Generor 登录注册

生成器公开

视频使用最新的 AI 模型生成和扩展视频——文本生成视频和图像生成视频

AI 模型

每个模型各有所长。所示费用为每秒视频的价格。

Hailuo 2 3-16

Advanced physics simulation for realistic complex movements. Supports end frame for loop creation.

Seedance 1.0 Pro Fast 3-12

Cheaper, faster variant of Seedance 1.0 Pro via BytePlus. ~50% cost savings vs the standard 1.0 Pro at the same resolutions.

Seedance 1.5 Pro 3-12

ByteDance Seedance 1.5 Pro via BytePlus. High-precision audio-visual sync, cinematic motion, and emotional expression. Supports first/last frames.

Seedance 1.0 Lite I2V 4-14

ByteDance Seedance 1.0 Lite Image-to-Video — efficient, cost-effective animation of source images. Direct via BytePlus.

Dreamina Seedance 2.0 Fast 5-20

Faster, cheaper sibling of Seedance 2.0 via BytePlus. Same multimodal capability set (text + image + video + audio references, native sync audio, multi-shot narration) at roughly 40% lower per-second cost. Trades a bit of fidelity for speed — ideal for iteration loops.

Seedance 1.0 Pro 6-30

ByteDance Seedance 1.0 Pro via BytePlus. Comprehensive and powerful video generation with strong motion control.

Dreamina Seedance 2.0 8-30

ByteDance flagship multimodal video model via BytePlus — accepts text + reference images + reference video + reference audio. Native synchronized audio, pro camera controls, multi-shot narration. Cheaper than the Replicate route and exposes capabilities Replicate hides.

P-Video 8

Pruna P-Video — fast, affordable text-to-video at $0.02/sec. Standard aspect ratios, 3-15 second clips. Strong bang-for-buck for prototypes and short-form content.

Grok Imagine Video 10

xAI Imagine API video — fast text-to-video and image-to-video at a flat $0.05/sec. Async polling pattern; supports 720p, 1-15 second durations. Native audio is included on every generation (cannot be turned off).

Wan 2.6 I2V Flash 10-15

Fast image-to-video with optional audio sync. Faster inference than standard Wan 2.6 I2V. Up to 15 seconds.

Happy Horse 1.0 I2V 14

Direct-to-DashScope Happy Horse 1.0 image-to-video. Strict consistency with the source image, fluent natural motion, native audio. 3-15 seconds.

Happy Horse 1.0 T2V 14

Direct-to-DashScope Happy Horse 1.0 text-to-video. Cheaper than the Replicate path. 3-15 second durations, five aspect ratios, native audio always on.

Happy Horse 1.0 R2V 20

Reference-to-video — combines up to 9 reference images for strong subject + scene consistency. Direct-to-DashScope only (Replicate proxy hides this).

Veo 3.1 Fast 20

Google's Veo 3.1 Fast with native audio and frame-to-frame generation. Supports start and end frames for seamless transitions.

Wan 2.6 I2V 20-30

Alibaba Wan 2.6 image-to-video with multi-shot storytelling, native audio, and precise lip-sync. Up to 15 seconds.

Wan 2.6 T2V 20-30

Alibaba's latest text-to-video model with multi-shot storytelling, native audio, and precise lip-sync. Up to 15 seconds.

Happy Horse 1.0 Video Edit 24

Local or global edits to an existing video using natural-language instructions and up to 5 reference images. Preserves original motion. Direct-to-DashScope only.

Kling v3 34-45

Kuaishou Kling v3 — cinematic text-to-video and image-to-video up to 15 seconds with native audio and lip-synced dialogue. Supports start and end frames. Standard mode = 720p, Pro mode = 1080p.

Veo 3.1 40

Google's flagship video model with strongest prompt adherence and cinematic motion. Synchronized native audio, reference images, and start+end frame control.

Kling v2.1 Master 56

Premium Kling model with enhanced quality and longer durations.

管理 Video 模型

质量

480P 5

720P 9

1080P 20

模式

文本转视频

图像转视频

视频转视频

宽高比

unspecified Unspecified

16:9 Wide

9:16 Tall

时长

5 seconds 25

6 seconds 30

7 seconds 35

8 seconds 40

10 seconds 50

12 seconds 60

15 seconds 75

起始图片

选择一张图片以制作成视频动画。AI 将根据你的提示词让它动起来。

上传起始图片

选择文件未选择文件

— OR —

输入网址

源视频

上传现有视频进行编辑。描述你想要的更改——替换、场景调换、重新着色。原有的运动会被保留。

上传源视频

选择文件未选择视频

— OR —

输入网址

或从您的素材库中选择：访问任意一个你的视频作品并点击“编辑视频”——您将进入此页面，并已预加载源文件。

结束帧（可选）

设置结束帧以引导视频过渡的方向。非常适合创建循环或可控动画。

将起始图片用作结束帧（平滑循环）

上传结束帧

选择文件未选择文件

— OR —

输入网址

参考输入 (可选)

在多次生成中固定角色的外观或动作风格。在提示中引用它们，格式为 [Image1], [Image2], [Video1].

