Deep Focus Prompts: Two Actions in One Drama Frame

A shopkeeper repairs a lantern as another person hangs its replacement

A shopkeeper mends a torn lantern by the counter while another person hangs a replacement outside. Each action explains the other. Cut to a close-up of the torn paper too early and the replacement looks like an unrelated decoration. A deep-focus composition can give the viewer enough room to read both actions in the same frame.

Build a short-film frame with two readable actions

Use the lantern workshop to plan a layered still, then decide whether to animate it. The sketches and prompts here are original preproduction examples. Approve each actual frame on its own.

Keep near and far readable without splitting the image

Start by deciding what needs to remain clear together. In this exercise, one frame has to show a repair and the temporary solution happening at the same time.

Deep focus versus shallow focus

The Columbia Film Language Glossary defines deep focus as keeping near and distant planes sharp within one composition. A shallow-focus treatment might isolate the shopkeeper’s fingers and turn the doorway into soft color. That treatment can be useful later, when the paper tear is the whole point. Here, the doorway carries an action the viewer needs in the same moment.

One continuous space rather than two panels

A near action and far action can share a single floor, doorway and source of light. Sketch the opening as part of the room, with the counter on one side and the person outside visible through the opening. Keep the aisle as continuous floor. Avoid an arrangement that reads as a second image pasted into the background.

Requested appearance versus simulated optics

In a generated still, write down the appearance you want. For example, the near fingers, torn seam, and distant hands should all have edges a viewer can see. A generator may produce those details unevenly. Describe the visible outcome and inspect the frame at the size it will be viewed. The exercise borrows a cinematographic idea for composition rather than setting a real lens or aperture inside the product.

A still can be sharp in the background and still hide the second action. The person may be too small, covered by a foreground shoulder, or surrounded by objects with the same shape. Put these checks after the focus request. Approving the image requires being able to read the raised arm and the hook together, not just recognizing that there is somebody near the door.

Place two actions that are worth seeing together

Both tasks should be needed to reopen the shop. The frame shows two people solving different parts of one practical problem.

A foreground task

On the near left, Mei presses a short strip of paper across the damaged lantern. Show her fingers resting on the join and make the tear large enough to identify. She can take up a good part of the frame while keeping one shoulder away from the center. The viewer should be able to point to the repair before reading an explanation.

A distant response happening at the same time

At the far right, Arun lifts a spare lantern to a hook outside the opening. Angle one forearm toward the hook and leave the other hand visible under the lantern. The lift gives the replacement a purpose. If the spare is already hanging, the viewer may see only a decorated doorway and miss Arun’s contribution.

An empty middle distance that separates the silhouettes

Draw three horizontal rows to plan a vertical frame. Bottom and near left: Mei, torn lantern, fingers on seam. Middle: a strip of open flooring, with no display stand or third performer. Top and far right: Arun and the spare lantern at the hook. Trace one path from the seam across the empty aisle to the raised hand. The aisle gives the two outlines room to read.

Label the distance between the seam and the raised hand with the quiet part of the image that separates them. In the proposed workshop, that is an open aisle with two floor lines leading outside. The floor lines can guide the eye without competing with either hand. Keep any bright sign out of that gap. The aisle should connect the tasks and make it easy to move attention from one to the other.

Spend contrast and size where the second action needs them

An early sketch can look clear on paper and turn into a knot at phone size. StudioBinder’s breakdown distinguishes arranging depth from controlling focus. Compare these two arrangements before writing the final prompt.

Keep the two versions comparable. Put the torn lantern and the spare at the same moment in the story, then alter their arrangement. If you also change what each person is doing, a viewer’s answer tells you little about which composition worked. Mark the doorway with a simple rectangle so you can judge the effect before adding surface detail.

A clean background silhouette

In the crowded sketch, the open door sits directly behind Mei’s shoulder. The spare lantern overlaps a rack of similar shapes. Move the opening to the right in the edited sketch. Give the raised arm a quiet wall or dark doorway as a backdrop. That change makes it possible to distinguish the person from the objects around them without brightening the whole background.

