Unveiling the full production process of viral dancing kitten AI: Nano Banana Pro image consistency + RunningHub dancing workflow complete guide

Dancing cat videos are taking social media by storm. Their catchy moves and fluid motions make it hard to believe they're actually AI-generated. So, how exactly is this tech wizardry achieved? This post reveals the full production process, focusing on the character consistency of the Nano Banana Pro API and the practical application of RunningHub's dancing workflow.

Core Value: By the end of this article, you'll master the complete technical solution for creating viral AI dancing videos from scratch. You'll learn how to use Nano Banana Pro to maintain character consistency and leverage RunningHub workflows for professional-grade dancing effects.

Core Technical Highlights of AI Dancing Cat Videos

Why are dancing cat videos so viral?

AI dancing videos have swept global social platforms in 2026, with billions of views across TikTok, Instagram Reels, and YouTube Shorts. Their explosive popularity stems from a few key factors:

1. Strong Visual Impact: Seeing pets like cats perform human dance moves creates a striking visual contrast.
2. Lower Production Barrier: AI tools allow regular users to create high-quality content without professional skills.
3. High Emotional Resonance: Pet content is naturally shareable and holds high emotional value.
4. Technical "Wow" Factor: Fluid movements and realistic lighting make it difficult to distinguish from reality.

🎯 Technical Tip: The secret to high-quality dancing videos lies in character consistency and motion fluidity. We recommend using the Nano Banana Pro API via the APIYI (apiyi.com) platform. They provide official relay services that support batch generation and guaranteed character consistency at a much better price point.

Core Tech #1: Nano Banana Pro Image Consistency Generation

What is Nano Banana Pro?

Nano Banana Pro is the Gemini 3 Pro Image Preview model released by Google DeepMind, specifically designed for professional-grade asset production. Its main edge is:

Multi-character consistency: In a single generation, you can mix up to 14 input images while maintaining the appearance of up to 5 different characters.

This capability makes it the go-to tool for creating animation storyboards, marketing materials, and AI dancing videos. Compared to other image generation models, Nano Banana Pro can keep details like facial features, fur color, and body shape highly consistent across multiple frames, avoiding the common "different cat in every frame" problem.

Key Technical Parameters

Using the Nano Banana Pro API to Generate Multi-Angle Cat Images

Here's an example of how to call Nano Banana Pro through the APIYI platform to generate multi-angle cat images:

Minimalist Code Example

import requests

# API Configuration
api_key = "YOUR_APIYI_API_KEY"
base_url = "https://vip.apiyi.com/v1/gemini"

# Generate multi-angle cat images
response = requests.post(
    f"{base_url}/generate-image",
    headers={"Authorization": f"Bearer {api_key}"},
    json={
        "model": "gemini-3-pro-image-preview",  # Nano Banana Pro
        "prompt": "An orange short-haired cat, full-body shot, standing pose, white background, high-definition photography",
        "reference_images": ["https://your-storage.com/cat-reference.jpg"],
        "consistency_mode": "character",  # Character consistency mode
        "num_images": 8,  # Generate 8 different angles
        "resolution": "1024x1024"
    }
)

# Get the generated images
images = response.json()["images"]
for i, img_url in enumerate(images):
    print(f"Angle {i+1}: {img_url}")

import requests
import time
from typing import List, Dict

class NanoBananaProGenerator:
    def __init__(self, api_key: str):
        self.api_key = api_key
        self.base_url = "https://vip.apiyi.com/v1/gemini"
        self.headers = {"Authorization": f"Bearer {api_key}"}

    def generate_consistent_images(
        self,
        reference_image: str,
        prompt_template: str,
        angles: List[str],
        resolution: str = "1024x1024"
    ) -> List[str]:
        """
        Generates consistent images from multiple angles.

        Parameters:
        - reference_image: URL of the reference image
        - prompt_template: Prompt template using {angle} placeholder
        - angles: List of angles, e.g., ["Front", "Side", "Back", "3/4 View"]
        - resolution: Output resolution

        Returns:
        - List of generated image URLs
        """
        generated_images = []

        for angle in angles:
            prompt = prompt_template.format(angle=angle)

            response = requests.post(
                f"{self.base_url}/generate-image",
                headers=self.headers,
                json={
                    "model": "gemini-3-pro-image-preview",
                    "prompt": prompt,
                    "reference_images": [reference_image],
                    "consistency_mode": "character",
                    "num_images": 1,
                    "resolution": resolution,
                    "guidance_scale": 7.5,  # Controls adherence to the prompt
                    "consistency_strength": 0.85  # Character consistency strength
                }
            )

            if response.status_code == 200:
                img_url = response.json()["images"][0]
                generated_images.append(img_url)
                print(f"✅ Successfully generated {angle} angle: {img_url}")
            else:
                print(f"❌ Failed to generate {angle} angle: {response.text}")

            # Avoid hitting rate limits
            time.sleep(1)

        return generated_images

    def batch_generate_dancing_poses(
        self,
        reference_image: str,
        dance_poses: List[Dict[str, str]]
    ) -> List[str]:
        """
        Batch generates dancing pose images.

