Skip to content
Tutorialbeginner9 steps · 18 min7 prompts

Shoot a 60-second sci-fi short with Kling 3.0 and Nano Banana 2

From a one-line idea to a graded, scored 60-second film. Build a character sheet, generate six shots, add dialogue and finish in CapCut.

Workflow

Still → Motion → Edit

Lock the look with stills first, animate each shot, then assemble. The backbone of every narrative tutorial on this site.

  1. 01 IdeateChatGPTLogline, beat sheet and shot list.
  2. 02 LookNano Banana 2Character sheet and one still per shot.
  3. 03 MotionKling 3.0Image-to-video, 8-second clips, one camera verb each.
  4. 04 EditCapCutAssemble, grade, export vertical and horizontal.

This is the guide most people start with. You will finish with a 60-second film you can post today.

Step 1 of 9

Write the logline

Start with one sentence. Who wants what, and what stands in the way? Generate five and pick the one you can shoot in six shots.

ChatGPT
Write five loglines for a 60-second [GENRE] short film. Each must name a protagonist, a want and an obstacle in one sentence. Tone: [TONE].
Step 2 of 9

Build the character sheet

A character sheet keeps faces consistent across every shot. Generate it once, upscale it, and reuse it as a reference image for the rest of the film.

Nano Banana 2Open
Character sheet of [YOUR CHARACTER], front, three-quarter and profile view, neutral grey background, soft studio light, consistent features, [STYLE: photoreal / anime / claymation], high detail --ar 3:2

Upscale the sheet and reuse it as a character reference in every later shot.

Step 3 of 9

Generate the establishing shot

The first shot sets the world. We generate a still in Nano Banana 2 first so we can lock the look, then bring it to motion in Kling 3.0.

Open the image generator, pick Nano Banana 2 and paste the prompt below. Change the bold variables to match your logline from step 1. Generate four, upscale the one with the cleanest horizon line.

Nano Banana 2Open
Wide establishing shot, [ABANDONED DRIVE-IN THEATER AT DUSK], single sodium lamp, 35mm anamorphic, volumetric haze, cinematic color grade, subtle film grain, no text --ar 16:9 --style raw --v 7
Kling 3.0Open
Slow dolly-in over 8 seconds, [DUST DRIFTING THROUGH THE LAMP LIGHT], screen flickers once, ambient wind, no dialogue, no camera shake
Step 4 of 9

Shots 2–6: coverage

Repeat the still-then-motion loop for the remaining five shots. Keep every camera move to one verb so the cuts feel deliberate rather than restless.

ChatGPT
Break this logline into five shots after the establishing shot: [YOUR LOGLINE]. For each, give shot size, camera move (one verb), and what changes in the story. No dialogue yet.
Step 5 of 9

Dialogue in Kling 3.0

Kling 3.0 generates lip-synced dialogue and native sound in the same pass. That is the hour-saver, so write for it: short lines, one speaker per clip.

Kling 3.0Open
Close-up, [CHARACTER DESCRIPTION], [SETTING BEHIND THEM], they say: "[LINE UNDER EIGHT WORDS]". Soft key light, shallow depth of field, handheld micro-movement

Keep spoken lines under eight words; longer lines drift out of sync.

Step 6 of 9

Score with Suno

Generate a 90-second instrumental so you have room to cut. Lay the music first and cut picture to it.

Suno v6Open
Instrumental only, [MOOD: tense / wistful / triumphant], [INSTRUMENTATION], 90 seconds, slow build with a drop at 0:45, no vocals, no drums until 0:30
Step 7 of 9

Assemble in CapCut

Drop the score on the timeline, then cut the six clips to it. Trim heads and tails hard: eight-second generations rarely earn eight seconds on screen.

Step 8 of 9

Color grade

One LUT across every shot, then match exposure by eye. Consistency reads as intent; per-shot grading reads as noise.

Step 9 of 9

Export for TikTok & YouTube

Export 9:16 and 16:9 from the same timeline. Reframe by moving the safe area, not by re-generating shots.

Prompt pack

Every prompt in this post

7 prompts, in the order they are used. Change the [VARIABLES] to match your idea.

  1. ChatGPT
    Write five loglines for a 60-second [GENRE] short film. Each must name a protagonist, a want and an obstacle in one sentence. Tone: [TONE].
  2. Nano Banana 2Open
    Character sheet of [YOUR CHARACTER], front, three-quarter and profile view, neutral grey background, soft studio light, consistent features, [STYLE: photoreal / anime / claymation], high detail --ar 3:2

    Upscale the sheet and reuse it as a character reference in every later shot.

  3. Nano Banana 2Open
    Wide establishing shot, [ABANDONED DRIVE-IN THEATER AT DUSK], single sodium lamp, 35mm anamorphic, volumetric haze, cinematic color grade, subtle film grain, no text --ar 16:9 --style raw --v 7
  4. Kling 3.0Open
    Slow dolly-in over 8 seconds, [DUST DRIFTING THROUGH THE LAMP LIGHT], screen flickers once, ambient wind, no dialogue, no camera shake
  5. ChatGPT
    Break this logline into five shots after the establishing shot: [YOUR LOGLINE]. For each, give shot size, camera move (one verb), and what changes in the story. No dialogue yet.
  6. Kling 3.0Open
    Close-up, [CHARACTER DESCRIPTION], [SETTING BEHIND THEM], they say: "[LINE UNDER EIGHT WORDS]". Soft key light, shallow depth of field, handheld micro-movement

    Keep spoken lines under eight words; longer lines drift out of sync.

  7. Suno v6Open
    Instrumental only, [MOOD: tense / wistful / triumphant], [INSTRUMENTATION], 90 seconds, slow build with a drop at 0:45, no vocals, no drums until 0:30