joined visual assets need equal dimensions. Append is the default, and so_0 at the applicable layer-apply component prepends. Private/authenticated layers require their delivery-type syntax.
Documentation-based guidance from the cited official pages captured on October 25, 2026. Scenarios and acceptance procedures are proposed, not executed tests. No account operation, measured demand, AI-citation frequency, processing benchmark or search-indexing outcome is claimed.
Prepared by Imagild Editorial, the company editorial team responsible for these guides. AI assisted research and drafting. This is documentation-based guidance with proposed examples, not a claim of hands-on testing or an independent product review. Report a correction.
Cover: AI-generated conceptual editorial illustration; not a real application screenshot, observed output or before-and-after test.