Vidu Multi Subject Consistency
Vidu is Shengshu Technology's video generation family, created for coherent motion and stable subjects across changing shots. The Q series supports multi subject control, intelligent camera switching, smooth motion, and synchronized audio and video output for cinematic storytelling. Atlas Cloud adds Day 0 access across the expanding model lineup, keeping new releases close to your production pipeline. Start building today.
Vidu is developed by ShengShu Technology. Atlas Cloud (operated by Atlas Cloud AI LLC) provides access to it and does not own it. All trademarks belong to their respective owners.
Explore the Leading Vidu
Atlas Cloud provides you with the latest industry-leading creative models.
Choose the Right Vidu API Workflow for Video Generation
Compare Vidu endpoints for reference guided video, start and end frame generation, and single image animation, including each model’s standard Atlas Cloud base price.
| Modality | Description |
|---|---|
| Vidu Reference-to-Video Q1 API (Reference to Video) | Reference imagery guides the Q1 endpoint as it generates video output. Atlas Cloud lists a standard base price of $0.40, making this model suitable for reference guided concept clips and visual storytelling workflows. |
| Vidu Reference-to-Video 2.0 API (Reference to Video) | For reference driven production with Vidu 2.0, supply reference imagery and receive generated video. Its standard base price is $0.20 on Atlas Cloud, which suits recurring character, product, and scene visualization tasks. |
| Vidu Start-End-to-Video 2.0 API (Start and End Frames to Video) | Provide opening and closing frames, and this Vidu 2.0 endpoint generates the video between them. Atlas Cloud records a standard base price of $0.075, positioning it for planned transitions, visual reveals, and storyboard sequences. |
| Vidu Image-to-Video 2.0 API (Image to Video) | A source image becomes generated video through the Vidu Image-to-Video 2.0 endpoint. With a standard Atlas Cloud base price of $0.075, it fits product image animation, social content, and still image motion workflows. |
Inside the Vidu API: Consistency, Sound, and Cinematic Control
From four image reference consistency to native audio and sixteen second 1080p clips, the Vidu API on Atlas Cloud puts Shengshu's entire Q series behind one pay-as-you-go endpoint.
Subject Consistency with the Vidu API
The Vidu API accepts one to four reference images and holds characters, products, and scenes visually consistent across every frame. That reliability makes serialized storytelling and brand safe product videos practical at scale.
Start and End Frame Interpolation
Upload a start frame and an end frame, and the model builds the smooth transition between them on its own. Motion designers use this for morphs, reveals, and precisely timed scene changes.
Native Audio in a Single Pass
Why render silent clips and score them later? Every Q3 endpoint can generate synchronized sound effects and background music in the same pass, so dialogue scenes and ads arrive ready to publish.
Sixteen Seconds, Up to 1440p
If a scene needs room to breathe, run clips up to 16 seconds at 540p to 1080p, with super resolution upscaling to 1440p. Fewer stitched shots means smoother stories and lighter post production.
Tunable Motion and Style
A movement amplitude setting shifts output from subtle cinematic drift to large dramatic action, while text to video endpoints switch between general and anime styles. Directors get shot level control without rewriting prompts.
Vidu vs Other Models - One Prompt
The same prompt, generated by Vidu and other leading video models
A lone stunt driver drifts a matte black muscle car through a neon-lit night market in the rain. Open on a low tracking shot skimming wet asphalt as the car slides around a corner, sparks flying off a guardrail. Whip pan to vendors yanking their stalls back, paper lanterns swinging wildly. Cut to a drone shot spiraling overhead as the car threads between two buses with inches to spare, then a hard cut inside the cockpit: hands snapping the wheel, eyes flicking to a mirror filled with flashing police lights. End on a slow-motion 180 degree spin into a steaming alley, engine roar and rain hiss filling the soundtrack. Photorealistic, cinematic lighting, 10 seconds, 16:9 aspect ratio.
Vidu
Seedance 2.0
Vidu Q2 Pro
Anime style: a young swordswoman faces a colossal storm dragon above a shattered floating shrine. Start with a fast dolly-in on her eyes as lightning splits the sky, then a whip pan following her sprint across a crumbling stone bridge, debris lifting around her feet. The camera orbits as she leaps, her blade igniting with blue flame, while the dragon's jaws crash down where she stood a frame earlier. Pull back to an epic wide shot as her strike cleaves a lightning bolt in half and the shockwave ripples through the clouds. Crisp 2D cel shading, dynamic speed lines, thunder and a swelling choir in the score. 8 seconds, 16:9 aspect ratio.
Vidu
Seedance 2.0
Vidu Q2 Pro
Vidu Reference Images Power Repeatable Visual Stories
Vidu uses reference images to anchor consistent characters, products, and visual worlds across episodic storytelling, branded campaigns, preproduction concepts, social series, and educational content through one pay-as-you-go API on Atlas Cloud.
