Edit your first video with Valmera
Upload footage you already have, describe the changes in chat, review the preview, and export the approved version in Studio. You can start without timeline-editing experience. Account creation and uploads are free; a subscription is required before indexing and agent editing begin.
How Valmera Works
Valmera is an agentic AI video editor. You upload real footage, the system analyzes it — building a word-level transcript, detecting shots, and mapping silences — and then you describe the edit you want in plain English. An AI agent performs it: cuts, captions, music, reframing, and polish. You refine the result with follow-up messages, just like working with a human editor.
You never have to touch a timeline to get a finished video — but there is one. Valmera includes a visual timeline and editable transcript — draggable blocks for inserts and music, cut markers, a scrubbable playhead — for the moments when you want direct control.
Valmera uses edit records and verification checks to help ground the agent's replies. Review the rendered preview yourself: confirm the requested cut, correct caption mistakes and listen to the audio. A completion message is not a guarantee that every change is right.
Step-by-Step
Create Your Account
Sign up at valmera.io with your email or Google account. Account creation and uploads are free; choose a subscription before indexing or editing begins.
Upload Your Footage
Attach the video you want edited — MP4, MOV, M4V, MKV, WebM and similar formats, up to 14 GB and 3 hours. Podcasts, screen captures, gameplay, vlogs, and long recordings all work.
Let the Analysis Run
Before any editing happens, Valmera builds a word-level transcript of everything said, detects shot changes, and maps the silences. Analysis runs once per upload — longer files take longer, and you can watch the progress in the Studio.
Describe the Edit
State the audience, target format, changes you want, and anything that must stay. For example: "Cut the long silences and add captions. Keep the speaker's full explanation and the original landscape framing." Start with a focused edit so the preview is easy to review.
The Agent Edits, You Preview
The agent performs the edit and renders a preview in the player. Watch and listen to the result, especially around cuts, names in captions, and changes in framing. Use the preview to check the edit rather than relying only on the chat summary.
Refine — by Chat or by Hand
Send follow-ups like "Make the intro shorter" or "Use bolder captions" and the agent re-edits. Or take direct control: drag blocks on the timeline, scrub the playhead, or fix a sentence in the transcript panel and the captions re-render.
Export at Full Quality
Previews render from a fast lower-resolution proxy, but Export renders the final video from your original full-quality file. Download the MP4 and post it anywhere.
Full details on formats and size limits are in Uploading Footage & Assets, and how edits are charged is covered in Understanding Credits.
A complete first-edit example
Use a short talking-head recording for your first project. A one-to-three-minute clip is a useful learning exercise: you can compare the whole source with the result. The prompts below are examples to adapt to your footage, not a claim that a particular recording has been tested.
1. Make a clean first cut
Remove the long pauses and obvious repeated starts. Keep complete sentences, natural breathing, and the speaker's meaning. Add readable captions. Keep the original framing and leave music out of this version.
This brief gives the agent both an editing job and constraints. A request such as “make it viral” leaves the audience, length, tone, and material to preserve unspecified. If a section must remain, identify it by its spoken phrase or source timestamp.
2. Review the parts that can change the meaning
- Listen to each cut for clipped syllables, missing context, and unnatural joins.
- Check names, numbers, product terms, and punctuation in the captions.
- Watch the opening and ending so the video starts with context and finishes a thought.
- Compare any removed repetition with the source; a repeated phrase can be intentional emphasis.
3. Revise the version you reviewed
The cut before “here is the example” feels abrupt. Restore a little of the pause before that phrase. Correct the caption spelling to “Valmera.” Keep the other cuts and caption style.
A narrow follow-up makes the intended revision clear. Use the project history when you need to revisit an earlier edit. For direct caption corrections, see the timeline and transcript guide.
4. Adapt it for a vertical destination
Make this edit vertical at 9:16. Keep the speaker visible and place captions away from their face. Use blurred padding if a tight crop would hide the demonstration.
Inspect the framing throughout the clip, particularly if the speaker moves or shows an object. Vertical video is a framing decision as well as a size change. Review crop, pad, and blurred-pad options before adapting footage that relies on details near the edges.
5. Export and check the downloaded file
Start the final export in Studio after reviewing the current version. The preview is a faster working copy; the final render uses the original upload. Open the downloaded MP4 and check its sound, captions, framing, and ending before publishing. The standard Valmera paid-plan policy omits overlay watermarks, but service settings can override it. Inspect the actual final file for branding and any end card. See preview and export behavior for details.
If your first edit does not go as expected
- The file uploaded, but editing has not started
- Check your subscription and the project progress shown in Studio. An upload can be stored before you subscribe; indexing and agent editing require a subscription. Larger recordings also take longer to analyze. Read the displayed state before uploading another copy.
- The captions contain mistakes
- Correct the exact word or phrase in the transcript or ask for a specific spelling correction. Listen against the source. Background noise, overlapping speech, names, and specialist vocabulary can affect transcription; English is the best-tested path.
- The AI removed something I wanted to keep
- Identify the missing phrase or source time range and ask to restore it. State why it matters and which other edits should stay. Review the revised transition so the restored section fits naturally.
- The preview looks softer than the source
- Previews use a lower-resolution working copy for speed. Evaluate pacing and composition there, then inspect the final MP4 for delivery quality. A final export is a separate action in Studio.
- I want an outside AI assistant to do the edit
- Read the MCP connection and access requirements first. Connector availability depends on authorization. MCP edits and reviews projects; final export is started by the user in Studio.
For your next project, use the editing prompt guide, credit explanation, and current plans.
Start Without a Video
You do not need footage to start. In a canvas project you pick an aspect ratio — 16:9, 9:16, 1:1, or 4:5 — and build a sequential timeline out of images, clips, music, and media you generate inside the edit. That is the path for a photo slideshow, a product carousel, or an image-plus-voiceover video.
Everything else works the same way: describe what you want, the agent assembles it, and you refine by chat or by dragging blocks. Stills get Ken Burns motion, on-screen text and overlays are available, and music mixes with automatic ducking.
The honest trade-off: with no main video there is no speech transcript, so canvas projects have no transcript captions, no speed changes, and no censor regions or reframing — details in Timeline & Transcript.
Tips for Better Results
- Say who the video is for. "A punchy 60-second vertical clip" and "a relaxed 15-minute YouTube video" get very different — and better — edits than "edit this."
- Iterate in small steps. Instead of rewriting your whole request, send one change at a time. The agent edits the existing cut rather than starting over.
- Upload your own music. Attach a track (up to 50 MB) and the agent lays it under your video with automatic ducking, so it never drowns out speech.
- Fix words in the transcript. If the transcript got a name or term wrong, edit the sentence in the transcript panel — the captions re-render from your correction.
- Be specific about caption style. Named caption presets, bundled fonts, karaoke word-pop, bottom/middle/top position, sizes from S to XL, colors, per-word emphasis, and entrance animations are all one sentence away.