Do I need a reference image, or can it work from text only?

The workflow supports both. For consistent identity and pose, load a reference image via LoadImage. If you prefer pure text, provide a clear description to the GeminiNode; the optimized prompt will drive Seedance 2.0 without a reference, though subject fidelity may vary.

How do I get a cleaner orbit and fewer artifacts around the subject?

Use a high‑resolution image with the subject centered, good contrast, and some background margin. Avoid heavy motion blur and occlusions. In ByteDance2ReferenceNode, increase reference guidance/strength if available to better anchor identity, and try a different seed for stability. Prompts that explicitly mention orbit direction, focal length (e.g., wide or telephoto), and suspended particles can improve coherence.

Why doesn’t it do a full 360° around my subject?

From a single image, the model infers unseen views; it can simulate an orbit but cannot reveal fully occluded sides reliably. You can suggest a partial arc in the prompt (e.g., “quarter‑orbit left to right”) and keep backgrounds simple to reduce hallucinations.

Do I need API keys or special hardware?

GeminiNode and ByteDance2ReferenceNode typically call external services and require valid API keys or configuration in their node settings. Check each node’s docs or your ComfyUI custom node configuration. Because the heavy lifting may be remote, local hardware demands are modest; generation time depends on the service and settings.

VFX - Bullet Time Effect - ComfyUI Workflow

This ComfyUI workflow builds a convincing bullet time sequence from either a reference image or a plain-text scene description. It does this in two stages. First, the Generate Bullet Time Seedance 2.0 Prompt group uses a GeminiNode to synthesize a single, highly focused prompt that emphasizes frozen action, slow‑motion energy, and a cinematic orbit around your subject. If you load a still via LoadImage, that visual context informs the prompt; otherwise, the node relies purely on your text description. A MarkdownNote in the graph explains what the node expects and how to phrase your ideas.

Next, the Generate Video group feeds the optimized prompt (and, when provided, your reference image) into the ByteDance2ReferenceNode, which drives Seedance 2.0 to render the shot. The node is the core video generator: it accepts the guidance prompt and optional reference to anchor subject identity and pose while simulating an orbiting camera and time‑frozen moment. PreviewAny lets you check the output right away, and SaveVideo writes the final clip to disk. The result is a short, loop‑ready bullet time pass you can iterate on quickly by adjusting the prompt, seed, or reference strength.

API

Use this workflow from code

Every Comfy workflow is a JSON graph. The payload below is this workflow, exactly as ComfyUI runs it — fetch it from the URL, keep it in version control, or load it in ComfyUI and run it node by node.

7a6ab8b6e694.json

Fetching workflow JSON…

Run it from TypeScript or Python with the Comfy SDK. The same code targets Comfy Cloud or a ComfyUI you host yourself — only the base URL changes.

// Install (beta)
npm i @comfyorg/sdk

// Run this workflow (TypeScript)
import { Comfy } from "@comfyorg/sdk";

const client = new Comfy({ apiKey: "comfyui-..." });
const wf = await client.workflows.fromFile("workflow_api.json");
const job = await client.run(wf);
await job.getOutputs("<output-node-id>")[0].toFile("output.png");

The SDK takes a workflow in API format: open this workflow in ComfyUI and use File → Export Workflow (API).

Comfy Cloud API access requires a plan with an API key.

SDK docs Get an API key

FAQ

Frequently Asked Questions