        Parameters:
        - reference_image: Reference photo of the cat
        - dance_poses: List of poses in format [{"pose": "Raise paw", "description": "Left paw raised..."}]

        Returns:
        - List of generated pose image URLs
        """
        pose_images = []

        for pose_data in dance_poses:
            prompt = f"""
A cat exactly identical to the reference image, {pose_data['description']},
white background, full-body shot, high-definition photography, clear details,
ensure the cat's fur color, patterns, and body shape are perfectly consistent with the reference.
            """.strip()

            response = requests.post(
                f"{self.base_url}/generate-image",
                headers=self.headers,
                json={
                    "model": "gemini-3-pro-image-preview",
                    "prompt": prompt,
                    "reference_images": [reference_image],
                    "consistency_mode": "character",
                    "num_images": 1,
                    "resolution": "1024x1024"
                }
            )

            if response.status_code == 200:
                img_url = response.json()["images"][0]
                pose_images.append(img_url)
                print(f"✅ Successfully generated pose '{pose_data['pose']}'")
            else:
                print(f"❌ Failed to generate pose '{pose_data['pose']}'")

            time.sleep(1)

        return pose_images

# Usage Example
generator = NanoBananaProGenerator(api_key="YOUR_APIYI_API_KEY")

# Example 1: Generate multi-angle images
angles = ["standing front", "left side", "right side", "back view", "3/4 view", "sitting front", "lying down", "jumping pose"]
prompt_template = "An orange short-haired cat, {angle} pose, white background, high-definition photography, maintain consistent appearance"

multi_angle_images = generator.generate_consistent_images(
    reference_image="https://your-storage.com/cat-reference.jpg",
    prompt_template=prompt_template,
    angles=angles
)

print(f"\nGenerated {len(multi_angle_images)} multi-angle images")

# Example 2: Generate dancing pose sequence
dance_poses = [
    {"pose": "Starting Stand", "description": "Standing on all fours, head slightly tilted up, ready to dance"},
    {"pose": "Left Paw Up", "description": "Left front paw raised to head height, right front paw naturally down"},
    {"pose": "Both Paws Up", "description": "Both front paws raised simultaneously, body leaning slightly back"},
    {"pose": "Spin Action", "description": "Body turned 45 degrees to the side, tail swishing"},
    {"pose": "Jump Pose", "description": "All four paws off the ground, body in mid-air, dynamic motion"},
    {"pose": "Landing Pose", "description": "Front paws touching down first, back legs bent for support"},
    {"pose": "Sway Motion", "description": "Body swaying left and right, tail moving in sync"},
    {"pose": "Ending Pose", "description": "Sitting down, front paws tucked in, satisfied expression"}
]

dancing_pose_images = generator.batch_generate_dancing_poses(
    reference_image="https://your-storage.com/cat-reference.jpg",
    dance_poses=dance_poses
)

print(f"\nGenerated {len(dancing_pose_images)} dancing pose images")
print("These images can be directly imported into a RunningHub workflow to generate a dancing video.")

💡 Character Consistency Tip: Explicitly using phrases like "exactly the same as the reference image" or "maintain identical appearance features" in your prompt can significantly boost Nano Banana Pro's consistency. We recommend using the APIYI (apiyi.com) platform, which supports the consistency_strength parameter to fine-tune the consistency level.

Core Tech II: RunningHub Dancing Workflow in Action

What exactly is RunningHub?

RunningHub is a cloud-based ComfyUI platform that offers thousands of ready-to-use workflows for text-to-video, image-to-video, and video-to-video creation. Its main perks include:

• Error-free execution: Every workflow is pre-tested, so you won't run into those annoying red error boxes or missing nodes.
• High-speed online generation: No need for local deployment; just use their cloud computing power to get results fast.
• Professional dancing workflows: It comes with several built-in workflows specifically designed for generating dance videos.

