Skip to main content

​
Image Models (Modify Frame)

← Closed-source video models · Model catalog · Upscale models →
Image models edit, stylize, or transform a single frame in the Modify Frame workspace. Models accept a source frame plus an instruction-based prompt. Separately, script projects have a Reference image model dropdown that generates reference images from a prompt alone (see Script project reference images).
💡 Preservation directives — For most edits, instruct the model to preserve what should not change: “Keep the original composition.” · “Keep the original outlines.” · “Keep the original character intact.” · “Do not change the lighting.” · _“Keep the original framing and proportions.” These work with GPT Image 2, Nano Banana 2, Nano Banana Pro, and Seedream 5.0 Pro.

​
GPT Image 2

OpenAI image model. Strong at precise editing; respects original content very well.
  • Output resolution: 2K.
  • Prompting: instruction-based.
  • Quality setting: medium (default) or high. High improves fine detail like small text. Doesn’t change output size but increases generation time. Most useful paired with a video model that can preserve that detail.
  • Image style reference: optional input guiding the visual style (up to 9 images).
  • Mask image input: add your frame with the area you want to edit painted in black; it is used as the mask reference.
  • Availability: Relaxed and Unlimited modes.
Use when — Precision editing of a specific element. Adding/removing objects. Text-sensitive edits (GPT Image 2 handles text better than most models).

​
Nano Banana Pro

Strong general-purpose image edit and style transfer model.
  • Prompting: instruction-based.
  • Image style reference: optional (up to 9 images).
  • Availability: Unlimited mode only — not available in Relaxed mode.
⚠️ Known behavior — When adding an image reference, Nano Banana Pro may confuse which input is the source and which is the reference. If results are wrong, retry with explicit phrasing: “use the first image as the reference” or “use the second image as the source”; if it still fails, invert the images in the prompt.

​
Nano Banana 2

  • Output resolution: 0.5K, 1K, 2K (default), or 4K. Higher resolutions cost more credits.
  • Prompting: instruction-based.
  • Image style reference: optional (up to 9 images).
  • Best for: higher-resolution outputs, since it supports 4K natively.
  • Availability: Unlimited mode only — not available in Relaxed mode.

​
Seedream 5.0 Pro

ByteDance image editing model. Strong at stylization and instruction-based transforms.
  • Cost: 120 credits per image.
  • Reference images: up to 9 reference images accepted as style or content guidance.
  • Availability: Unlimited mode only — not available in Relaxed mode.
  • Drawn mask: paint a red mask by hand on the frame (from a video source) to limit the edit to a specific region. Tip: include “in the red area” in your prompt.
Use when — Style transfer or visual transformation where you want a different aesthetic look while preserving overall composition.

​
Qwen Image Edit +

Alibaba image editing model.
  • Prompting: instruction-based.
  • Reference images: up to 9, providing style/content context.
  • Availability: Unlimited mode only — not available in Relaxed mode.
Use when — General instruction-based edits or stylization, similar territory to Seedream 5.0 Pro.

​
Script project reference images

Script projects have a Reference image model dropdown that generates reference images from a prompt alone — no source frame needed. It is separate from the Modify Frame model list. Reference images generated from the script project always use Credits mode.
Use when — You need a generated reference image from scratch — a character design, background, or scene that doesn’t yet exist in your project.

​
Selection guide

← Closed-source video models · Model catalog · Upscale models →