nano-banana/pro/edit
Turn sketches into precise 2K-4K visuals with smart correction and seamless creative control.
- Consistent subject and style transfer from multiple reference images
- High-fidelity 4K output with sharp detail retention
- Strong prompt adherence for fine-grained edits
- Fast, production-ready inference on managed GPUs
- Deterministic reproduction via seeding
Vidu Q2 is an image-to-image model that turns reference images and prompts into high-quality 4K visuals with precise, controllable edits. It is optimized for production pipelines that need speed, consistency, and repeatability on RunComfy.
RunComfy provides a managed environment for Vidu Q2 so you can prototype in minutes and scale to production without wrangling infrastructure. You get consistent performance, simple versioning, and reproducible runs.
https://www.runcomfy.com/models/vidu/q2/reference-to-image/apiBelow are the supported inputs for Vidu Q2.
| Parameter | Type | Default/Range | Description |
|---|---|---|---|
| prompt | string | "" (max 1500 chars) | Natural language instruction guiding how the model should transform and stylize the output. Be explicit about subject, style, lighting, camera, and materials for best control. |
| reference_image_urls | array of image URLs | [] | One or more https URLs to reference images for identity/style consistency. Use 1-3 high-quality images of the same subject; ensure clear, well-lit framing for reliable transfer. |
| Parameter | Type | Default/Range | Description |
|---|---|---|---|
| aspect_ratio | enum string | 16:9 (choices: 16:9, 9:16, 1:1) | Target aspect ratio for the generated image. Match this to your composition (e.g., 9:16 for vertical portraits, 16:9 for landscapes, 1:1 for square creatives). |
| seed | integer | 0 (random) | Controls determinism. Use a fixed seed to reproduce a result; set 0 to randomize each run during exploration. |
Required parameter: prompt
These tips help get the best results from Vidu Q2 for image-to-image workflows:
Turn sketches into precise 2K-4K visuals with smart correction and seamless creative control.
Fast bilingual image creation engine with depth and pose guidance for precise, photoreal visual design.
Replace a photo’s background with a new scene using Ideogram 3.
Edit and fuse images into high quality results with Seedream 4.0.
Sync image edits, remixes, reframe, and background swaps for film.
Create detailed visual assets from prompts with scalable, high-speed precision
Vidu Q2 image-to-image is governed by its original license from ShengShu Technology, often aligned with permissive Open RAIL or similar frameworks allowing research and possible commercial use under stated conditions. RunComfy simply provides API and cloud hosting access — it does not alter or override the model’s original licensing structure. Users must review and adhere to Vidu Q2’s stated license before selling or distributing generated content.
Vidu Q2 image-to-image operates on cloud-hosted GPUs managed by RunComfy, offering low latency (around 3–6 seconds for 1080p) and stable concurrency through automatic GPU scaling. RunComfy manages GPU load balancing so users can run multiple concurrent generations without setup. Local execution isn’t supported due to high hardware demands.
Yes, Vidu Q2 image-to-image supports outputs up to 4K resolution, though 1080p is free-tier optimized. Each prompt is limited to roughly 512 tokens, and reference inputs via ControlNet or IP-Adapter are capped at four. Aspect ratios follow 1:1, 16:9, and 9:16 presets to ensure stability across generations.
You can prototype in the RunComfy Playground (https://www.runcomfy.com/playground) using Vidu Q2 image-to-image, then integrate your workflow via the RunComfy API. The API mirrors playground parameters for inputs, references, and post-processing options. For production use, you’ll need to purchase usd credits, authenticate with your RunComfy API key, and adjust concurrency or callback URLs for automated processing.
Vidu Q2 image-to-image vastly improves consistency, handling multiple references with refined spatial and identity control. It’s up to 2× faster than Vidu Q1 and handles both text-to-image and full-image editing in the same unified architecture. This makes it reliable for professional pipelines across advertising, animation, and concept art where visual continuity is critical.
RunComfy runs Vidu Q2 image-to-image entirely on managed GPU clusters in the cloud, eliminating the need for users to provision hardware. Resources are dynamic—users experience minimal latency and autoscaled throughput even at high demand. This managed approach separates inference performance from local computing limitations.
Yes, new RunComfy users receive free trial usd credits that can be used with Vidu Q2 image-to-image for generating outputs up to 1080p. This allows users to evaluate quality and performance before upgrading to paid usage for higher resolution or production-scale workflows.
For support or feedback regarding Vidu Q2 image-to-image usage, integration questions, or GPU availability, you can contact RunComfy directly at hi@runcomfy.com. The platform team assists with both technical troubleshooting and billing inquiries related to usd consumption or API access.
RunComfy is the premier ComfyUI platform, offering ComfyUI online environment and services, along with ComfyUI workflows featuring stunning visuals. RunComfy also provides AI Models, enabling artists to harness the latest AI tools to create incredible art.