Episodic Characters with Vidu
Reference images help Vidu preserve a character’s identity while prompts place that subject in new scenes and actions. Series creators can build episodic shorts, character campaigns, and connected story chapters with stronger visual continuity.
Product Stories from Shared References
Use product reference images to keep shapes, colors, and details present as the setting and camera direction change. Commerce teams can produce campaign concepts, catalog motion clips, and launch stories around the same item.
Reusable Brand Worlds
Logos, mascots, and signature environments can serve as references, guiding Vidu toward consistent brand elements across fresh scenes. Marketing studios gain reusable visual foundations for seasonal campaigns, localized variants, and multi-shot brand storytelling.
Preproduction Concept Testing
When a script needs visual proof, combine reference subjects and written direction to generate coherent concept clips before production. Directors can evaluate casting, blocking, scene continuity, and campaign ideas without committing to a shoot.
Recurring Social Personas with Vidu
Could one recognizable persona anchor a whole social series? Vidu uses reference images to maintain that subject across prompted situations, helping creator tools produce recurring hosts, virtual personalities, and serialized short-form content.
Vidu for Connected Learning Stories
Anchor educational narratives with reference images of artifacts, settings, or recurring guides, then prompt new actions while helping preserve their visual identity. Museums and learning teams can create explainers, cultural stories, and lesson sequences.
Vidu Models Versus Multimodal Video Alternatives
Compare Vidu generation workflows, supported reference inputs, and standard base prices with selected reference-to-video models available on Atlas Cloud.
| Model | Input Workflow | Reference Inputs | Standard Base Price |
|---|---|---|---|
| Vidu Reference-to-Video 2.0 | Reference to video | 1 to 3 images | $0.20 |
| Vidu Start-End-to-Video 2.0 | Start and end frames to video | 2 images: start and end frames | $0.075 |
| Image-to-video-2.0 | Image to video | 1 source image | $0.075 |
| Vidu Reference-to-Video Q1 | Reference to video | 1 to 3 images | $0.40 |
| Seedance 2.5 Reference-to-Video | Multimodal reference to video | Up to 30 images, 10 videos, and 10 audio clips | $0.167 |
| Wan-3.0 Reference-to-video | Multimodal reference to video | Up to 20 mixed image, video, and audio references | $0.05 |
How to Use Vidu on Atlas Cloud
Get started in minutes — follow these simple steps to integrate and deploy models through Atlas Cloud's platform.
Create an Atlas Cloud Account
Sign up at atlascloud.ai and complete verification. New users receive free credits to explore the platform and test models.
Why Use Vidu on Atlas Cloud
Combining the advanced Vidu models with Atlas Cloud's GPU-accelerated platform provides unmatched performance, scalability, and developer experience.
Performance & flexibility
Low Latency:
GPU-optimized inference for real-time reasoning.
Unified API:
Run Vidu, GPT, Gemini, and DeepSeek with one integration.
Transparent Pricing:
Predictable per-token billing with serverless options.
Enterprise & Scale
Developer Experience:
SDKs, analytics, fine-tuning tools, and templates.
Reliability:
99.99% uptime, RBAC, and compliance-ready logging.
Security & Compliance:
SOC 2 Type II, HIPAA alignment, data sovereignty in US.
Vidu API FAQ for Consistent Video Generation
Vidu is a family of AI video generation models developed by ShengShu Technology in collaboration with Tsinghua University. It supports text to video, image to video, reference to video, and start and end frame workflows, including models designed to preserve subject identity across a clip.
Use reference images to place recognizable characters, products, objects, or environments into a generated video. This workflow is useful for recurring characters, branded product scenes, short narratives, advertisements, and other projects that need visual continuity.
Reference to video uses supplied images to guide subject identity without treating them strictly as the opening frame. Image to video animates a source image, while start and end frame generation creates the motion connecting two specified frames.
Create an Atlas Cloud API key and send a POST request to /api/v1/model/generateVideo with Bearer authentication. Include the selected model identifier and its required inputs, such as a prompt and reference images. The response provides an ID, status information, and output URLs when available.
Accepted reference formats include PNG, JPEG, JPG, and WebP, supplied through accessible URLs or valid Base64 data. Limits depend on the endpoint: Reference to Video 2.0 accepts one to three images, while Q3 Mix accepts one to four. Check the selected model schema before constructing the request.
For better consistency, use clear reference images and describe each subject, action, setting, and camera movement explicitly. Keep identities and clothing visually compatible across inputs, and avoid combining conflicting scene instructions in one prompt.
Atlas Cloud lists standard per generation prices of $0.40 for Reference to Video Q1, $0.20 for Reference to Video 2.0, and $0.075 for either Image to Video 2.0 or Start and End to Video 2.0. Other models have their own pay as you go rates, so review the selected endpoint before submitting a job.
