How to Match a Reference Video’s Editing Style with ChatGPT

Adapting a reference video can mean replaying the same few seconds dozens of times. You study the cuts, search your recordings for a suitable action, trim it to fit, rebuild a title and adjust the sound. Then a longer reveal changes the timing of everything around it.

ChatGPT can help carry out that work when connected to a compatible video editor through MCP. The connection exposes editing controls so the assistant can inspect material and change the timeline. For this workflow, conversation alone does not reliably perform the edits.

You can choose any editor with a compatible MCP connection; setup and capabilities differ. But if you're just looking to get started the free invideo Editor MCP is a useful option. Its style and motion reference workflow supports adapting pacing, transitions, effects, typography, animation and sound to your footage on a professional-grade, frame-accurate timeline. The full manual editing suite stays available, so you can shape the same edit yourself and continue with ChatGPT.

Ordinary editing through invideo MCP does not consume credits, however your ChatGPT subscription and usage allowance apply separately. Stock sourcing by the invideo agent and generating images, video or audio use invideo credits.

Here’s how to use a reference to direct the choices you want ChatGPT to make, then refine them on your own timeline.

Connect ChatGPT to an editing suite

Open invideo’s connection page, choose ChatGPT and follow the setup instructions for your account. In ChatGPT on the web, open Plugins, select + and choose Add custom MCP server. Name it “invideo” and enter this server URL:

https://mcp.invideo.io/mcp

Complete the invideo sign-in, create and install the plugin, then select invideo in your editing conversation. OpenAI’s custom MCP setup guide explains the account and workspace requirements. Follow the linked instructions if your setup options differ. Keep invideo Editor open alongside ChatGPT for playback and manual editing.

Keep invideo Editor open beside ChatGPT to play the edit and make manual adjustments.

Bring in the footage and explain the film

Upload your recordings, supplied music, logo and closing card directly in invideo Editor. You can also attach files through ChatGPT on web or desktop and ask it to import them into the intended project. For an existing edit, give it the project link and exact timeline name.

Share a reference file or an accessible YouTube link, and explain what your video should communicate. Upload the reference if its link is private or access-restricted.

Give ChatGPT the project and creative context before asking it to edit. You can use this brief template:

Open [project link] and inspect [uploaded footage, music and graphics] for [timeline name]. We’re making a [duration], [aspect ratio] video about [subject] for [audience], intended to [purpose]. Use [reference link or file] as direction for [pacing, sequence, transition, effect, typography, animation or sound]. The key moments are [actions or messages], and the ending should [closing shot, message or call to action]. Preserve [approved choices]. First inspect the material and propose an approach; leave the timeline unchanged until I ask you to assemble it.

Choose what to recreate from the reference

A reference can guide your pacing, shot order, transitions, effects, typography, animation or sound. Tell ChatGPT which elements you want to recreate. It can study those choices and apply them to your footage, with editable controls for you to refine.

To recreate the pacing and rhythm

Suppose you’re cutting a night-run teaser and want the energy of Nike’s “Nothing Beats a Londoner”. Ask ChatGPT to compare the reference’s quick cuts with its longer holds, then shape your preparation details and launch shot around that rhythm. Identify the passage to study and make clear which actions need time to play out:

Study the cutting at 01:00–01:20 in https://www.youtube.com/watch?v=mEB1C59hCvs. Adapt the energy to the opening of “Run the Night v1”: quick laces, watch and breathing details, then a longer shot of the runner launching into the street. Find useful source ranges and keep the first stride understandable. Build the opening from our recordings, without inserting the reference’s footage, branding or audio. Preserve the rest of the timeline and its 40-second total.

Illustration: short preparation clips give way to a longer moving shot, so the launch has room to register.

Play the opening through the launch. The fast details should build anticipation without making the first stride unreadable. Ask ChatGPT to shift a cut or hold the action longer, or choose the exact frames manually.

