AI is moving at the speed of light. In just months, creative tools have gone from “kind of impressive” to generating high-quality visuals, audio, and animations that rival studio-level production.
At Guardz, we’re not just watching it happen. We’re in it, testing it, breaking it, and seeing what’s possible. Because if we’re going to stay ahead and help others do the same, we need to know which tools are worth the hype and which are just noise.
This is the story of one of those experiments.
The Challenge: Find the Best AI Tools for Visual Content Creation
It started with a simple goal: put today’s top generative tools to the test.
Liran Viler, our Lead Marketing Designer, took the mission personally. Using a small gift we gave him for his Guardziversary, he jumped into a multi-step creative challenge: take a basic photo and turn it into a fully animated, high-quality, audio-backed scene powered entirely by AI.
The subject? A LEGO character.
The vibe? Full-blown DJ rave.
The goal? Learn what works, what doesn’t, and how fast these tools are improving.
Step-by-Step Breakdown: Tools We Tested
Step 1: Start with a real photo
Liran Viler, our creative lead, took a picture of a LEGO DJ figure. Simple and analog.
No filters. No staging. Just a minifigure, a record, and a boombox.
Step 2: Use clear, focused prompting
Then came the first big test. Could AI interpret a real image and turn it into something fully staged and visually complex?
The prompt:
- Use this image
- Isolate the LEGO character
- Place him on top of a stage as the DJ
- At the bottom, show thousands of LEGO characters dancing in a rave
Tool tested: Midjourney
First results? Not bad. It got the concept, but the execution was a little off.
Step 3: Improve the output
To fine-tune the results, Liran tried to make the DJ match the source figure more closely.
This meant uploading both the photo and the earlier output, plus refining the prompt.
Bonus moment: he accidentally wrote “bird” instead of “beard” and Midjourney returned a DJ bird. Ridiculous, hilarious, and a reminder that AI will do exactly what you tell it.
Step 4: Try again with a clean start
Lesson learned. Don’t just keep poking the same conversation. Start fresh and get clear.
The new prompt:
- Fewer characters surrounding the main DJ
- All the characters are dancing
- Zoom out of the scene
- Add a bar
- Add lady LEGO dancers
- Add a smoke effect to the background
Result: more control, better spacing, improved visual storytelling.
Step 5: Refine for realism
To avoid flatness and make the scene more cinematic, Liran adjusted:
- The lighting (swapped gradients for realistic smoke and beams)
- Added purple and green lights
- Reinserted the original LEGO face from his real photo for continuity
Tools tested: Photoshop and Midjourney combined
Step 6: Animate the whole thing
With the visual locked, it was time to make it move.
Liran moved the still scene into Sora, OpenAI’s text-to-video tool, using this prompt:
“Take this image and turn it into a live party. The characters are dancing while the main character is dancing to the beat. The smoke in the background is slowly moving.”
Tool tested: Sora
Result: A fully animated scene where the LEGO rave comes alive.
Step 7: Add a custom soundtrack
A great visual needs an equally strong soundtrack. Liran used Suno to generate a custom beat based on Guardz’s brand energy. He pulled everything together in Adobe Premiere for final polish.
Tools tested: Suno and Adobe Premiere
What This Experiment Taught Us
This wasn’t just a creative detour but a real-world test of AI’s current visual capabilities.
We learned:
You have to experiment constantly to keep your edge. The AI space doesn’t wait. That’s why we’re always experimenting with marketing visuals, product design workflows, and video content. This LEGO rave wasn’t a one-off. It was one of many experiments we’re running to understand the creative AI stack inside and out.
Prompting is an art. Simpler and clearer get you better results.
Different tools have very different strengths. You need to stack them smart.
AI is evolving so fast that what you learn this week might change next week.
We’ll keep sharing what works, what breaks, and what you need to watch next.
And shoutout to the tools from this round:
- Midjourney (image generation)
- Sora (video animation)
- Suno (AI soundtrack)
- Adobe Premiere (final polish)
Stay tuned for more tests, more learnings, more unexpected bird-DJs coming your way soon.
- Share On:
