Skip to main content

​
Editor Workflow

product loop

​
Select. Prompt. Refine. Accept. Continue.

The Iris editor is built around changing a specific part of a specific moment, then preserving enough continuity that the result feels authored instead of isolated. This prompt-driven loop is designed to be iterative, allowing for precise control and refinement at each step.

​
The Iris Editing Loop

The core workflow in Iris can be broken down into these key stages:

​
1. Select

Identify the specific segment or region within your video that you want to modify. This could be a time range on the timeline, or a bounding box around an object or area within a frame.

​
2. Prompt

Describe the desired change using natural language. Iris’s AI agent interprets your prompt to understand the intent, leveraging its understanding of the video content and available tools. The agent can also help you understand the current state of your reel using tools like get_timeline, preview_frame, and preview_strip.

​
3. Refine (New!)

After an initial generation, Iris provides a suite of tools to refine the edit without needing to regenerate from scratch. The agent can assist with:
  • Timeline Manipulation: split_segment, trim_segment, and delete_segment allow for precise control over the video’s structure.
  • Visual Adjustments: color_grade and grade_preview enable fine-tuning of visual properties like brightness, contrast, and saturation.
  • Variant Evaluation: score_variant provides detailed AI-based feedback on how well a generated variant meets the prompt or compares to the original, helping you choose the best take or guide further iterations.
  • Continuity Checks: The agent can also use score_continuity to ensure edits blend seamlessly.
  • Snapshots: Before making significant changes, the agent can snapshot_timeline to preserve a state for easy reverts.
This iterative refinement process ensures you achieve the exact look and feel you envision.

​
4. Accept

Once you’re satisfied with a generated variant or a refined edit, you accept it. This decision is then propagated through the timeline, ensuring continuity and consistency across your project.

​
5. Continue

With the edit accepted, you can move on to the next localized change, or export your finished video using export_video and track its progress with get_export_status. The goal is to build up a complete, polished video through a series of focused, prompt-driven adjustments.

​
Key Concepts

  • Localized Edits: Focus on small, specific changes rather than global adjustments.
  • Prompt-Driven: Use natural language to describe your creative intent.
  • Continuity-Aware: Iris helps maintain visual and temporal consistency across edits.
  • Iterative Refinement: Tools and agent capabilities support a detailed, step-by-step approach to perfecting your video.