Ever since I can remember, I've had VERY vivid dreams. I'm using AI to recreate them to be as close as I can to the real dream. I'm exploring with different tools to see which AI model creates the closest visuals.
Tools I use: Leonardo AI | Flux | Krea | Luma | Minimax | Runway
Dream 1
I dreamt...
I was in god mode, watching some ships sail along a forest floor towards the horizon of a magical forest.
I was in god mode, watching some ships sail along a forest floor towards the horizon of a magical forest.
How I re-made my dream
I wanted to try image-to-video and text-to-video to find the most accurate output.
For image creation I used Leonardo AI, Flux, Krea.
For video creation I used Luma, Minimax, Runway, Kling
I thought the most dream-accurate output was Luma.
I wanted to try image-to-video and text-to-video to find the most accurate output.
For image creation I used Leonardo AI, Flux, Krea.
For video creation I used Luma, Minimax, Runway, Kling
I thought the most dream-accurate output was Luma.
Best motion prompt - Luma
A fleet of small ships with full masts, sail in a line, through the grass on a forest floor, towards a sunny horizon, where there is a sunny meadow. The camera is a slightly above shot. The forest is dense on either side, the towering trees are tall and straight, and line both sides of the ship’s path, dappled sunlight streaming through the leaves. The camera moves slowly, capturing the ships sailing towards the golden coloured light, in a serene forest. The light source comes from the horizon, in centre of the frame.
A fleet of small ships with full masts, sail in a line, through the grass on a forest floor, towards a sunny horizon, where there is a sunny meadow. The camera is a slightly above shot. The forest is dense on either side, the towering trees are tall and straight, and line both sides of the ship’s path, dappled sunlight streaming through the leaves. The camera moves slowly, capturing the ships sailing towards the golden coloured light, in a serene forest. The light source comes from the horizon, in centre of the frame.
Most dream-accurate visuals
Krea did a great job at recreating text-to-image at the correct scale and perspective of my dream. But when it came to adding motion, it didn't capture the movement correctly.
Krea did a great job at recreating text-to-image at the correct scale and perspective of my dream. But when it came to adding motion, it didn't capture the movement correctly.
I tried to animate the images with motion on Runway.
It was very close to the dream, and even recreated the ship bobbing:
It was very close to the dream, and even recreated the ship bobbing:
Dream 2
I dreamt...
I had itchy eyes, and went to the bathroom and saw I had radicchio leaves instead of eyeballs.
I had itchy eyes, and went to the bathroom and saw I had radicchio leaves instead of eyeballs.
How I re-made my dream
I used 76 different prompts on Minimax (Hailio AI)
Pro tip: You can use time code to prompt shift in MiniMax. 00:01 - Prompt 1, 00:03 - Prompt 2. Basically like using images as start/end frames.
I used 76 different prompts on Minimax (Hailio AI)
Pro tip: You can use time code to prompt shift in MiniMax. 00:01 - Prompt 1, 00:03 - Prompt 2. Basically like using images as start/end frames.
Best motion prompt - Minimax
A front on shot, close-up of hyper-detailed eyeballs layered in cabbage leaves. We clearly see the skin of her eyelids and eyelashes on top of the cabbage. The red cabbage texture is flush against the small eyeballs, inside both eyes of a pretty 24 year old woman, smoothly wrapped around the eyeball's surface like a second skin. She shifts her gaze left to right. Her face is illuminated softly, with the focus remaining solely on her eyes, emphasizing the surreal yet hyper-realistic look of the red radicchio-covered eyeballs.
A front on shot, close-up of hyper-detailed eyeballs layered in cabbage leaves. We clearly see the skin of her eyelids and eyelashes on top of the cabbage. The red cabbage texture is flush against the small eyeballs, inside both eyes of a pretty 24 year old woman, smoothly wrapped around the eyeball's surface like a second skin. She shifts her gaze left to right. Her face is illuminated softly, with the focus remaining solely on her eyes, emphasizing the surreal yet hyper-realistic look of the red radicchio-covered eyeballs.
Failed attempts of Minimax, Luma
Minimax was struggling to make the eyes literally made of radicchio:
Minimax was struggling to make the eyes literally made of radicchio:
Luma's style of replicating humans felt too animated – likely from training the model on animation inputs:
Dream 3
I dreamt...
I was in the elevator of a decaying building in a war-torn desert landscape. I was moving upwards in this caged elevator which stopped at one of the floors which made a noise and alerted the killer robots, that turned to see me in the elevator and started to attack me.
I was in the elevator of a decaying building in a war-torn desert landscape. I was moving upwards in this caged elevator which stopped at one of the floors which made a noise and alerted the killer robots, that turned to see me in the elevator and started to attack me.
How I re-made my dream
I thought the most dream-accurate output was Luma.
I thought the most dream-accurate output was Luma.
Best motion prompt - Luma
On the 6th floor of a building, large scary Terminator robots stand on a metal grate floor, where you can clearly see many flayers of the building floors below. The building is decaying exposes a post-apocalyptic urban landscape, a vast chaos of rolling sand dunes, desert wasteland, engulfed in darkness at night. The robots resemble The Terminator, their large powerful mechanical body gleaming. The robot's face is a big, black camera lens instead of a head, surrounded by a glowing red LED ring, facing the camera.
On the 6th floor of a building, large scary Terminator robots stand on a metal grate floor, where you can clearly see many flayers of the building floors below. The building is decaying exposes a post-apocalyptic urban landscape, a vast chaos of rolling sand dunes, desert wasteland, engulfed in darkness at night. The robots resemble The Terminator, their large powerful mechanical body gleaming. The robot's face is a big, black camera lens instead of a head, surrounded by a glowing red LED ring, facing the camera.