A script into a run of clips that cut together
Every shot gets a still first, because a still costs a fraction of a clip.
Your agent breaks the script into shots, generates one frame per shot on a single image model so they match, waits for you to approve the frames, and only then spends the video budget animating them.
You say: "Turn this 30 second script into shots and give me a clip for each one."
Three steps to have your agent running this
- Connect Bildy at
https://bildy.ai/mcpand sign in with your Bildy account. Setup per client. - Copy the text below and save it as
.claude/skills/storyboard-to-clips/SKILL.md, or paste it into whatever your agent reads for instructions. - Ask for what you want. Your agent reads the skill and follows it.
Bildy tools this skill calls
- images_generate
- images_get_job
- videos_make_storyboards
- videos_list_models
- videos_generate
- videos_get_job
- videos_extract_frame
- videos_compose
- videos_get_edit
The whole file
--- name: bildy-storyboard-to-clips description: Take a script, a product idea or a rough outline and turn it into a run of short clips that cut together, by generating a frame for every shot first and animating each one. Use when someone wants a short ad, an explainer, a trailer or any video made of several shots rather than one clip. --- # A script into a run of clips A video model gives you one shot at a time, usually four to ten seconds. A video with five shots in it is five generations that have to look like they came from the same camera. You get that by making all five still frames first, on one image model, and animating them afterwards. ## Why the frames come first An image costs a fraction of what a clip costs, and you get it in seconds instead of minutes. Getting the look right on stills, where a re-run is cheap, saves you re-running clips at twenty times the price. Five frames from one image model with one prompt style also match each other far better than five clips generated independently. ## Steps 1. Get the script, or write one with them. Break it into shots and keep each shot to one action. Five to eight shots is a sensible short video. 2. Write a style line you will paste into every shot prompt: the film stock, the light, the colour, the lens. Something like "shot on 35mm, overcast daylight, muted greens, shallow depth of field". 3. Generate one frame per shot with `images_generate`, each prompt being the style line plus that shot's description. Use the same model and the same `ratio` for all of them. Show the person all the frames together. 4. Fix whichever frames they do not like before you animate anything. 5. For each approved frame, call `videos_generate` with the frame's link as `imageUrl` and a prompt describing only the motion in that shot. 6. Poll `videos_get_job` on each. Clips take minutes, so tell them where you are rather than going quiet. 7. Join the clips into one file with `videos_compose`, passing the links in shot order. Poll `videos_get_edit` until it is done and hand over the finished video. Links expire after an hour. If they would rather cut it themselves, skip step 7 and hand over the clips in shot order with the shot number on each. ## Planning a one-clip video on a storyboard If the whole video fits in one generation, you can plan its shots on a storyboard and skip the separate frames. Call `videos_make_storyboards` with the prompt, the clip length and its shape. Poll `images_get_job`, show the person the storyboards and let them pick one. Then call `videos_generate` with the same prompt and the chosen storyboard's url as `storyboardUrl`. The model follows the panels in order, so one clip can hold two or three shots with cuts between them. We send the storyboard to the model as a reference picture, so pick a model with `maxReferenceImages` above 0 in `videos_list_models`. The storyboard's link expires after an hour, like every other link here. ## Costing it up front Call `videos_list_models` and multiply the per-second price by the total seconds before you generate anything. A five shot video at six seconds a shot is thirty seconds of video, and the difference between a cheap model and an expensive one across thirty seconds is large. Tell them the figure and let them pick the model. ## Making consecutive shots carry on from each other Two clips generated from the same style line still jump at the join, because the second one invents its own opening frame. To have one carry on from the last, take the closing frame of the first with `videos_extract_frame` and pass its link to `videos_generate` as the `imageUrl` for the second. The second clip then starts on the frame the first one ended on. Do this where the same character or the same room appears on both sides of a cut. Do not do it where you want a cut to a different place, because there the jump is the point. ## Cutting between shots Give consecutive shots different framings, wide then medium then close. Two clips at the same distance cut badly. Keep camera motion in one direction across a sequence. A push in followed by another push in reads as one move; a push in followed by a pull back reads as a mistake. Leave a beat at the start and end of each clip with little motion. An editor needs somewhere to cut, and a clip that is moving hard at both ends is hard to place.
Other skills for the same tools
Product photos from one picture
A photo you took on a desk becomes a set of clean shots for a shop listing.
Ad variants in every placement size
Your agent composes each crop separately, with room left for the headline.
A clip from a picture you already have
A still becomes frame one and the model carries on from there.
The same prompt on several models at once
Your agent runs one prompt on four models and lays the results out with what each cost.
A video longer than one generation
Each clip starts on the frame the last one ended on, so the run carries on instead of jumping.