👶 Face of Time 👵 Made with #LumaDreamMachine 1.5 Sound on 🔊 Complete workflow ⤵️

The Workflow : Creating the Images: Back in March, I used ChatGPT for prompts and Midjourney for generating the images in this video. You can find the detailed workflow in the post linked here.

Using LumaLabs for Transitions: LumaLabs has an awesome keyframes feature that makes it easy to transition between two images. Steps: Upload your start and end frame images, as shown below.

Prompting Tips: There's a Prompt Guide that’s super helpful. I recommend checking it out and experimenting. You can even load the whole guide into an LLM and customize it for your own prompts. For smooth transitions between two frames, I tried testing with and without prompting.

Results from Prompting: Without a prompt, the success rate was lower, especially when the subjects were facing different directions. Here’s the prompt that worked best for me: Prompt: slow seamless transition

Creating the Video: I generated a video for each pair of adjacent age frames (e.g., 1-2, 4-6, 42-46, etc.) using @LumaLabsAI Without prompts, and when the images were very different, I got some funny fails too. A couple of example below. First fail: smooth transition Second fail: slow smooth transition

Editing the Final Video: Once I had all the videos I was happy with, I used a video editor to stitch them together, adding age numbers and titles.

My video is inspired from the below post.

Adding Audio: For the music, I used Suno AI. I asked ChatGPT for a prompt to fit the theme. Prompt used for audio generation : “Create a gentle, heartwarming melody that evolves over time, reflecting the journey of life. The music should start with soft, innocent tones representing the early years, gradually growing in richness and depth as the years progress. Incorporate uplifting and nostalgic elements, with a subtle piano and strings theme, symbolizing the beauty and passage of time. The overall tone should evoke feelings of love, warmth, and reflection, ending on a hopeful and serene note.” Final Touches: The generated audio was longer than the video, so I cropped it down using simple online tools, then merged the audio with the video.

Summary: Images: ChatGPT and Midjourney for consistent visuals. Video: LumaLabs with keyframes (Better results with effective prompts. I had 4-5 fails and tweaked prompts immediately). Music: Suno AI. Editing: Used video editor tools to bring it all together. Thank you for reading! If you found this helpful, consider following me @umesh_ai and sharing the post!

Great work! This will help transition from one scene to another, but randomness is also present in AI - that's both the challenge and the beauty of it. 😅 Here's one of my shots that I particularly like, which works well with @LumaLabsAI photos from #GROK2.