curiousrefuge: Stop Motion Felt Mountain Adventure AI Art

Consistency is the “holy grail” of AI filmmaking. We tested Nano Banana 2’s ability to generate images using character references for both single and multiple characters, and the results were pretty impressive. Not every character stays perfectly consistent all the time, especially in multi-character scenes, and realism doesn’t hold up in every scenario. But the ability to upload up to 14 reference images opens up real possibilities for narrative storytelling. #ai #generativeai #aifilmmaking #aivideo #aiadvertising

Curious Refuge

@curiousrefuge

INSTAGRAM · 2026-03-13Source

286likes

9comments

Remix This

Recreate with AI Image Generator

Make your own AI viral video

Prompt

[Subject] A polished AI demo graphic showing a multi-person image-to-image workflow. In the top card, display five vertically cropped reference portraits of different people: a gray-haired man, a red-haired child or teenager, a smiling Black woman in a rainbow-striped outfit, a bearded man in a black turtleneck and orange knit cap, and an older man in a patterned traditional-style hat. In the middle card, show the generated output: all five transformed into handcrafted stop-motion felt characters standing together on a mountain-climbing adventure scene. At the bottom, include a prompt card describing the transformation into a stop-motion wide shot of felt characters.

[Environment] Premium teal-and-dark-green UI presentation with rounded rectangular panels outlined by a subtle mint or teal neon stroke. The top panel contains the reference portraits labeled “REFERENCE IMAGES,” the middle panel contains the stop-motion mountain result labeled “NANO BANANA 2,” and the bottom panel contains a dark green prompt box with white text. The entire layout should feel like a refined social-media product demo for an AI image tool.

[Composition/Camera] Vertical three-panel infographic composition. Top panel: five portrait slices aligned side by side. Middle panel: wide stop-motion mountain scene with the felt versions of all five people walking or standing on a mountain trail, surrounded by miniature pine trees, rocky slopes, small clouds, and a snow-capped peak. Bottom panel: prompt text centered inside a rounded UI box. Keep even spacing and consistent card styling throughout.

[Lighting] Soft interface glow around the card edges, with the reference portraits lit naturally and the generated felt scene lit like a bright handcrafted miniature set. The stop-motion output should have gentle daylight, soft shadows, tactile felt texture, and clean studio clarity.

[Style/Rendering] High-end AI workflow showcase, product-marketing graphic, stop-motion craft aesthetic in the output section, clean UI mockup styling, editorial tech-demo composition, and premium social content design.

[Detail constraints] Preserve clear identity correspondence between the reference portraits and the felt characters, keep the mountain scene handcrafted with visible wool or felt textures, maintain the teal card outlines, retain readable labels and prompt panel hierarchy, and avoid cluttered dashboard elements or unnecessary interface chrome.

Negative prompt: messy UI, unreadable text, low-resolution portraits, generic cartoon mountain, plastic toy textures, broken felt anatomy, extra buttons, noisy gradients, crowded interface widgets, low-detail trees, warped identities, muddy colors, photoreal mountain instead of handcrafted set.

Suggested parameters: stylize 105, quality 1, aspect ratio 4:5, crisp UI cards, soft felt miniature textures, readable typography, premium product-demo finish, subtle teal glow.

Delta prompt strategy: If the layout feels disorganized, add "three clean stacked rounded cards with equal spacing and teal outlines." If identity matching weakens, add "same five people preserved as corresponding felt characters in the mountain scene." If the felt output looks generic, add "handcrafted stop-motion felt figures, wool textures, miniature clouds and pine trees." If the UI feels too plain, add "high-end AI tool showcase with subtle neon-teal edge accents." If the prompt box disappears, add "bottom dark green prompt card with clear white instructional text." If the top panel dominates too much, add "reference portraits smaller and flatter than the central generated result." If the mountain scene lacks charm, add "storybook miniature mountain adventure with soft clouds and toy-like trees." If the whole image stops reading like a demo, add "before-and-after AI workflow graphic, not a standalone stop-motion poster." If text becomes cluttered, add "minimal labels only: REFERENCE IMAGES, NANO BANANA 2, and PROMPT." If the felt texture gets lost, add "visible wool fibers, sewn edges, and handcrafted stop-motion material detail."