Generation runs asynchronously, so retain the task ID returned after submission and monitor its status. Once processing succeeds, retrieve the generated video from the output URL in the result. Store completed media promptly if your application requires long term availability.
First verify the API key, model identifier, required parameters, image accessibility, file format, and endpoint specific limits. If identities drift, try clearer references with consistent lighting and fewer conflicting details, then simplify the prompt. Use the returned error information to diagnose rejected or failed requests.
Explore More Families
Seedance 2.5
The Seedance 2.5 API gives developers ByteDance Seed’s joint audio and video model, generating 4 to 30 seconds in one pass from text, a first frame, or up to 50 references across images, video, and audio. It edits selected regions, extends clips, and follows white-model and green-screen inputs. Atlas Cloud, an authorized U.S. reseller, adds a 480p draft mode, native 1080p finals, and enhanced 4K delivery with pay-as-you-go pricing.
Kling 4.0
The Kling 4 API delivers Kling 4.0, the video model Kuaishou released on October 15, 2026. It generates up to 30 seconds in a single pass, follows up to 10 keyframes, blends 15 images, videos, and subjects through Omni Reference, and adds stereo audio with lip sync. Atlas Cloud serves all four endpoints, including video editing, on one OpenAI-compatible key from $0.11 per second. Start building today.
Seedream 5.5
Seedream 5.5 is ByteDance Seed's image generation and editing family. Guide edits with up to 10 reference images, or split an image into a base image and up to 99 editable layers, with optional OCR for text layers. On Atlas Cloud, generation, editing, and layer decomposition each cost a standard $0.02 per image, helping developers plan image workflow budgets. Start building today.
Wan 3.0
Wan 3.0 is Alibaba’s newest video model, producing up to 30 seconds of 1080p video at 30fps with audio in one generation. Steer it with a text prompt, a first frame, up to 10 reference images, or a document. Standard costs less per second, while Prime keeps the same capabilities and returns results faster. Both run on the same Atlas Cloud key as Wan 2.7, 2.6, and 2.5.
GPT Image 2.6
The GPT Image 2.6 API gives developers OpenAI’s image model released on November 4, 2026, in two versions: Flare for fast iteration and Sunburst for detailed visual work. It generates images 50% faster than before and adds animated GIF output, SVG export, markup edits that target the area you circle, and a complexity setting. Atlas Cloud serves both models on one OpenAI-compatible key with pay-as-you-go pricing.
MiniMax H3
MiniMax H3 is MiniMax's multimodal video family for text, image, and reference guided creation. Across supported routes, it preserves subjects from reference media, offers flexible aspect ratios, and pairs generated sound with visuals through H3 Developer, with output profiles selected by endpoint. Atlas Cloud unifies the family behind one OpenAI-compatible key with transparent pay-as-you-go pricing from the standard rate of $0.038 per second. Start building today.
GPT Image 2.5
GPT Image 2.5 is OpenAI’s image generation and editing model, released on September 8, 2026 in two API versions. Flare is the default, with up to 50% lower latency than GPT Image 2, while Sunburst takes longer to give finer control over detailed edits. Both support up to 16 reference images, five quality tiers, transparent backgrounds, and custom sizes up to 3840px. Atlas Cloud serves the GPT Image 2.5 API at OpenAI’s per token rates, or at a flat price per image by 1K, 2K, or 4K tier.
Seedream 5.0 Pro
Seedream 5.0 Pro API gives developers ByteDance's controllable image editing model on Atlas Cloud. It places edits precisely with anchors and coordinates, separates images into editable layers, fuses multiple references, and matches exact colors and materials, with multilingual text at 2K and 3K. On Atlas Cloud you reach it through one key!
Seedance 2.0
Seedance 2.0 is ByteDance’s production video model for precise shot creation. Turn prompts into video, animate a first-frame image with optional last-frame guidance, or shape results with reference media and optional web search. Atlas Cloud brings these workflows into one unified API with transparent pay-as-you-go pricing and one OpenAI-compatible key. Start building today.
Gemini Omni 1.2 Flash
Kling V3.0
From Kuaishou, the Kling V3 API family supports text, image, and reference video workflows. Transfer motion while preserving character identity, guide generation with visual references, and edit videos through natural language commands. Atlas Cloud offers Standard and Pro endpoints for different production needs. Update: Kling V4 is now available on Atlas Cloud. Explore it on the[Kling V4](https://www.atlascloud.ai/zh/models/kling-v4) page.
Grok Imagine Video
Grok Imagine Video gives developers access to xAI’s video family for creating, animating, extending, and editing clips. Generate 1 to 15 second videos from prompts, guide people, objects, or styles with up to seven reference images, and work at 480p or 720p. Atlas Cloud brings these capabilities together through OpenAI-compatible endpoints with transparent pay-as-you-go pricing. Start building today.
Recommended Articles
Guides, tutorials, and product updates to help you get the most out of Atlas Cloud.