[Image1] 参考图片

选择文件未选择文件

— OR —

输入网址

[Image2] 参考图片

选择文件未选择文件

— OR —

输入网址

[Image3] 参考图片

选择文件未选择文件

— OR —

输入网址

[Video1] 参考视频

选择文件未选择文件

— OR —

输入网址

视频提示词 *

计算中…

提示：描述你想要的场景、动作和风格。上传图片以进行图片转视频转换。

生成音频

否

是

图像来源

上传图片

从文本生成

图像模型

P-Image 1

Pruna P-Image (direct) — sub-second text-to-image built for production. Strong prompt adherence and crisp text rendering at standard aspect ratios. 阅读更多 →

Z-Image Turbo 1

Login required to use this model. Alibaba Z-Image Turbo — lightweight text-to-image with bilingual (Chinese+English) text rendering. Fast, cheap, flexible resolutions from 512×512 up to 2048×2048. 阅读更多 →

Flux Schnell 2

Ultra-fast image generation at a fraction of the cost. Perfect for quick iterations and testing ideas. 阅读更多 →

P-Image Edit 2

Pruna P-Image Edit — focused image editing with 1-5 reference images, sub-second inference, and exact-prompt adherence. Strict image-to-image (requires at least one source image). 256-1440 px output in standard aspect ratios. 阅读更多 →

Wan 2.6 Image 2

Login required to use this model. Alibaba Wan 2.6 text-to-image — strong cinematic photography and artistic styles. Renders at 1280×1280 to 1440×1440 with five aspect ratios. 阅读更多 →

Grok Imagine 4

Login required to use this model. xAI's Imagine API image model — fast generation and editing with strong prompt fidelity. 阅读更多 →

Qwen Image 4

Login required to use this model. Alibaba Qwen Image — flagship text-to-image with strong text rendering, multilingual prompt support, and crisp realism. 阅读更多 →

Wan 2.7 Image 4

Login required to use this model. Alibaba Wan 2.7 image — text-to-image, image editing with up to 9 reference images, and multi-image reference generation. Up to 2K resolution. Was previously via Replicate at ~30× the cost. 阅读更多 →

SeedEdit 3.0 6

Login required to use this model. BytePlus SeedEdit 3.0 — focused image-editing model. Takes a source image plus an instruction and applies the edit in-place ("make the bubbles heart-shaped", "change the sky to sunset"). Adaptive output sizing follows the input image dimensions. Strict image-to-image only (no pure text-to-image). 阅读更多 →

Seedream 4.0 6

Login required to use this model. BytePlus Seedream 4.0 — multimodal image creation with text + single-image + multi-image inputs. Multi-image blending up to 10 references, sequential image-set generation up to 15 outputs, 4K ultra-HD support. Direct via BytePlus. 阅读更多 →

Seedream 5.0 Lite 7

Login required to use this model. BytePlus Dola-Seedream 5.0 Lite — flagship image model with web-connected retrieval, multi-image fusion, image-set generation, and strong consistency preservation. Supports text-to-image and image-to-image (with up to 4 reference images). Up to 2K output. Direct via BytePlus. 阅读更多 →

Qwen Image Edit 8

Login required to use this model. Alibaba Qwen Image Edit — image-to-image with multi-reference editing, text-in-image rendering, and example-based composition. 阅读更多 →

Seedream 4.5 8

Login required to use this model. BytePlus Seedream 4.5 — refined image model with strong editing consistency, multi-image fusion, finer detail control, natural small-text and face rendering, and 2560×1440 to 4096×4096 output. Supports text-to-image and image-to-image with up to 4 reference images. 阅读更多 →

Wan 2.7 Image Pro 8

Login required to use this model. Alibaba Wan 2.7 image Pro — premium variant with 4K text-to-image, thinking mode for higher quality, image editing, and multi-image reference. 阅读更多 →

Seedream 4.0 9

Login required to use this model. Next-generation image model with unified generation & editing. Supports sequential image generation for coherence and multi-reference workflows. 阅读更多 →

Grok Imagine Quality 10

Login required to use this model. xAI's higher-quality Imagine image model — sharper detail and stronger composition than the standard Grok Imagine, at a higher per-image cost. 阅读更多 →

Gemini 2.5 Flash Image 12

Login required to use this model. Google's fast image model with good quality and speed. Excellent at following natural language prompts. ⚠️ This model only supports Square (1:1) and Match Input aspect ratios. 阅读更多 →

Seedream 4.5 12

Login required to use this model. Upgraded Bytedance image model with stronger spatial understanding and world knowledge. 阅读更多 →

Nano Banana 2 14

Login required to use this model. Google's latest fast image model with high-efficiency production-scale visual creation. Supports multiple resolutions up to 4K. 阅读更多 →

Nano Banana Pro 27

Login required to use this model. Google's professional design engine with a reasoning core for studio-quality 4K visuals, complex layouts, and precise text rendering. 阅读更多 →

管理 Image 模型