MiniMax H3 Max: Creating 24-Yen SNS Motion Graphics with Multi-Shot Prompt Techniques

Japanese AI video creator @ponzponz15 (ぽんず | AI映像) recently shared a practical prompt structure and multi-shot directing framework on X, demonstrating how to generate a polished picture-book motion graphics commercial using the MiniMax H3 Max foundation model (480p) for an ultra-low production cost of just 24 yen (approximately 0.16 USD).

Image source: @ponzponz15 on X

Many creators attempting AI video generation default to hyper-realistic visual styles or compute-heavy high-resolution rendering, frequently encountering steep credit consumption and anatomical inconsistencies across sequential cuts. By contrast, the author embraced 480p resolution and paired it with a flat 2D picture-book aesthetic, an explicit semantic color token system, and sub-second multi-shot timeline choreography, producing an eye-catching commercial for mobile social feeds at remarkable economic efficiency.

24-Yen Cost Efficiency and Flat 2D Motion Graphics Design

The defining strength of this workflow is prioritizing intentional art direction over sheer pixel count.

Vertical (4:5) video commercials distributed across mobile feeds must deliver narrative clarity within seconds on compact smartphone displays. Instead of running 1080p or 4K passes that rapidly exhaust GPU credits, the author opted for the 480p mode of MiniMax H3 Max, bringing the single-generation cost down to roughly 24 yen. Lower resolution often risks blurry textures or muddy rendering, but the author smartly turned this constraint into a stylistic virtue by mandating a 2D flat illustration aesthetic enhanced by subtle static washi paper grain and bold hand-drawn ink outlines.

The opening directives in the prompt establish unambiguous visual boundaries and narrative foundations:

• Format and Genre: A 12-second vertical 4:5 Japanese SNS housing commercial styled as a miniature Three Little Pigs picture book.
• Narrative Roles: The pigs are established as clear protagonists, while the wolf functions as their primary narrative obstacle.
• Negative Constraints: Explicitly banishing 3D rendering, gradients, and floating particle effects (no 3D, gradients or particles) while maintaining generous negative space and cute, rounded character silhouettes.
• Animation Grammar: Avoiding erratic camera swings in favor of static framing, crisp cut transitions, solid color-field wipes, and 2D pose-to-pose animation to preserve clean motion graphics pacing.

Semantic Color Token Locks and Cross-Shot Character Consistency

Preserving consistent character wardrobe and environmental color palettes across cuts remains one of the most stubborn hurdles in generative video. The author tackled this challenge by locking the global palette to five capitalized semantic color tokens (Palette only).

The five dedicated color tokens defined in the prompt operate as strict rules:

• PAPER: A warm cream tone serving as the base paper background, the pigs' skin foundation, and the exterior walls of the third house.
• INK: Sumi-black ink applied to outer character contours, object line art, the wolf's silhouette, and bold Japanese typography.
• COOL: Deep indigo blue allocated consistently to the third pig's overalls, the wolf's scarf, the wind gust trajectories, and the pitched roof of the sturdy home.
• WARM: Golden-yellow reserved specifically for the first pig's overalls and the straw hut.
• HOT: Vibrant vermilion red designating the second pig's overalls, the brick cottage, and dynamic exclamation typography.

By cementing these tokens in uppercase (PAPER, INK, COOL, WARM, HOT) and tying each directly to individual character outfits and building materials, the model is prevented from hallucinating random hue shifts across camera cuts, locking visual continuity end to end.

Millisecond Multi-Shot Timeline and Kinetic Typography

The core achievement of this MiniMax H3 prompt lies in orchestrating a continuous 12-second multi-shot sequence with millisecond-level precision inside a single integrated_multimodal_description generation call. Rather than generating detached fragments and assembling them in an external NLE, the prompt dictates exact timestamps, scene cuts, physical interactions, and typography transitions.

