How to use Starpop and Nano Banana Pro contact sheeting to create cohesive AI videos

Starpop is an AI content tool that lets you turn Nano Banana Pro contact sheets into cohesive, shippable video sequences in one place. This guide walks through the exact Starpop workflow for generating Nano Banana Pro grids, sending them to Kling for image-to-video, and assembling them into performant AI ads that feel like a single real shoot.

The Logic of Contact Sheets and Nano Banana Pro

Making a single image look professional is straightforward, but turning that one look into a sequence that feels like a continuous moment is difficult. Without continuity, the result often looks like unrelated renders stitched together. A contact sheet solves this by forcing the model to reason about multiple frames in a single pass. This process ensures the wardrobe stays locked, the light direction remains stable, and the set geometry reads consistently from every angle. When those frames are animated, the motion feels observed rather than artificial.

Nano Banana Pro (NBP) is essential for this workflow because it can reason across the entire grid simultaneously. Instead of generating and matching frames one by one, you ask NBP to inventory everything that must not change. It then places the camera around the subject using per-frame logic. This global awareness allows lens choice, angle, depth of field, and composition to evolve from frame to frame without contradiction. This consistency makes the contact sheet a dependable bridge from stills to image-to-video (I2V) animation.

Contact sheeting involves generating six to nine keyframes arranged as a single composite grid with strict continuity across panes. Each pane represents a resting frame, which is an end position after an implied camera move, rather than an animation. Because the grid is treated as one image, NBP maintains identity and environment while intelligently varying camera placement. Two layouts are particularly useful. A 2×3 fashion or editorial grid suits a single subject where dynamic camera placements and detail cutaways are required. The panes share the same aspect ratio, usually 3:2 or 1:1, and function like one lookbook scene. A 3×3 cinematic storyboard provides nine frames that mirror standard film set coverage. The top row covers wide to knee-up shots, the middle row focuses on performance, and the bottom row handles detail and angle work. Both approaches rely on the same core concept demonstrated here: a studio fashion portrait with an oversized blazer and hard flash.

Essential Tools and Assembly

The tool stack is straightforward. Starpop, an AI content tool, serves as the orchestration layer for both the contact sheeting and the video generation. Use NBP to generate the contact sheet and perform any targeted refinements. Optionally, wire the process inside Node Banana to keep the grid, slice, and export steps repeatable. Kling turns the stills into motion. Use Kling 2.6 on the web for the best quality and first-plus-last-frame guidance, or use the Kling 2.5 API for batching and throughput control. Clips are assembled, easing is added, and subtle sound is layered using easypeasyease.

Generate contact sheets at 2K to 4K to ensure each pane holds sufficient detail. Keep square grids for 3×3 layouts and consistent 3:2 or 1:1 panes for 2×3 layouts. Helpful resources include Willie Falloon’s walkthrough for fashion and editorial contact sheeting, the cinematic 3×3 guide from Nanoprompts, the Node Banana repository, the Kling web UI, and easypeasyease.

Establishing Continuity

Before generating, decide what elements cannot drift and force NBP to enforce them. Wardrobe and styling are the first priority, including the exact blazer silhouette, fabric sheen, belt hardware, glasses, jewelry, and shoes. Hair and makeup follow, specifically parting, flyaways, and the exact reflectance desired on the skin. Light direction and quality must remain stable. If a hard on-axis flash with slight falloff is used, it needs to persist across all frames without mysterious fill light appearing.

Set geometry and environmental markers anchor believability. In a studio setting, this might be the position of the seamless background, a standing apple box, or a visible corner of a softbox reflection in the glasses. The color grade should be a constant. A Velvia-style saturation with grain and a slight magenta bias creates a completely different atmosphere than a cool cyan lift. The aspect ratio needs to be decided in advance so each pane aligns with the final deliverable. Finally, bake camera logic into the continuity plan. Focal length and subject distance should explain the depth of field and background compression. If a side-on profile uses a longer lens, the background should compress and the depth of field should deepen. Conversely, close portraits should bloom into shallow depth of field without changing the light direction.