To recreate the sequencing

For a pottery-studio film, you might use Brompton’s “Made in London” as a reference for showing workmanship. Ask ChatGPT to study how the reference connects details, actions and the finished object, then arrange your pottery footage so viewers can follow the cup taking shape:

Study the workshop details and actions at 00:00–00:15 in https://www.youtube.com/watch?v=tWKs6IQraEg. Build “Clay to Cup” from our pottery recordings: centre the clay, shape the cup, attach its handle, glaze it, then reveal the finished piece. Search the sources for clear moments of each stage and group the shots so viewers can follow the object changing. Hold the finished cup long enough to appreciate its shape and glaze. Use our footage and supplied sound, preserve the agreed duration, and flag missing stages instead of inventing them.

Illustration: the ordered timeline follows the material from raw clay to a finished cup.

Watch whether each stage makes the next one understandable. A close-up of glaze looks different once viewers have seen the cup it belongs to. ChatGPT can search your footage by action and arrange the relevant ranges; you can revise the order manually too.

To recreate a particular transition

Suppose your food edit cuts from a falling cherry tomato to a salad bowl. Daniel Schiffer’s lime-transition chapter at 02:49 provides a reference for connecting shots through a moving object. Ask ChatGPT to study the object’s position, scale and direction on either side of the cut, then find suitable frames at the join in your edit:

Study the lime-transition editing passage beginning at 02:49 in https://www.youtube.com/watch?v=AR2425w1X7k. Adapt the object-matching idea to “Fresh Salad”: cut from our falling cherry tomato to the tomato landing in the salad bowl. Find source frames that match its screen position, scale and direction across the cut. Keep the landing readable. Use our recorded coverage and editable transforms where needed; flag an unusable angle rather than generating a bridging shot. Change only this join, preserving all other cuts, text, sound positions and duration.

Illustration: the red tomato holds a similar screen position and scale across two different camera angles.

Play across the join at normal speed. The tomato should continue through the cut without a distracting jump. Adjust the source frames or transforms, or ask ChatGPT to refine only this join. The transitions guide explains the wider workflow.

To recreate a particular effect

For a night-drive film, Luuk Alleman’s built-in-agent example at 03:45 could serve as a reference for light trails. Ask ChatGPT to study how the effect follows the highlights and leaves the subject clear, then adapt it to your selected shot with controls you can refine:

Study the light-trail treatment and controls at 03:45 in https://www.youtube.com/watch?v=8T_xABE3VD8. On the selected tail-light shot in “After Hours Drive,” create short trails from the bright lamps and street reflections. Keep the car body clear and retain our colours. Give me editable controls for highlight threshold, softness, trail length, direction and mix. Apply the effect only to this shot, preserving its duration and audio.

Illustration: the same car frame is shown untreated and with short trails from bright highlights.

Compare the effect on and off. Reduce the mix or trail length if the car becomes obscured. Available controls depend on the effect built; the custom-effects guide covers refining the treatment.

To recreate typography and graphic layout

If you’re making a desert music-festival announcement, the opening of Apple’s “Don’t Blink” offers a reference for bold, high-contrast typography. Ask ChatGPT to study the text’s hierarchy, spacing and placement, then create a layout that fits your wording and footage:

Study the typography at 00:00–00:06 in https://www.youtube.com/watch?v=jk6sz25OZgw. Adapt its short wording and contrast to our “Desert Echo” festival announcement. Create a large stacked “DESERT ECHO” headline over the clear left side of the sunset-dunes shot, with “12 OCT · SUNSET STAGE” as a smaller supporting line. Use our supplied condensed font, cream text and burnt-orange accent on “ECHO.” Keep the headline and event information on separate editable layers. Keep the stage visible and preserve the footage, sound and duration. Leave animation out of this pass.

Illustration: headline scale, contrast and placement create a festival-poster hierarchy over the landscape.

Check the composition at the size viewers will watch. The event name should read first, then the date and venue. Move or resize the text manually, or ask ChatGPT to revise the spacing while preserving the approved wording, font and colours.

To recreate the animation

A skincare launch could draw on the opening word changes in Apple’s “Don’t Blink” for its title animation. Ask ChatGPT to study how the words appear, hold and change, then adapt that timing to your product message while giving the complete phrase time to read:

Study the word changes at 00:00–00:06 in https://www.youtube.com/watch?v=jk6sz25OZgw. On the serum-bottle shot in “Daily Ritual,” bring in “CLEAN,” then “CALM,” then “GLOW,” using a short scale-and-opacity entrance on each editable word layer. Keep earlier words visible as the phrase builds. Hold all three words together for three seconds, then fade them out. Use our approved sage-green layout and supplied font, keep the bottle visible, and preserve the product footage, sound and total duration.

