The NEW Lighting Workflow Changing Filmmaking (Full AI Tutorial)
Provides AI filmmakers with a concrete, practical technique to change time-of-day and lighting setups in post-production using reference images and video-to-video diffusion models.
What this lesson covers
Caleb from Curious Refuge demonstrates an AI-driven workflow for relighting video clips and gray box studio footage in post-production. The process uses image generation models to alter a single reference frame's lighting and then applies video-to-video tools like Seedance 2.5 to transfer the new lighting across the entire scene.
Key takeaways from the creator
AI-extracted notes, not independently verified product claims. Timestamp links let you check each point in the original video.
- 01:44 ↗
Relighting video requires extracting a single start frame screenshot and pairing it with a visual style reference image that represents the desired lighting.
- 03:19 ↗
GPT image 2.5 Sunburst can be used to transfer complex lighting styles and color grades onto a reference image frame prior to video generation.
- 04:05 ↗
Applying both the source video clip and the re-lit reference frame into Seedance 2.5 enables consistent video-to-video relighting.
- 04:38 ↗
Current AI video relighting models may introduce technical artifacts such as exposure drift, diagonal banding, and resolution capping at 1080p.
- 05:27 ↗
Combining gray box studio filming with AI relighting allows filmmakers to substitute low-cost indoor props with complex lit environments like cliffs at sunset.
Workflow outlined in the video
- Capture a clean start frame screenshot from your target video clip.
- Select a high-quality style reference image containing your target lighting and color grade.
- Input the screenshot and style reference into an image generator like GPT image 2.5 Sunburst to generate a modified reference frame.
- Upload the original video clip and the newly lit reference frame into a video generator like Seedance 2.5.
- Prompt the video model to apply the reference frame's lighting and style onto the target video clip.
- Review the generated output for artifacts like exposure variation or diagonal banding, upscaling if needed for final distribution.
Before you use this workflow
These notes describe the source video at its publication date. Model access, pricing, connectors and interfaces may have changed. Check the original source and the provider’s current documentation before installing an add-on, connecting an account or spending credits.
ReelStack has not independently tested this workflow. Preview one representative shot and check motion, continuity and output quality before applying it to a full production. No result, cost saving or model capability is guaranteed.
Explore the tools
How to interpret the numbers
16,931 views captured 5 November 2026. ReelStack recommendation score: 52/100. Formula: radar-v5. This is ReelStack’s calculation, not a YouTube rating or a measure of factual accuracy.
Show the calculation
- 65% performance: 0.88× current views vs the channel's recent median; same-age history is not yet available. View-evidence factor 94% (views / (views + 1,000))
- 25% freshness: 76/100 with a 14-day half-life
- Momentum pending: collecting daily snapshots; its 20% weight goes to observed performance, not free points
- 10% engagement: 45/100; likes + 4× comments, smoothed with a 500-view neutral prior