Back to Nodes
Qwen Edit Multiangle

Qwen Edit Multiangle

Official

Camera‑aware image editing with multi‑angle LoRA – rotate, tilt, zoom, or wide‑angle for quick, realistic edits.

Nodespell AI
AI / Image / Qwen

Camera‑aware image editing with multi‑angle LoRA – rotate, tilt, zoom, or wide‑angle for quick, realistic edits.

Model Overview

A single‑image editor that understands camera motion. By supplying rotate_degrees, move_forward, vertical_tilt, and use_wide_angle sliders, the model generates a camera‑directive and then runs the Qwen‑Image‑Edit‑2509 pipeline in Lightning mode with the multi‑angle LoRA. The result keeps subjects, lighting, and textures consistent while applying the requested viewpoint change.

Best At

  • Smooth orbital moves – easy to rotate around a subject.
  • Dolly‑zoom and perspective changes – adjust framing with move_forward and use_wide_angle.
  • Subtle tilts – low‑angle or top‑down shots via vertical_tilt.
  • Fast generation – Lightning + LoRA give an 8‑step quick preview; set go_fast = false for a more detailed 40‑step pass.

Limitations / Not Good At

  • Extremely large angles or extreme tilts can produce distortion or artifacting.
  • Heavy style or content prompts may conflict with the camera directive, leading to hallucinations.
  • Not designed for adding or removing objects; best for viewpoint changes.

Ideal Use Cases

  • E‑commerce – rotate or zoom product shots for catalog pages.
  • Portrait retouching – create low‑angle hero shots or top‑down angles.
  • Scene re‑framing – change the camera viewpoint in photos of interiors or landscapes.
  • Quick style tweaks – add lighting, film grain, or color‑grading after adjusting the camera.

Input & Output Format

Input: JSON with required image (URL or file path) and optional parameters such as prompt, camera sliders, aspect_ratio, output_format, etc.

Output: An array of image URLs (single string per result) pointing to the edited image.

Performance Notes

  • Default go_fast = true gives an 8‑step Lightning run – roughly 5 × faster than the 40‑step base.
  • Switching go_fast = false enables 40 steps for maximum detail at the cost of speed.
  • Memory usage is moderate; a single image edit fits comfortably on a mid‑range GPU.
  • Quality can be set with output_quality when using JPEG or WEBP.
Inputs (2)

Image

String

Image file uploaded to Cog (jpeg, png, gif, or webp).

Min: 0Max: 100

Prompt

String

Optional text instruction appended after the camera directive.

Multi InputMin: 0Max: 100
Parameters (16)

Seed

Number

Random seed. Set for reproducible generation.

Default: -1

Prompt

String

Optional text instruction appended after the camera directive.

Default:

Go Fast

Boolean

If num_inference_steps is omitted, true runs the 4-step fast preset and false runs the 40-step detailed preset.

Default: true

Lora Scale

Number

Strength applied to the selected LoRA.

Default: 1.25

Aspect Ratio

String

Aspect ratio for the generated image

Default: match_input_image

Lora Weights

String

LoRA weights to apply. Pass a Hugging Face repo slug like 'dx8152/Qwen-Edit-2509-Multiple-angles' or a direct .safetensors/zip/tar URL (for example, 'https://huggingface.co/flymy-ai/qwen-image-lora/resolve/main/pytorch_lora_weights.safetensors', 'https://example.com/lora_weights.tar.gz', or 'https://example.com/lora_weights.zip').

Default: dx8152/Qwen-Edit-2509-Multiple-angles

Move Forward

Number

Move the camera forward (zoom in). Higher values push toward a close-up framing.

Default: 0

Output Format

String

Format of the output images.

Default: webp

Vertical Tilt

Number

Vertical camera tilt. -1 = bird's-eye, 0 = level, 1 = worm's-eye.

Default: 0

Output Quality

Number

Quality when saving the output images, from 0 to 100. 100 is best quality, 0 is lowest quality. Not relevant for .png outputs.

Default: 95

Rotate Degrees

Number

Camera rotation in degrees. Positive rotates left, negative rotates right.

Default: 0

Use Wide Angle

Boolean

Switch to a wide-angle lens instruction.

Default: false

Num Inference Steps

Number

Explicit denoising step count (1-40). Leave blank to use the go_fast presets (4 or 40 steps).

Default: 1

True Guidance Scale

Number

True classifier-free guidance scale passed to the pipeline.

Default: 1

Disable Safety Checker

Boolean

Disable safety checker for generated images.

Default: false

Advanced

String
Default:
Outputs (1)

Output

Inferred

Output

Nodespell

Nodespell

📍 London

Building the future. Join us!

Type

Node

Status

Official

Package

Nodespell AI

Category

AI / Image / Qwen

Input

ImageText

Output

Image

Keywords

Image EditAspect ControlResolution Control
Use in Workflow