RunningHub Dancing Workflow Types

Generating a Dancing Kitten with RunningHub

The full step-by-step process:

Step 1: Get your assets ready

1. Full-body cat photo: Use the multi-angle consistent images generated by Nano Banana Pro.

   • Recommended resolution: 1024×1024 or higher.
   • Requirements: White or solid color background, with the full cat visible.
   • Pose: Ideally standing or sitting with all limbs clearly shown.

2. Reference dance video: Prepare a clip of a human dancing.

   • Duration: 3-10 seconds (roughly 60-250 frames).
   • Movement: Moderate movement is best; avoid extreme tumbling or rolling.
   • Background: A simple background makes it easier for the AI to track the motion.

Step 2: Upload to RunningHub

1. Head over to the RunningHub platform: runninghub.ai
2. Select the "AI Animals Dancing" workflow.
3. Upload your prepared cat image.
4. Upload your reference dance video.
5. (Optional) Upload a custom music file.

Step 3: Configure generation parameters

- **Frame Settings**: Auto-match based on the dance video (≤250 frames)
- **Motion Intensity**: Set to Medium
- **Smoothness**: Set to High to ensure fluid motion
- **Background Handling**: Choose "Keep original background" or "Transparent background"
- **Resolution**: 1080p (Full HD)

Step 4: Generate and export

1. Click the "Run" button to start the process.
2. Wait about 30-90 seconds (depending on the frame count).
3. Preview the generated result.
4. Download the video in MP4 format.

Pro Tip: Splicing Multiple Dance Clips

If you need a long video that exceeds 250 frames, you can use a segmented generation strategy:

# Pseudo-code example: Logic for multi-segment dance generation
segments = [
    {"cat_image": "pose_1.jpg", "dance_video": "dance_part_1.mp4"},
    {"cat_image": "pose_2.jpg", "dance_video": "dance_part_2.mp4"},
    {"cat_image": "pose_3.jpg", "dance_video": "dance_part_3.mp4"}
]

generated_videos = []
for segment in segments:
    video = runninghub_generate(
        image=segment["cat_image"],
        reference=segment["dance_video"]
    )
    generated_videos.append(video)

# Use a video editing tool to stitch them together
final_video = merge_videos(generated_videos)

🚀 Efficiency Suggestion: For batch generation, we recommend first using the APIYI (apiyi.com) platform with Nano Banana Pro to generate 10-20 images of the cat in different poses. Then, import them into RunningHub in batches to create multiple video clips and stitch them into a final masterpiece. This approach allows for much richer movements and longer video durations.

Step-by-Step Production Case Study

Case Study: Creating a Video of an Orange Cat Breakdancing

Project Goal: Create a 15-second orange cat breakdancing video with smooth motion and character consistency, optimized for TikTok.

Technical Stack:

• Image Generation: Nano Banana Pro (via APIYI apiyi.com)
• Video Generation: RunningHub "AI Animals Dancing" workflow
• Post-processing: Adding music and captions

Detailed Steps:

Step 1: Generate Consistent Cat Images (Nano Banana Pro)

# Using the previously mentioned NanoBananaProGenerator class
generator = NanoBananaProGenerator(api_key="YOUR_APIYI_API_KEY")

# Define the breakdance pose sequence
street_dance_poses = [
    {"pose": "Start", "description": "Standing pose, front paws hanging naturally, ready to dance"},
    {"pose": "Left paw up", "description": "Left front paw pointing straight up, right paw on hip, head slightly tilted up"},
    {"pose": "Squat prep", "description": "Squatting down, both front paws on the ground, gathering power"},
    {"pose": "Jump", "description": "Jumping off the ground with all four limbs, body stretched out"},
    {"pose": "Turn", "description": "90-degree turn in mid-air, tail swinging"},
    {"pose": "Single paw support", "description": "Left front paw on the ground, right paw raised, body tilted"},
    {"pose": "Crossed paws", "description": "Both front paws crossed in front of the chest, cool pose"},
    {"pose": "Ending pose", "description": "Sitting down, one paw raised, with a proud expression"}
]

# Generate 8 pose images
cat_poses = generator.batch_generate_dancing_poses(
    reference_image="https://your-storage.com/orange-cat-ref.jpg",
    dance_poses=street_dance_poses
)

Step 2: Prepare the Reference Dance Video

1. Search for "breakdance tutorial short videos" on YouTube or TikTok.
2. Pick a clip with clear movements and a simple background.
3. Use a video editing tool to crop it down to 8-10 seconds.
4. Make sure the video includes a complete dance loop.

