Vidu Q4 Preview Reference-to-Video Spicy
Vidu Q4 Preview Reference-to-Video Spicy is a high-performance, uncensored video generation model designed for dynamic storytelling and multi-subject visual synthesis. By combining Vidu's next-generation Q4 preview architecture with the expressive "Spicy" rendering engine, it delivers high-fidelity cinematic motion, aggressive action sequences, and precise camera controls while supporting multiple reference images for character and scene continuity. As an uncensored model, it operates without restrictive safety filters, enabling creators to realize bold visual concepts, complex character interactions, and high-impact commercial assets. Supported across resolutions from 1080p to 4K, it offers a powerful foundation for cinematic production, animation workflows, and high-engagement social media content.
Related Models
Vidu Q4 Preview Reference-to-Video
Vidu Q4 Preview Reference-to-Video is a high-performance video generation model engineered for complex visual storytelling and consistent character animation. By leveraging Vidu's advanced Q4 preview architecture, it allows creators to input multiple reference images to maintain strict character features, stylistic tone, and environmental details across dynamic scenes. The model excels at rendering fluid, high-action motion, natural physics, and smooth camera path transitions across flexible aspect ratios and resolutions up to 4K.
Read Me
Vidu Q4 Preview Reference-to-Video Spicy API
Vidu Q4 Preview Reference-to-Video Spicy is a next-generation multimodal video generation model developed by Vidu (Shengshu Technology). It excels in high-consistency multi-subject reference generation and flexible first-frame/first-last-frame image-to-video generation. As a Spicy global version, it provides relaxed content policies for global users while maintaining the original generation capabilities.
On the iCreat platform, this model operates via a unified endpoint supporting both Reference-to-Video and Image-to-Video modes. It offers five resolution tiers from 540p to 4K, native audio synthesis, and flexible aspect ratios. Pricing is based on the output video duration at $0.045 to $0.39 per second.
Model Positioning
Vidu Q4 Preview Reference-to-Video Spicy is positioned as a high-consistency, multi-subject reference video generation model tailored for global users requiring flexible content policies. By unifying Reference-to-Video and Image-to-Video modes under a single endpoint, it serves developers and creators needing complex multi-subject consistency, native audio, and up to 4K resolution outputs.
Core Capabilities
Unified Dual-Mode Generation
Supports both Reference-to-Video (mode: "reference") for multi-subject reference generation and Image-to-Video (mode: "image") for first-frame/first-last-frame transitions via the same endpoint, simplifying integration and development.
High-Capacity Reference Inputs
Reference mode supports up to 15 reference images and up to 2 reference audio URLs, enabling highly consistent multi-subject video generation with native audio synthesis.
Flexible Resolutions and Aspect Ratios
Offers five resolution tiers (540p, 720p, 1080p, 2K, 4K) and five aspect ratios (16:9, 9:16, 3:4, 4:3, 1:1) to meet diverse creative and production needs.
Native Audio Synthesis
Generates native audio tracks synchronized with the video content. Audio output is enabled by default and can be explicitly disabled using the audio parameter.
Relaxed Content Policies
As a Spicy global version, it provides relaxed content policies for global users, expanding the range of creative expression while maintaining the original model's generation quality.
Pricing
| Resolution | Unit Price (USD/sec) | 5-Second Cost |
|---|---|---|
| 540P | $0.045 | $0.225 |
| 720P | $0.095 | $0.475 |
| 1080P | $0.12 | $0.60 |
| 2K | $0.19 | $0.95 |
| 4K | $0.39 | $1.95 |
Total Cost = Unit Price × Output Video Duration.
Application Scenarios
- Multi-subject character consistency video generation for storytelling and short dramas.
- First-frame and first-last-frame transition video creation for animation and visual effects.
- High-definition marketing and promotional video generation up to 4K resolution.
- Native audio-synchronized video generation for social media content.
- Global creative projects requiring relaxed content policies and flexible expression.
Model Comparison
Comparison Table 1: Multi-Subject Reference Video Model Comparison
| Model | Developer | Resolutions | Duration | Reference Inputs | Key Features | Pricing (USD/sec) |
|---|---|---|---|---|---|---|
| Vidu Q4 Preview R2V Spicy | Vidu | 540p–4K (5 tiers) | 4–16 sec | 15 images + 2 audios | Dual-mode unified endpoint, native audio | $0.045–$0.39 |
| Seedance 2.5 Spicy | ByteDance | 480p/720p | 4–30 sec | Image/video/audio multimodal | Video editing extension, web_search | $0.084–$0.303 |
| Vidu Q4 Preview Highlights | Vidu | 540p–4K | 4–16 sec | 15 images + 2 audios | Highest reference image limit, lowest 540p starting price | $0.045–$0.39 |
Note: Vidu Q4 Preview offers the highest reference image limit (15), the lowest 540p starting price, and its 4K tier is roughly half the price of Seedance 2.5's reserved tier.
