Summer Road Trip in Four Shots

Tested running a full 4-shot summer road trip sequence in one single prompt with MiniMax H3 + ComfyUI. 🎬

Instead of generating disconnected clips and stitching them together on a timeline, I wanted to see if the model could handle multi-scene continuity on its own. 

⚙️ Setup & Specs:
• Workflow: Text-to-Video (MiniMax H3 + ComfyUI)
• Prompting: Structured into 4 scenes using the official H3 prompt skill
• Local Render: RTX 4090 24GB @ 768P (~1,000s)
• Upscale: Pushed to crisp 2K via MiniMax Regenerate API (~6 mins)

Prompt & GitHub below ⬇️

### 作者补充

[补充1]
Tested running a full 4-shot summer road trip sequence in one single prompt with MiniMax H3 + ComfyUI. 🎬

Instead of generating disconnected clips and stitching them together on a timeline, I wanted to see if the model could handle multi-scene continuity on its own. 

⚙️ Setup & Specs:
• Workflow: Text-to-Video (MiniMax H3 + ComfyUI)
• Prompting: Structured into 4 scenes using the official H3 prompt skill
• Local Render: RTX 4090 24GB @ 768P (~1,000s)
• Upscale: Pushed to crisp 2K via MiniMax Regenerate API (~6 mins)

Prompt & GitHub below ⬇️

[补充2]
integrated_multimodal_description: [Shot 1] Live-action, exactly 15.00 seconds, vertical 9:16, a bright midsummer day on a black-volcanic-rock shoreline, filmed as a first-person smartphone POV. There are exactly four shots and four captured summer views. Each shot begins with 0.6 to 0.8 seconds of an unobstructed establishing view with mild natural handheld sway. The same physical handmade camera prop then rises smoothly into the foreground from the lower-right edge, settles, and remains steady for the photograph. The prop is a flat horizontal compact-camera silhouette cut by hand from corrugated cardboard, exactly 22 cm wide and 13 cm high. Visible tan paper fibers and slightly crooked hand-cut edges remain exposed beneath a cheerful locked color design: a wide turquoise band surrounds the landscape viewing hole, the top strip is coral pink, the right control panel is sunflower yellow, the grip is lime green, the four-way control is lavender with a white center, and two small buttons are sky blue and orange. Doubled matte-black marker outlines define every edge. Small hand-drawn decorations—a white five-point star, an orange sun, and three dark-blue wave marks—appear in fixed positions on the body without any letters or numbers. The large landscape rectangular viewing hole occupies the left two-thirds; a small raised faux hot-shoe shape sits along the top edge. It has no real lens, glass, screen, electronics, readable text, or brand. The same adult right hand holds its upper-right and lower-right edges in all four shots, with the thumb resting on the white center of the lavender control pad. The prop's colors, decorations, dimensions, edge shape, button positions, and hand grip remain identical in every shot and never redesign themselves. After the cardboard settles, the phone focus locks on the cardboard and hand about 25 cm away: cardboard fibers, painted-pencil texture, black marker strokes, fingertips, and colorful controls are sharp, while the live-action landscape behind and through the empty viewing hole becomes softly defocused but remains clearly recognizable by shape, color, and layout. At every shutter, only the thumb pad depresses the control by about 3 mm. One short mechanical camera click-clack occurs, immediately followed by a very brief two-frame full-screen white exposure flash. When the live-action view returns, the entire viewing hole has been replaced by one fully opaque, motionless hand-drawn paper illustration of the composition that was inside the hole; zero photographic pixels remain inside the hole, while the live-action scenery outside it continues normally. There is no morph, dissolve, animated drawing process, zoom, recoil, prop redesign, or intermediate half-photo state. All four illustrations share one locked handmade recipe closely matching a casual travel sketch: warm cream paper, slightly uneven black fineliner outlines, simple colored-pencil fills in sky blue, deep blue, turquoise, coral orange, sage green, warm ivory, lavender, lime green, pink, sunflower yellow, tan, brown, white, and black, light diagonal pencil strokes with visible paper gaps, simplified shapes, mildly awkward perspective, and moderate detail. They are charming but visibly handmade, never photorealistic, glossy, vector-clean, 3D, anime, painterly, or heavily shaded.

The opening view is a small sunlit seaside cove. Exactly two young adult East Asian women stand from head to upper thigh on a rough gray concrete edge in the lower half of the frame, leaving the blue sea, basalt, and sky clearly visible around them. Behind them, clear deep-blue water moves through irregular charcoal-black basalt rocks; short white foam gathers in the narrow channels. A low green grassy strip and one small white canopy are far left, beneath a saturated cobalt-blue sky with a few small soft cumulus clouds. The woman on the left has glossy black hair gathered into a neat low bun with two soft face-framing strands. She wears an elegant coral-ora