Today around noon, I spontaneously felt like creating a spec ad. No long preparation. Just the decision to take an idea and execute it straight away, and see how far I could get in a short amount of time with a clean AI based workflow. Pretty early on, I knew I wanted to do something with Amazon. The next step was finding the right scenario. I wanted an environment that creates tension immediately. Since I already had Artemis on my mind, a space setting felt like the obvious choice. That led to the core idea: if something breaks while being out in space, what would a solution look like if it had to arrive fast and get straight to the point? I started by building the astronaut in Nano Banana 2. That gave me a clearly defined main character right from the beginning, and I knew exactly how he should look across all shots. I also generated the rest of the images with Nano Banana 2 and built the visual world step by step. In the next stage, I turned the shots into moving footage with Kling 3.0. Deliberately without sound and without music, because I wanted to control image, voice, and sound design separately. I then created the music with Suno. The dialogue between the astronaut and Houston was generated with ElevenLabs. For that, I picked two voices that worked well together for me. After that, I edited the voice tracks in CapCut and added a distortion effect so the whole thing sounded more like real radio communication. I also added a few radio and interference sounds to make the atmosphere feel more believable. In the end, everything came together in Premiere. That is where I did the final edit and assembled the visuals, music, dialogue, and sound design. From the initial impulse to the finished spot, the whole process took me around three hours. Once the idea is in place and the tools work well together, you can build a complete piece in a short amount of time that feels coherent from start to finish. I am happy with the result.

I started with two to three reference photos and generated the base stills in Nano Banana 2. After that, I converted everything to black and white because the overall look felt stronger that way. Then I brought the images into Kling 3.0 and Grok. The prompts were kept very simple. I mainly wanted multiple angles and close ups, then let the models run with it. In total, I got five short clips ranging from about five to fifteen seconds. I edited everything together in Premiere. One issue was that some of the Grok clips pushed the black values too hard, so parts of the footage shifted away from a normal photorealistic look and started leaning more into a high contrast Sin City type aesthetic. That was hard to fully correct in post, so I added grain across the full piece to help unify the material and make the overall look feel more consistent. After that, I generated the voice over in ElevenLabs and finished the full edit in Premiere. For Instagram and TikTok, I added a Hellboy soundtrack directly from each platform’s music library. Since I already knew the track beforehand, I was already cutting with that rhythm in mind. Total work time was around twenty minutes. Pretty fast process, very simple prompts, and still a fun result. What surprised me most is how absurdly well this Hellboy makeup actually suits my face.

Hey Folks! Practice doesn't make things perfect, just hopefully better. Let's keep practicing!

Hey Folks! This is not the ideal voice for such a power creature, i'm testing prompts. Check out the look on Junior!

Do you remember SPARK? What you might not know is that Sumowtawgha Junior and SPARK are best friends and are having a fantastic time in the urban jungle ⚽😎