Workflow A: Fashion and Editorial 2×3

1. Base image generation. Start with a decisive still that defines the look. For this example, prompt a full-body low-angle studio portrait using a 55mm lens on full-frame at f/4 for moderate separation. Use a hard on-axis flash with a subtle edge falloff, Velvia-style saturation, and visible grain. The model glances past the camera with a bored, raised-brow expression and one hand tapping the aviators. Frame at 3:2 for editorial flexibility and let the oversized blazer dominate the silhouette.
2. Contact sheet generation. Ask NBP to silently inventory identity, wardrobe, materials, hair and makeup, light direction and quality, environment geometry, color grade, and aspect ratio. Instruct it to keep all these elements unchanged. Describe only the six end positions rather than the moves. Request an intimate beauty close-up, a high-angle three-quarter shot from overhead and off-center, a low oblique full-body shot that elongates the legs, a side-on compression frame using a longer focal length, an unexpected-height portrait, and an extreme detail cutaway on the blazer’s belt hardware from a non-intuitive angle. One styled still becomes six coherent frames with identity, set, light, and grade locked while the camera explores the space. [Figure: 2×3 contact sheet of the example look].
3. Slicing and labeling. Extract each pane cleanly without resampling. Preserve the 3:2 ratio and name files in order to reduce friction later, such as A01_CU_Beauty.jpg through A06_ECU_BeltDetail.jpg. Keep a duplicate at original resolution for image-to-video (I2V) and a working copy for notes.
4. Targeted refinements. Use NBP for small, local adaptations that respect the locked elements. If the beauty close-up needs a cleaner cheekbone highlight or the detail frame should rotate five degrees, run a targeted pass to adapt that pane with explicit constraints. Specify the same wardrobe, light direction, grade, and face. Avoid global re-prompts that tempt NBP to reinterpret the scene.
5. Image-to-video in Kling. On Kling 2.6 web, load first and last frames for each shot when bookends are available and ask for a barely-there move. Favor gentle boom or pan language while keeping subject motion minimal and deliberate. For larger batches, the Kling 2.5 API provides better throughput to process a folder of panes overnight. Restraint is key, so hold expressions, limit blinking, and avoid big focal length shifts mid-clip.
6. Assembly. Trim the heads and tails to hide artifacts and apply easing so moves start and stop naturally. Set an ease-in and ease-out curve on each clip in easypeasyease so the motion breathes. Lay in a dry room tone and a faint fabric rustle to give the sequence air. Play it through to verify the illusion of six resting positions around one subject in one studio observed over a few seconds.

Workflow B: Cinematic 3×3 Storyboard

1. Analysis and panel generation. Keep the subject, environment, and light behavior consistent. Ask NBP for a 3×3 storyboard that follows classic coverage. Row one establishes the space with an Extreme Long Shot, Long Shot, and Medium Long Shot. Row two captures performance with a Medium Shot, Medium Close-Up, and Close-Up. Row three adds detail and drama with an Extreme Close-Up, a low-angle hero frame, and a bird’s-eye high angle. The wardrobe, light direction, set geometry, and grade remain identical to the 2×3 grid.
2. Slicing and naming. Use shot-type codes in filenames to speed sequencing and notes, such as B01_ELS.jpg through B09_High.jpg. Place them in a B bin within the project and maintain a spreadsheet linking pane numbers to intended moves.
3. Animation. Midjourney’s short animation feature serves as a stylized option for painterly motion or graphic transitions, particularly on the Extreme Close-Up and Close-Up where texture carries the frame. Use Kling for photoreal continuity and smoother camera moves across the grid. Keep motion restrained with slow dolly-ins on the Medium Shot and Close-Up, a fractional tilt on the high angle, and a gentle ground-up boom on the low angle. Avoid whip pans or rack focuses that break studio logic.
4. Editing. Assemble wide to tight to sell space before performance, or cluster by theme for a rhythm of alternating angles. Use match cuts on geometry, such as the blazer’s shoulder line or the glasses rim, to thread frames together. A minimal sound bed of room tone, a soft shutter click, or one bass swell per transition adds realism without calling attention to itself.

Adaptable Prompt Templates

Use the following prompt for the initial look of a single styled still:

Create a high‑fashion studio portrait of one model wearing an oversized charcoal blazer with strong shoulders and silver‑rimmed aviators, on a saturated cobalt seamless. Low angle, full body. 55mm on full frame, f/4, hard on‑axis flash with edge falloff. Velvia‑style saturation, visible grain, rich primaries, slight magenta bias. The model glances past camera, bored expression, one hand touching the glasses. 3:2 aspect ratio; clean floor sweep; no props beyond the seamless.

Use this prompt for the NBP 2×3 fashion contact sheet:

Silently analyze the input still; lock identity, exact wardrobe, hair/makeup, set geometry, lighting direction/quality, color grade, and aspect ratio. Output one 2×3 contact sheet—six panes with identical style/grade. Only describe final frames, not motion:
1) Beauty CU near face, slight high or low offset, shallow DoF.
2) High‑angle three‑quarter from overhead, off‑center.
3) Low oblique full‑body, elongating silhouette and shoes.
4) Side‑on long‑lens compression (profile or near‑profile).
5) Intimate portrait from unexpected height (editorial offset).
6) ECU of belt/glasses fabric detail from a non‑intuitive angle.
DoF must vary naturally with focal length and distance; lighting and environment are unchanged.

Use this prompt for the NBP 3×3 cinematic storyboard:

Analyze the input image and keep subjects, wardrobe, lighting, environment, color grade, and aspect ratio strictly consistent. Generate a 3×3 grid covering:
Row 1: ELS, LS, MLS (knee‑up).
Row 2: MS, MCU, CU.
Row 3: ECU, Low angle (worm’s eye), High angle (bird’s eye).
Maintain correct focal length logic and realistic DoF; photoreal textures; identical palette and light direction across all nine frames.

Use this compact I2V motion prompt for Kling:

Very slow, smooth camera move; subject remains mostly still and intentional. Prefer subtle boom or pan; no abrupt racks, no zooms, no dolly zoom. Keep focal length constant; hold expression; minimal hand micro‑adjustments only. Ease in for the first 15% and ease out for the last 15%. Preserve lighting direction and grade; avoid flicker and geometry drift.