The sequence breakdown across the three shots unfolds as follows:

• [Shot 1] Introduction and Construction (00:00.000 – 00:01.800): All three pigs are positioned across the lower screen beside their miniature homes: the straw hut (WARM) on the left, the brick cottage (HOT) in the center, and the finished house with an indigo roof (COOL) on the right. In three staggered beats, the first two pigs fit their roof sections, while the third merely removes a paper cover from an already completed home. All three raise their arms in pride. At exactly 00:00.600, kinetic Japanese typography reading "家、できた！" (House, finished!) snaps into place overhead via a paper-card fold assembly.
• [Shot 2] The Wolf's Attack on the Straw Hut (00:01.800 – 00:03.200): At 00:01.800, a clean cut highlights the first pig standing outside its straw doorway. A compact wolf enters from the far left, draws a quick breath, and expels three indigo wind lines. The straw panels fly offscreen, the pig's ears perk up in shock, and a bright vermilion exclamation "うわーっ！" pops above its head. The pig sprints right toward its sibling's house with ears trailing, completely free of violent debris or dust clouds.
• [Shot 3] Crisis at the Brick Cottage (00:03.200 – End): At 00:03.200, the camera cuts to the brick cottage with both pigs huddled together outside the entrance. The wolf blows a second gust; the brick structure shakes and collapses cleanly, prompting the pair to make another frantic escape toward the final sturdy home.

The complete prompt text as shared by the author is structured as follows:

integrated_multimodal_description: [Shot 1] Create a 12.000-second vertical 4:5 Japanese SNS housing commercial told as a miniature Three Little Pigs picture book. The pigs are protagonists; the wolf is their obstacle. Flat illustration, bold hand-drawn outlines, subtle static washi grain, cute round silhouettes, generous negative space. Palette only: warm cream PAPER, sumi-black INK, indigo COOL, golden-yellow WARM, vermilion HOT. Japanese kinetic type uses paper-card assembly and compact settles. Static shots, crisp cuts, color-field wipes, pose-to-pose animation; no 3D, gradients or particles. Three PAPER pigs keep distinct WARM, HOT and COOL overalls. One small INK wolf with pointed ears and COOL scarf blows rightward from the left edge. House order: STRAW, BRICK, sturdy final home. On PAPER, arrange the three pigs across the lower half beside their separate miniature houses: WARM straw hut left, HOT brick cottage center, PAPER house with COOL pitched roof and thick INK frame right. In three staggered beats, the WARM and HOT pigs place their own roof pieces; the COOL pig merely pulls off a paper cover from its already-finished house. All three thrust arms up proudly. Crisp pop-in kinetic type above: Japanese text reads "家、できた！" at 00:00.600 in three paper-card snaps. [Shot 2] At 00:01.800, cut to the WARM pig proudly standing just outside its straw doorway at right. The smaller wolf appears at far left, takes a brief breath and blows three COOL wind lines rightward. On contact, three large straw panels fly offscreen right; the pig's ears flick up and both arms rise in alarm. Exact HOT "うわーっ！" pops above the pig, not the wolf. The pig dashes right to its sibling's house, ears trailing. No injury or debris cloud. [Shot 3] At 00:03.200, cut to the HOT brick cottage at right, now with the WARM and HOT pigs together in front of its doorway. The wolf blows again; the brick cottage shakes and collapses cleanly, prompting another escape.

This prompt engineering technique demonstrates that practical adoption of generative AI in commercial advertising does not require chasing photorealism at all costs. Instead, defining rigorous art direction, locking semantic color palettes, and scripting millisecond cut transitions can deliver professional narrative coherence and exceptional cost efficiency.

Original source

• @ponzponz15 (ぽんず | AI映像) on X: MiniMax H3 Max 24-Yen SNS Motion Graphics 'Three Little Pigs' Prompt & Video — Primary announcement thread detailing 480p generation settings, the 5 semantic color tokens (PAPER, INK, COOL, WARM, HOT), and the millisecond-accurate 3-shot prompt breakdown for mobile social advertising.