Illustration: three states show the phrase building; staggered text clips share a complete-title hold.

Watch at normal speed. Each word should arrive clearly, and the complete phrase should hold long enough to read. Ask ChatGPT to delay a word or soften its entrance without changing the layout, or refine the animation keyframes yourself.

To recreate the sound design

When shaping the sound for a café opening film, you could take direction from Nike’s “Nothing Beats a Londoner”. Ask ChatGPT to study how the reference combines music, ambience and action sounds, then build a treatment around the recorded actions in your café footage:

Study 01:00–01:20 in https://www.youtube.com/watch?v=mEB1C59hCvs for the relationship between action sounds, music and pauses. On “Morning Coffee,” use our uploaded grinder, espresso, steam and cup-contact recordings. Keep the coffee-making sounds on A1, the cup-contact accent separately on A2, and supplied music on A3. Place the sounds against their matching actions, lower the music beneath the final cup landing, then let the ending settle. Preserve the picture edit and total duration. Use our audio rather than the reference’s audio; do not source or generate new sounds.

Illustration: separate sound layers and a music-level envelope show the proposed arrangement. Playback is needed to judge timing and balance.

Listen for whether the grinder, steam and cup accent land on their actions and whether the music change feels natural. Adjust a cue’s position or level, or make the mix decisions manually. The beat-sync guide develops timing against music.

Refine the treatment and export your edit

Review the treatment with its surrounding shots. Give ChatGPT a specific change and identify the decisions it should preserve. You can use the same structure for any of the examples above:

On [timeline name], revise [time range or named shot] so that [desired change]. Preserve [approved footage, layout, animation, sound or other choices]. Keep [duration and fixed cue points] unchanged. If the source material cannot support the change, explain which boundary needs to move. Report the revised cue times.

Play through the changed passage and its joins. If you make a creative adjustment manually, ChatGPT can continue from the updated timeline. The video-feedback guide covers directing a scoped revision.

Once the treatment works in the full edit, identify the timeline and export settings:

Export the current [timeline name] as a finished [duration], [aspect ratio] H.264 MP4. Preserve the approved footage, titles, transitions, effects, sound and ending. Keep the editable timeline available for further revisions.

The H.264 MP4 view provides settings for the finished video. Request an editable-project export separately for a handoff to another editor.

You can export manually too. Open the rendered file and check the revised passages, text, sound and ending before sharing it. The export guide covers finished-video and editable-project handoffs.

For the wider workflow, see editing videos with ChatGPT.

Frequently asked questions

Can ChatGPT match a reference video’s editing style?

Yes, for this timeline workflow it needs a compatible editor connected through MCP. In invideo Editor, you can direct it to study selected pacing, shot sequences, graphics, effects or sound and adapt them to your footage, then refine the editable result.

Do I need to identify every shot in my footage?

No. Describe the actions or details the edit needs, and ChatGPT can search material in the connected project. Review the selected ranges for complete actions and useful framing. You can identify the reference passage without having to mark every matching source shot yourself.

Can I borrow only a title animation or transition?

Yes. Give the timestamp in the reference, the destination in your current timeline and what should stay unchanged. Ask for the treatment’s useful controls to remain editable, then review it with the surrounding shots.

Will it recreate the reference exactly?

The reference guides editing choices using your footage and brief. Judge whether the pacing, treatment and structure serve your video. A different action may need a longer hold or a different shot grouping, even when the reference provides the creative direction.

Does this use invideo credits?

Ordinary timeline editing and footage analysis through invideo MCP do not consume invideo credits. Your ChatGPT subscription and allowance apply separately. Stock sourcing by the invideo agent and generating images, video or audio use invideo credits.

Can I keep editing manually after ChatGPT works on it?

Yes. invideo Editor’s full manual suite remains usable on the same timeline for cuts, text, keyframes, effects, colour and sound. ChatGPT can continue from the updated edit. Tell it which manual decisions are approved and should be preserved.