Skip to main content
Panorama

Panorama · screenshot coming soon

What this node does

Panorama is ComfyTV’s 360° environment entry point. Its output is not a flat photo but an equirectangular panorama — imagine a flat “world map” where the horizontal axis wraps 360° and the vertical axis runs from sky to ground. The node opens a 3D viewer: you sit at the center and drag to look in any direction. Three ways to bring a panorama in, highest priority first:
  1. 📤 Upload — pick .jpg / .png / .webp / .hdr / .exr. The viewer loads instantly with a manual badge. Upload overrides everything else; clear it with ✕ before generating.
  2. Image-to-panorama — workflow Qwen-Image-Edit 2511 Image-to-Panorama, wire a photo to image. The source becomes the front view; the model pushes out the full 360°.
  3. Text-to-panorama — workflow Qwen-Image 2512 + 360 LoRA, fill main_prompt only, no reference image.
Panorama only shows the 3D viewer — no flat thumbnail. Use Panorama · Current View or Multi-View downstream for flat frames.

When to use it

  • HDRI / skybox references for 3D, games, or VR
  • Expand a regular photo into a look-around scene
  • Generate a 360° environment from text alone
  • Shared environment base for storyboard or multi-view shots

How ComfyTV designed this

  • Stage = a step with ▶ Run. Run executes this node only — it does not queue the entire ComfyUI graph.
  • Snapshot: after Run, the panorama URL is stored in the project. Downstream Current View / Multi-View reads that snapshot; it does not re-run Panorama generation automatically.
  • workflow dropdown = the ComfyUI sub-workflow JSON (workflows/panorama/). ComfyTV maps main_prompt and image; the stage auto-prefixes equirectangular 360 degree panorama, .
  • Upload mode needs no Run — pick a file and it loads immediately. AI generation requires Run.

Types (COMFYTV_* vs native ComfyUI)

Conversion: native IMAGE → ComfyTV via → ComfyTV Image; snapshots back to tensor via ← ComfyTV Image. See bridges.md.

Parameters

workflow

Backends from workflows/panorama/. Qwen-Image 2512 + 360 LoRA (text, ~2048×1024); Qwen-Image-Edit 2511 Image-to-Panorama (image input required). Click ▶ Run after choosing; skip when uploading only.

main_prompt

Scene description, e.g. “sunset over a mountain lake”. Do not repeat 360° trigger words. For image-to-pano, add extension hints (“expand outward into grassland and distant mountains”).

image (optional)

Reference COMFYTV_IMAGE for image-to-pano. Wire Image Stage or Image Picker.

manual_source (hidden)

Set by 📤 Upload. When non-empty, skips workflow. Files land in ComfyUI input/. ✕ Clear upload to generate instead.

custom_params

JSON overrides for workflow sidebar params. See sidebar-config-editor.md.

Outputs

Step by step

  1. Search Panorama, drop on canvas.
  2. Upload: 📤 Upload panorama, pick equirectangular / HDRI, drag in 3D.
  3. Text: workflow Qwen-Image 2512 + 360 LoRA, write main_prompt, ▶ Run.
  4. Image: Qwen-Image-Edit 2511 Image-to-Panorama, wire COMFYTV_IMAGE, prompt, Run.
  5. For flat frames: add Current View or Multi-View, connect panorama, operate viewer, Run.
This page covers one node only. For end-to-end workflows, multi-stage pipelines, type conversion, and design rationale, see the ComfyTV user guides on GitHub:

Repository and workflows

FAQ

Q: Uploaded file — Run does nothing new?
A: Upload wins. ✕ Clear upload, pick workflow, Run.
Q: What is equirectangular?
A: A projection that unwraps a 360° sphere onto a flat 2:1 rectangle. The viewer maps it on the inside of a sphere; you look outward from the center.
Q: No flat output — how do I edit?
A: Current View first → COMFYTV_IMAGE → edit stage.
Q: Workflow Run fails?
A: Check models.md for Qwen-Image / 360 LoRA; or upload instead.
  • Panorama · Current View / Multi-View — flat captures
  • Image Stage — reference for image-to-pano
  • → ComfyTV Image — plugin IMAGE into ComfyTV