Cindy Zhu.
← all free guides
Content creation Multi-tool

Edit a whole video with AI (Arcads + Google Omni 1.1)

Hey, it's Cindy 🌱 Here's the full prompt from the reel. One tripod shot goes into Arcads, and Google's Omni 1.1 adds the paper texture backgrounds, the handwritten title, the brown captions, the paper graphics and every camera move. πŸ‘©β€πŸ’»
the setup

The whole thing in 6 steps πŸ—ΊοΈ

  1. πŸŽ₯ Film one clean take One shot, one tripod, one angle. Keep the camera still, use good light, and leave a little space above your head for the title to pop into.
  2. πŸ†• New Project, then Video In Arcads, click New Project, then click Video.
  3. ⚑ Choose the Omni 1.1 Flash model This is the model that does the camera work and the on-screen elements, so pick it before you paste anything.
  4. πŸ“‹ Paste in your clip and the prompt Upload your clip and paste the full prompt below. Swap the example title and topic for your own.
  5. πŸ’‘ Make lots of drafts at 360p Drafts at 360p cost about a third. Every generation comes out different, so make several and compare them.
  6. ✨ Upscale the keepers, then stitch Upscale only the drafts (or the parts of drafts) you like to 4K, then cut the best moments together in your editor.

the full prompt

The paper scrapbook prompt πŸŽ₯

This is the look from my reel: paper texture backgrounds, a handwritten title that pops up at the start, brown captions at the bottom, paper cut-out graphics and camera angle changes. Copy the whole block, then change the two examples in quotes to your own title and topic.

πŸ“œ the full prompt
Edit this clip into a finished, high-quality short-form video in a warm paper scrapbook style. Keep my face, my voice, my words and my timing exactly as they are. Do not smooth my skin, change my features or change my clothes.

Background: switch between two looks. In some shots, keep my real room. In others, cut me out cleanly and replace the background with a cream paper texture: soft visible paper grain, torn paper edges in the corners in beige, tan and charcoal, and small cut-paper shapes (circles, stars and triangles) around the edges of the frame. It should feel like a handmade scrapbook page, matte and tactile, never glossy or digital.

Title: in the first second, a big handwritten title pops in above my head with a quick, bouncy pop: thick dark brown marker letters with a cream outline, saying "Did you KNOW?". For the whole video, keep a small strip of torn notebook paper taped across the top of the frame with the topic written on it in handwritten marker: "Omni flash 1.1".

Captions: brown captions at the bottom. A rounded pill in warm caramel brown, centred in the lower third, with one to three words at a time in a white, rounded handwritten font. The words match exactly what I am saying and appear in sync with my speech. Keep the pill the same size and position in every shot.

Graphics: when I mention something, show it as a paper cut-out that pops in beside me, never covering my face. For example: paper bar charts, line charts and donut charts on little paper cards; a paper laptop with a short phrase on its screen; big chunky cut-out letters for key words, in dark brown with a cream outline. Each graphic pops in on the word it relates to and clears before my next point.

Camera: create camera moves that were never filmed. Mix medium shots, close-ups and a slightly angled side shot, with a slow push in on my most important line. Change angle only on a natural pause in my speech, and keep my face sharp and in frame the whole time.

Colours: cream, tan, caramel brown, terracotta and charcoal everywhere, so every shot feels like the same scrapbook.

Quality: clean edges around my hair and hands, no flicker, no warping, no extra fingers, and consistent lighting from shot to shot. Every element should look deliberate and polished.
🌱 Make it yours. Change "Did you KNOW?" to your hook, and "Omni flash 1.1" to your topic. Keep everything else the same so the style holds.

scene by scene

One prompt per scene 🎬

If the full prompt gives you a mix you don't love, generate one scene at a time instead. Paste the first paragraph of the full prompt (the line about keeping my face, voice and words), then add one of these.

✍️ the hook
Keep my real room. In the first second, a big handwritten title pops in above my head with a bouncy pop: thick dark brown marker letters with a cream outline, saying "Did you KNOW?". Brown caption pill at the bottom with white handwritten words in sync with my speech.
πŸ“Š paper charts
Replace the background with a cream paper texture with torn paper edges and cut-paper circles and stars. Move me to the right side of the frame in a medium close-up. On the left, three paper cards pop in one after another: a bar chart, a line chart and a donut chart, in tan, caramel and charcoal. Brown caption pill at the bottom.
πŸ”€ big cut-out words
Replace the background with a cream paper texture with torn paper edges, cut-paper stars and circles. Frame me as a close-up on the left. On the right, the words "TEXT EFFECTS" pop in as big chunky cut-out letters in dark brown with a cream outline, slightly tilted. Brown caption pill at the bottom.
πŸ’» paper prop
Keep my real room. Above my head, a paper cut-out laptop pops in with "edited 100% by AI" written on its screen in simple black type. Keep it clear of my face. Brown caption pill at the bottom.

the real process

What it actually takes 🧠

This isn't one click, and I want you to know that before you start.

πŸ’‘ Lots of draftsI made a lot of drafts at 360p, because each one comes out different. Drafts cost about a third, so this is where you experiment.
✨ Upscale only the keepersPick the drafts you actually like and upscale only those to 4K. That's where the real credits go.
βœ‚οΈ Stitch the best partsOften one draft nails the hook and another nails the charts. Cut the best moments from different generations together in your editor.
πŸ› οΈ Expect some hand tweaksI still fixed a few things by hand at the end, like a caption or the timing of a cut. AI does the heavy lifting, you do the final polish.

camera moves

Camera moves you can swap in πŸŽ₯

The movement is whatever you ask for. Replace the Camera: paragraph in the full prompt with any of these.

πŸ”™ pull back to reveal
Camera: start close on my face, then slowly pull back to reveal the whole room, ending on a wide shot with me still centred and sharp. One smooth move, no cuts during it.
πŸ”„ slow circle
Camera: slowly circle around me in a half orbit, from my left side to my right, keeping my face centred and in focus the whole way. Keep the camera at eye level and the speed constant.
⬆️ overhead
Camera: rise smoothly from eye level to a high overhead angle looking down at me, then hold. Keep my face visible and readable at the top of the move.

too short?

Extend a clip that runs short ⏩

If your clip is too short, tell it to keep going. Omni 1.1 holds 10 seconds of memory instead of one, so a shot can run up to 40 seconds and still hold together.

⏩ keep going
Keep going: continue this shot from its final frame for another [8] seconds. Keep the same camera move, speed, framing, lighting, background and graphics style so the join is invisible. Keep my face, clothing and the room exactly as they are.

quick fixes

When a draft is nearly right πŸ› οΈ

Leave the rest of the prompt alone. Add the line that matches what went wrong and regenerate.

πŸ’¬ the captions are off
Make the captions match my exact words, one to three words at a time, appearing as I say them. Keep the brown pill the same size and in the same place in every shot.
πŸ“œ the background looks digital
Make the paper background more physical: visible paper grain, uneven torn edges with soft shadows, matte colours. It should look like real paper photographed on a desk.
πŸ™‚ it keeps changing my face
Treat my face as locked reference. Do not smooth, slim, relight or alter my features in any frame. If a camera move would require inventing detail on my face, choose a smaller move instead.
🐌 the moves are too fast
Slow every camera move down by half. The camera should drift rather than travel, like a tripod head being panned gently by hand.

The links πŸ”—

🎬 Arcads: arcads.ai

Follow @cindiezhu for more AI tips every single day 🌱