Creativity + AI = Limitless Visual Possibilities

Visual content creation has become much easier and faster with AI tools. We now can produce high quality images, designs, and videos without traditional design and production expertise. In our latest YouTube video Lilac Lane's Instagram Ad, we used multiple AI tools to produce this visually engaging story from start to finish.

Here’s the step-by-step process that we followed:

AI Tools for video generation: Tools tested

We tested the multiple AI tools to find the right mix of ai tools for video generation.

AI Image Generators

To generate characters and scene images.

1. Midjourney

2. Dall-E

3. Leonardo AI

4. Adobe Firefly

AI Video Generators

To convert images into animated video scenes.

1. Kling

2. Hailuo

3. Pika

4. Runway

5. Hedra

AI Voice Generators

To generate voiceovers for each character.

1. ElevenLabs

2. MotionArray

3. Murf AI

AI Design Generators

To generate banners, ads, and thumbnails.

1. Sivi

2. Microsoft Designer

3. Canva

AI Text Generators

To refine the script and improve image and video prompts.

1. Gemini

2. ChatGPT 4o

3. Perplexity

Once we finalized the ai tools, we used them with paid subscriptions for advanced features and high-quality output.

The Step-by-step process of how we created our YouTube video With AI tools

We created a structured workflow and followed it to ensure easier and faster production. Here's the behind the scenes of how we created:

The Vision: Conceiving the idea and story

First things first, we started with brainstorming and coming up with story ideas. The story needs to be simple, engaging, and fit within a short-form video, approximately 30 seconds. The video has to be relatable to our global audience of small businesses, agencies, ecommerce stores, and freelancers. We took inspiration from the designs generated with Sivi by our users. We wanted to create stories that highlight how these businesses across verticals are easily generating ad creatives, social posts, email designs, and more using Sivi, the multilingual AI design tool. We created fictional story ideas around these use cases.

Once the idea was finalized, we gave it a name and wrote the script and characters.

Describing the characters & generating character images (Text to Image)

Once the story was finalized, we visualized the characters. Using Midjourney, we generated multiple versions of the characters and selected the best ones. Since Midjourney allows for detailed text-to-image creation, we provided detailed descriptions of each character’s look, attire, expressions, and settings. Midjourey helped us create consistent characters across scenes.

These AI-generated images were used as the foundation for our animated video. Once satisfied with the character images, we proceeded to scene creation.

Writing & refining the script (Text to Text)

Once the characters are finalized, we wrote an initial draft of the script. We used ChatGPT 4o to refine and enhance the dialogues, ensuring they sounded natural, engaging, and concise. We made sure that the final script fit within 30 seconds.

Creating the storyboard

Before moving into full production, we created rough pencil sketches for each scene.

1. This helped visualize the scene transitions, character interactions, and overall flow of the video.

2. It also helped ensure that the AI-generated images and animations would align with the story.

Once the rough sketches were complete, we began generating AI-powered scene images.

Generating scene images using AI image generator (Image to Image)

With character images in place, we generated scene-specific images that visually matched the story.

Using Midjourney’s image-to-image generation, we:

1. Used character images as a base.

2. Added prompts to adjust their expressions, backgrounds, and poses.

3. Created different scene variations to pick the best ones.

This helped us transform the static images into different settings while maintaining the original character designs.

Building the storyboard with generated images

We then organized the generated images into a visual storyboard based on the script. This helped us map put the story visually before video generation and served as a blue print for the final video..

Turning images into video using AI video generator (Image to Video)

With the storyboard complete, we moved to animating the images. We used Hailuo for image-to-video generation.

1. Uploaded scene images into Hailuo.

2. Provided detailed motion prompts for each scene to define camera angles, movements, and animation effects.

3. Adjusted timing and movement to make character expressions feel natural.

This resulted in smooth AI-generated video clips for each scene.

Generating voiceovers for each character using AI voice generator (Text to Audio)

Now that the visuals were ready, we needed to add voiceovers.

Using ElevenLabs, we:

1. Selected realistic AI-generated voices that matched the characters.

2. Adjusted the tone, pitch, and emotion to create a natural delivery.

3. Converted each dialogue into natural-sounding speech

Creating custom designs for the video using AI design generator (Text to Design)

To generate the banners for the small business, we used Sivi:

1. Added the brand colors, fonts, logos, and assets to Sivi

2. Added the prompt to generate the designs

With the brand details and prompt, Sivi generated the ads, banners, posters, etc. needed for the small business.

With Sivi’s text to design, we also generated thumbnails for this video.

Editing & post-production

With all assets ready, we moved to post-production, where we:

1. Combined video clips, voiceovers, and designs.

2. Synced audio with video timing for natural flow.

3. Added subtle effects, transitions, and background music for engagement.

4. Reviewed the final cut to ensure all elements aligned with the creative vision.

The result? A seamless, AI-generated YouTube video!

Sharing the story: YouTube publishing

We optimized the video for YouTube Shorts with an engaging title, description, thumbnail created with Sivi, and relevant hashtags.

Final thoughts

Combining multiple AI tools - Midjourney for images, Hailuo for scenes, ElevenLabs for audio, Sivi for design, and ChatGPT 4o for script, enabled us to create a compelling, AI generated YouTube video effortlessly and streamlined production. This is about the future of AI in storytelling, where creativity meets automation.

The learning curve with each of these tools were relatively low. Good creativity, visualization, and expressing it in text prompts is all it takes. The result? Faster production, high-quality output, cost-effective, and creative control. Anybody can design with AI! As generative AI continues to evolve, we believe that AI-assisted visual content creation will become more accessible to produce stunning content.

We’re excited to create more AI-driven content and push the boundaries of creativity! Stay tuned for more innovations from the Sivi Design Team. Follow us for more AI-powered creativity!

Want to know the comparison of these AI tools, why we chose this combination of AI design tools, the don’ts of using these AI tools, or how long it took us to produce one video? Just subscribe to Sivi blog.