Try showing a reviewer only the crowded sketch and asking them to mark where the lantern will go. Then show the edited sketch and ask the same thing. In a real comparison, record which sketch they saw first because the first drawing can teach them what to look for in the second. The sketch is a planning aid. Use a final crop to judge the actual display.

A gap around the important hand action

Leave a little clear area around the fingertips where the paper joins. Remove the second tool or coil of string if it competes with the gesture. If the seam still disappears, adjust its position before enlarging the entire lantern. In a narrow crop, a small shift in the hand can restore the action while preserving the far plane.

A foreground object that does not fill the frame

A giant round lantern fills the lower center and can hide the hook above it. Shrink the near prop or place it farther to the left. Keep the near prop large enough to see the torn seam and the spare large enough to see the lift. This relative-size note is easier to share than a lens number when the image is going to be drafted from a prompt.

Write the layers in the order the eye should find them

Give the near action one clause, the far action a second, and the shared space a third. Keep the initial still simple enough to review before adding motion.

Near subject and its task

Try this in the APOB short-film workflow with an original reference you can use: “Vertical view inside a small paper-lantern workshop. Near left, an adult shopkeeper presses a patch across the torn seam of an unbranded paper lantern, with her fingers clearly visible.” The placement keeps the repair on one side and leaves room for the second event.

Far subject and its readable gesture

Continue the same prompt: “Far right, through the open frontage, another adult holds a replacement lantern up toward a hook, both hands and the lift easy to read. Keep an empty aisle between the tasks. Both actions have clear outlines and readable detail.” The verb “holds up” gives a reviewer something to look for beyond the presence of a second character.

A shared environment with restrained motion

Once the frame reads at phone size, use APOB Image to Video for a restrained follow-up: “Keep the approved framing. The shopkeeper presses the paper patch once as the person outside raises the spare lantern a little. The aisle stays clear and the camera stays still.” Inspect the generated motion where either hand crosses a prop. Compare it with the approved still and save the exact prompt and take.

If the first still puts both people at the same distance, revise the layout clause before changing the style. Keep the counter, doorway, and eye-level perspective stable. If the still reads but the video swallows Arun’s hand during the lift, shorten the movement or give him a second shot. Working from one approved reference gives you a stable picture to compare against when a movement creates the problem.

Simplify the frame when the far action becomes decoration

Use a blank check card before sharing a frame. Enter frame ID ___, near action named ___, far action named ___, obstructing edge ___, and next edit ___. Ask someone to look at the same crop you plan to deliver.

The background figure is only a blur

Ask “What is each person doing?” before pointing to either lantern. If the answer is “somebody is standing outside,” check the raised hand at final display size. Clear away a conflicting shape or adjust the opening. If the arm remains soft, return to the still and establish a more readable background reference before spending effort on movement.

An enlarged foreground masks the second task

If the near lantern blocks the doorway, move it to the side or choose a different crop. Inspect the seam and hook together after the change. An enlarged near object can make one action more impressive and the other less intelligible. The useful test is whether the viewer can describe both within the same picture.

On a phone, look at the exact part of the frame that carries the small action. In this scene, the viewer should see where the spare lantern meets the hook. If you have to pinch to zoom to see it, decide whether to bring the doorway closer or move the hook away from the other shapes. Run the two-action question again after each material change.

Separate coverage when both actions cannot read

If both gestures cannot be read together, take separate coverage. Establish the room in a wide shot, then show the repair and the lift with matching direction and visible props. Keep the two tasks concurrent in the edit. The APOB guide to linking short-drama story beats can help when a shot change also changes when the audience learns a fact. The check card stays empty until someone has actually reviewed a frame.

Use the same check after any crop change. A vertical cut may need the near hand lower in the frame and the far lantern closer to the center. An uncropped still can pass while a thumbnail hides the only evidence of the second action. Save the delivery crop beside the working frame and label the version a reviewer actually saw.

Related reading: Anamana Studio Opens: Keep Script Direction Intact

Sources

HappyHorse 1.1 vs HappyHorse 1.0: Feature Analysis for AI Video GenerationAPOB AI generated AI imageSoft gradient background of the APOB AI call-to-action banner

Create a dreamlike vision with APOB

No credit card needed

CREATE FOR FREE

You may also be interested in