Quality Control, Iteration, and Delivery

If the wardrobe shifts, such as an extra belt loop appearing or the blazer lapel changing width, re-run NBP on the offending pane with a hard constraint to replicate the original garment silhouette, stitching, and closures exactly with no additions or removals. Feed it the base still as a reference alongside the target pane so it can reconcile differences. For accessories like glasses, explicitly mention material and reflection behavior to freeze specular highlights. Identity drift requires tighter facial anchors. Add anatomical descriptors or use a face-locked reference crop from the base still. In NBP, call out a perfect facial match to the reference with no reinterpretation of age, ethnicity, or expression. Keep pose notes minimal since micro-poses are safer than new gestures.

To enforce consistency in set geometry and light direction, put geometry into words. Note that the cobalt seamless meets the floor at one-third of the frame from the bottom, or that the on-axis hard flash comes from the camera with slight vignetting at the edges. If the I2V introduces moving shadows or background shifts, shorten the clip and increase easing. Flicker usually hides when the move is gentle, duration is 1.5 to 2.5 seconds, and rack focus is avoided. If softness appears, regenerate the contact sheet at the higher end of the 2K to 4K range rather than upscaling. Aggressive upscaling before I2V can amplify artifacts that Kling will then animate.

When delivering, work backward from the platform. A 3:2 source grid can finish as 9:16 vertical if panes are composed with safe lateral headroom. A 1:1 ratio is easier if the destination is a square carousel. Decide the final aspect ratio before generating so NBP composes each pane accordingly. Stick to 23.976 or 24 frames per second for a cinematic feel, or 30 frames per second for snappier social pacing. Duration per shot typically lives between 1.5 and 3 seconds. Name files so the timeline reads like a shot list to keep order and intent visible. Subtle room tone, a remote shutter chirp, or a light fabric creak finishes the illusion. Eased heads and tails on short clips hide I2V hiccups and make restrained moves feel natural. [Figure: timeline showing eased camera moves].

Scaling the Process and Case Study

Node Banana turns a personal recipe into a reusable pipeline. Build a template that ingests a base still, runs an NBP contact sheet prompt, slices panes to files, and exports both originals and working proxies. Commit it so collaborators can drag the graph and match settings without guesswork. Kling 2.6 on the web offers the fastest way to get high-quality results on a few selects, while the Kling 2.5 API provides throughput control for dozens of shots. Standardize shot lists and prompt skeletons so the team shares the same end-position vocabulary, continuity locks, and I2V motion language. A small style guide prevents unintentional drift. Inside Starpop, this same recipe becomes a saved workflow you can rerun on any new base still—drop in your image, and Starpop handles NBP prompting, slicing, and handoff to Kling so the contact sheet and video chain stays consistent.

In a practical example, begin with the styled Velvia studio still. Ask NBP for a 4K 2×3 grid to receive six panes with perfect wardrobe and light continuity. Slice and label them, then adapt two more detail angles from the same setup. Animate eight total clips in Kling 2.6 with slow boom and pan moves, using first-plus-last-frame guidance where available. In easypeasyease, stack the sequence wide-to-tight and weave the two details where the geometry match-cuts cleanest. A whisper of room tone and a single low piano note under the transition to the low oblique full-body give it shape. The finished piece is a 25-second lookbook that feels like one uninterrupted studio session.

Final checklist

• Look locked: wardrobe, light direction and quality, set geometry, color grade, aspect ratio.
• Contact sheet generated at 2K–4K with consistent pane ratios; frames sliced and named in order.
• Prompts emphasize final camera positions and minimal subject motion.
• I2V clips trimmed; easing applied; subtle audio added; final aspect ratio matches the plan.

Try this workflow in Starpop

If you want to use this exact Nano Banana Pro contact sheet workflow without wiring the tools yourself, you can run it directly inside Starpop. Starpop ships with templates and prompt skeletons for AI ads, so you can plug in your brand, upload a base still, and generate a full contact-sheet-to-video sequence in minutes. Sign up for Starpop to start building AI ads campaigns using Nano Banana Pro, Kling, and this contact sheet approach.

References and further reading:

• Willie Falloon, Contact Sheet Prompting (fashion/editorial workflow, Kling and easypeasyease usage): https://www.willienotwilly.com/contact-sheet-prompting
• Nanoprompts.org cinematic 3×3 guide: https://nanoprompts.org/prompt-handbook/trending-prompts/cinematic-storyboard-contact-sheet
• Node Banana repo: https://github.com/shrimbly/node-banana
• Kling web UI: https://klingai.com
• easypeasyease: https://easypeasyease.vercel.app

Acronyms used: Nano Banana Pro (NBP), image-to-video (I2V), depth of field (DoF), Extreme Long Shot (ELS), Long Shot (LS), Medium Long Shot (MLS), Medium Shot (MS), Medium Close-Up (MCU), Close-Up (CU), Extreme Close-Up (ECU).