This image works because it sells transformation through structure, not just through spectacle. The top panel gives the raw source material, the middle panel delivers the imaginative payoff, and the bottom panel explains the instruction that connects the two. That sequence makes the AI workflow legible in seconds. The viewer can understand both the technical claim and the creative leap without reading a long explanation.

The transformation itself is also well chosen. The reference portraits are highly varied in age, styling, and facial features, which makes the output challenge more convincing. Turning those distinct people into stop-motion felt adventurers on a mountain is visually surprising, but still simple enough to read. The output is not just “stylized”; it has a very specific handcrafted medium, and that specificity is what makes the demo memorable.

How curiousrefuge Created This Stop Motion Felt Mountain Adventure AI Art — and How to Recreate It

Signal	Evidence (from this image)	Mechanism	Replication Action
Workflow clarity	The graphic is divided into reference images, generated result, and prompt text	Sequential layout reduces cognitive friction and makes the transformation instantly understandable	Present AI demos in explicit stages rather than relying on caption text alone
Identity continuity	Each person in the portrait strip appears as a recognizable felt counterpart in the output	One-to-one mapping increases trust in the tool’s consistency	Choose reference material where the identities can be clearly tracked into the result
Medium specificity	The output is not generic animation but a handcrafted stop-motion felt world	Specific material language makes the model capability feel more impressive and intentional	Use transformation prompts that target a concrete medium, not a vague style label
Product polish	Rounded cards, teal outlines, and clean labels create a unified showcase aesthetic	Refined presentation makes the workflow feel premium and credible	Use consistent panel styling and restrained UI accents to frame the transformation

Observed Style Choices

Style Choice	Observed Effect
Top portrait strip	Creates fast reference diversity and sets up the transformation challenge
Central felt mountain scene	Acts as the emotional and visual payoff of the graphic
Teal rounded borders	Unify the three sections and signal modern product design
Handcrafted miniature textures	Give the generated result charm, warmth, and medium specificity
Bottom prompt box	Explains the exact instruction that generated the output

Prompt Technique Breakdown

Technique	Why It Matters Here	How To Phrase It
Reference-to-output mapping	The core value of the demo is showing that multiple identities can be transformed together	"all of these people reimagined as felt stop-motion characters"
Medium locking	The output becomes stronger when the handcrafted medium is explicit	"stop-motion wide shot, felt characters, miniature mountain set, wool textures"
Workflow framing	The image is presenting a tool capability, not just artwork	"AI product demo with reference panel, generated panel, and prompt card"
Scale and scene simplification	The mountain adventure remains readable because the scene is miniature and contained	"storybook mountain trail with small trees, clouds, and a snow-capped felt peak"
UI restraint	Too much interface decoration would weaken the clarity of the showcase	"minimal rounded panels, teal edge accents, clean typography, no extra dashboard clutter"

Execution Notes

To recreate this kind of demo, design the information hierarchy first. Decide which part proves identity, which part delivers the imaginative jump, and which part explains the prompt. Once that structure is stable, the actual generated image has room to shine. Here the middle panel is clearly the hero, while the top and bottom panels function as evidence and explanation.

Most weak versions will either use a transformation that is too vague or an interface that is too noisy. Fix that by choosing a very specific medium like felt stop-motion and simplifying the presentation to three clean panels. If the result feels random, strengthen the one-to-one identity mapping between the portraits and the output characters. The best version feels like a product demo that is both technically trustworthy and creatively delightful.

How curiousrefuge Created This Stop Motion Felt Mountain Adventure AI Art — and How to Recreate It

Observed Style Choices

Prompt Technique Breakdown

Execution Notes

Related AI Generator