MetaView Novel View Synthesis

MetaView Novel View Synthesis

Renders a scene from a new camera angle using one uploaded image, with yaw and pitch controls. An interactive demo of the MetaView method for researchers, 3D artists and developers.

Made by hugging-apps on Hugging Face. These notes are Mindket's, written from its Space page and README; the app's own files and license are on Hugging Face.

About

What it is: This is an interactive demo of MetaView, a method for monocular novel view synthesis. Give it one picture and pick the camera direction you want, as left-right yaw and up-down pitch, and it produces the scene from that new position. The page ties the method to an ECCV 2026 paper. It runs on zero-a10g hardware at Hugging Face.

Who it suits: Vision and graphics researchers, 3D and visual effects artists who want a new angle on a still, and developers who want to study the pipeline.

Industry | Who uses it | How they use it | Example
Research | Vision researchers | Try single-image view synthesis | Turn the camera a little to the left
Visual effects | Artists | Get a new angle on a still for reference | Tilt the view up on a room photo
Games | Game developers | Look at a scene from another viewpoint | Check how a scene reads from the side
Software | Developers | Read the pipeline code in the repository | See how the models are combined

Skip it if you need several input photos, since the app works from a single image.

Where to use it:

• A web browser, on the Space's page at Hugging Face
• Your own computer, as a local copy, once Python 3.10 is installed
• The listing carries an MCP tag, but the page does not say how to use it that way

Use case | Who | What they get back
Render a new viewpoint | Artists and researchers | An image of the scene from the camera angle you chose
Study the pipeline | Developers | The code for the models and their combination

How to set it up:

• Nothing at all if you only use the online Space
• Python 3.10 and pip, for a local copy
• The Python packages listed in requirements.txt

1. Open the Space's page at Hugging Face to use it online.
2. For a local copy, install the Hugging Face hub package and download the Space.
3. Move into the new folder and install Gradio at the pinned version.
4. Install the packages from requirements.txt.
5. Start the app with Python.

pip install -U huggingface_hub
hf download hugging-apps/metaview --repo-type space --local-dir metaview
cd metaview
pip install gradio==5.49.1
pip install -r requirements.txt
python app.py

The repository is large, at 211 files, and includes examples. The page does not say what a local run needs beyond the packages.

How to use it: Upload one image, set the target camera yaw for left or right and pitch for up or down, then run it. The app renders the scene from the new viewpoint.

Tool | What it does
Yaw | Turns the target camera to the left or right
Pitch | Tilts the target camera up or down
python app.py | Starts the Gradio app on your computer

Example: You upload a photo of a living room, set the yaw a little to the right and leave the pitch alone. The app returns the room rendered from that shifted viewpoint, and you compare it with the original.

For newcomers and experts: If you are new, try a clear photo and a small yaw change first. Large turns ask the model to invent more of the scene, so results can drift from reality.

If you know the field, the pipeline pairs the Qwen-Image-Edit MM-DiT backbone with Depth-Anything-3 for implicit 3D priors and dense depth. A modified RoPE called PRoPE anchors the metric scale of the camera pose. The model list also names FLUX and other models that this description does not cover.

You upload one image
You set the target yaw and pitch
Depth priors anchor the geometry
The MM-DiT backbone renders the new view

Good to know:

• License: apache-2
• Cost to run: none named, though the Space runs on zero-a10g hardware at Hugging Face
• Data it reaches: the image you upload, and the model and dataset files it pulls from Hugging Face
• Not for: inputs with several photos, since it is built for a single image

On Hugging Face, 13 November 2026

• Likes: 8
• Built with: gradio
• License: apache-2.0

Links

• Space: https://huggingface.co/spaces/hugging-apps/metaview
• Files: https://huggingface.co/spaces/hugging-apps/metaview/tree/main

Try it

Open the Space in your browser, or copy it to your own Hugging Face account from its page. Nothing in this kit runs by itself.