Step 3: Generate the Dancing Video on RunningHub

1. Visit runninghub.ai/ai-detail/1882704909102469121/hid (AI Animals Dancing workflow).
2. Upload the first cat pose image (cat_poses[0]).
3. Upload your reference breakdance video.
4. Configure parameters:
   • Frames: Auto-detect (approx. 200-240 frames, 8-10 seconds)
   • Motion Intensity: High
   • Smoothness: High
5. Click Run and wait 60-90 seconds.

Step 4: Post-Processing Optimization

1. **Add Music**:
   - Pick some high-energy breakdance music.
   - Use DaVinci Resolve or Premiere Pro to sync the music and video.
   - Make sure the moves align with the beat.

2. **Add Captions and Effects**:
   - Add a "Master Orange Cat Breakdancer" caption at the start.
   - Add "Explosion" or "Flash" effects at key moves.
   - Add a "Like and follow for more" prompt at the end.

3. **Color Grading**:
   - Boost saturation and contrast to make it pop.
   - Add a slight sharpening effect.
   - Ensure the cat's fur color is vibrant and eye-catching.

4. **Export Settings**:
   - Resolution: 1080x1920 (Vertical 9:16)
   - Frame Rate: 30fps
   - Codec: H.264, High Bitrate
   - Optimized for TikTok / Instagram Reels specs.

Cost and Time Estimate

💰 Cost Optimization: By batch-calling the Nano Banana Pro API through the APIYI (apiyi.com) platform, you can access enterprise discounts. For content creators and studios, the platform offers monthly packages that can drop the cost per image to $0.05-$0.10, significantly cutting production costs.

FAQ & Solutions

Technical Summary: Creating AI Dancing Cats

Core Technology Review:

1. Image consistency is key: By leveraging Nano Banana Pro's multi-character consistency capabilities, you can ensure the same cat maintains a uniform look across different frames.
2. Choosing the right workflow: RunningHub's AI Animals Dancing workflow is specifically optimized for animals, yielding much better results than generic workflows.
3. Asset quality determines the outcome: Success starts with high-quality cat images (clean backgrounds, distinct features) and appropriate dance reference videos.
4. Post-optimization is essential: Adding music, subtitles, and special effects can boost a video's viral potential several times over.
5. Controllable costs: Producing a single video costs roughly $0.80-$2.00, making it ideal for individual creators and studios looking to scale up.

Practical Advice: For creators who want to batch-produce AI dancing videos, we recommend the "Nano Banana Pro (APIYI platform) + RunningHub + Post-editing" combo. By using the Nano Banana Pro API through the APIYI (apiyi.com) platform, you can take advantage of bulk discounts and enhanced consistency features, significantly boosting your production efficiency and quality. Combined with RunningHub's high-speed workflows, one person can churn out 10-20 high-quality dancing videos a day, making large-scale content production a reality.

References:

1. Google AI for Developers – Nano Banana Image Generation Docs

   • Link: ai.google.dev/gemini-api/docs/image-generation
   • Description: Official API documentation and user guides for Nano Banana Pro (Gemini 3 Pro Image).

2. Google DeepMind – Nano Banana Pro Product Page

   • Link: deepmind.google/models/gemini-image/pro
   • Description: Nano Banana Pro technical specs, capabilities, and use cases.

3. RunningHub – ComfyUI Cloud Platform

   • Link: runninghub.ai
   • Description: A cloud-based ComfyUI platform offering ready-to-use workflows for dance video generation.

4. RunningHub – AI Animals Dancing Workflow

   • Link: runninghub.ai/ai-detail/1882704909102469121/hid
   • Description: A specialized workflow for animal dancing videos that supports one-click uploads.

5. FlexClip – AI Cat Dancing Technical Analysis

   • Link: flexclip.com/learn/ai-cat-dancing.html
   • Description: A guide on production techniques and best practices for AI dancing cat videos.

6. GoEnhance – AI Cat Dancing Generator

   • Link: goenhance.ai/ai-dance/cat-dancing
   • Description: AI cat dancing video generation tool and its underlying technical principles.

Author: APIYI Technical Team
Technical Support: If you're looking for Nano Banana Pro API batch solutions or technical consulting on AI video generation, head over to APIYI (apiyi.com) for professional support and customized services. We offer free test credits to help you quickly verify your technical setup.