Comparison Table 2: Same-Tier Multi-Resolution Video Model Comparison
| Model | Developer | 540P | 720P | 1080P | 2K | 4K |
|---|---|---|---|---|---|---|
| Vidu Q4 Preview R2V Spicy | Vidu | $0.045 | $0.095 | $0.12 | $0.19 | $0.39 |
| Seedance 2.0 Spicy | ByteDance | N/A | $0.164 | $0.409 | N/A | $0.83 |
| Kling V3 Omni | Kuaishou | N/A | $0.084 (5s=$0.42) | N/A | N/A | N/A |
Note: Vidu Q4 Preview offers the lowest unit price across all available tiers. The 540P tier is a unique low-cost entry point exclusive to Vidu.
Why Choose Vidu Q4 Preview Reference-to-Video Spicy?
- Unified endpoint supporting both Reference-to-Video and Image-to-Video modes.
- Highest reference capacity: up to 15 images and 2 audio tracks.
- Flexible output from 540p to 4K with native audio synthesis.
- Relaxed content policies for global users via the Spicy version.
- Cost-effective pricing starting at just $0.045 per second for 540p.
Specifications
| Field | Value |
|---|---|
| Model Name | Vidu Q4 Preview Reference-to-Video Spicy |
| Developer | Vidu (Shengshu Technology) |
| Model ID | vidu/q4-preview-global |
| Model Type | Multimodal Video Generation Model |
| Endpoint | https://api.icreat.ai/v1/task/submit/vidu/q4-preview-global |
| Task Result Endpoint | https://api.icreat.ai/v1/task/result |
| Authentication | Authorization: Bearer ${ICREAT_API_KEY} |
| Generation Modes | reference, image |
| Input Modalities | Images, Text, Audio |
| Output Modalities | Video, Audio |
| Supported Resolutions | 540p, 720p, 1080p, 2K, 4K |
| Aspect Ratios | 16:9, 9:16, 3:4, 4:3, 1:1 (reference mode only) |
| Duration per Request | 4–16 seconds |
| Reference Limits | Reference mode: 1–15 images + 1–2 audios; Image mode: 1–2 images |
| Native Audio | Supported (default true) |
| Billing Unit | USD per second |
| API Mode | Asynchronous two-step (submit task → poll result) |
Architecture
The model utilizes an asynchronous two-step API architecture. Users submit a generation task via the endpoint, receiving a task_id. The client then polls the task result endpoint using the same task_id until a terminal state (SUCCEEDED, FAILED, or NOT_FOUND) is reached. The request body uses a top-level flat JSON structure without nested input or parameters objects.
Notes
- Both steps use the same
task_id. While processing, theresultarray is empty ([]); continue polling. - FAILED and NOT_FOUND are terminal states. Do not continue polling; check your request parameters or media URLs.
- Generated media URLs are temporary. Download and persist them immediately; do not hardcode temporary URLs.
- The request body must be a top-level flat JSON. Do not wrap fields in
inputorparameters. - The
modeparameter is case-insensitive. Inreferencemode,promptis required and non-empty; inimagemode, it is optional. soundsis only supported inreferencemode (up to 2 items). Passing a non-emptysoundsarray inimagemode will return an HTTP 400 error.aspectRatiois only valid inreferencemode. Do not pass it inimagemode (aspect ratio follows the first frame).isRecis only available inimagemode. Do not pass it inreferencemode.audiodefaults totrue. Set tofalseto explicitly disable audio output.seedcan be specified to fix the generation seed for reproducibility.
Frequently Asked Questions
How do I call Vidu Q4 Preview Reference-to-Video Spicy?
It uses a two-step asynchronous process. First, submit the task to the endpoint with Bearer authentication and a top-level flat JSON body. Then, poll https://api.icreat.ai/v1/task/result using the returned task_id until the task succeeds or fails.
How is the cost calculated?
Total Cost = Unit Price × Output Video Duration, billed based on the selected resolution tier. For example, generating a 5-second video at 720p costs $0.095 × 5 = $0.475. The costUSD field is returned in the response upon success.
What is the difference between reference mode and image mode?
Reference mode supports 1–15 images plus up to 2 audios, requires a non-empty prompt, and allows setting aspectRatio. Image mode supports 1–2 frame images, makes prompt optional, automatically follows the first frame's aspect ratio, and allows the isRec parameter. Passing sounds in image mode will result in a 400 error.
What resolutions and aspect ratios are supported?
Supported resolutions are 540p, 720p, 1080p, 2K, and 4K. Supported aspect ratios are 16:9, 9:16, 3:4, 4:3, and 1:1. Aspect ratios can only be specified in reference mode.
What are the limits for reference materials?
In reference mode, you can provide up to 15 reference images and 2 reference audios. In image mode, you can provide 1–2 frame images. All provided URLs must be publicly accessible.
Are the generated result links permanent?
No, the generated media URLs are temporary. It is recommended to download them as soon as possible and persist them using the download_url.