P-Video
输出预览
任务完成后,最新视频会直接出现在这里。
还没有视频
先写提示词,如果需要再加起始图或音频,然后生成第一条测试片段。
用 P-Video 把提示词、单帧图片和音频一起变成动态镜头
P-Video 适合快速创意迭代。你可以先用更低成本的草稿模式试镜头,用静态图或上传音频去约束画面,再在方向确定后切到更清晰的 1080p 输出。
为快速视频迭代做的一套产品化工作台
这个页面把 P-Video 包装成更顺手的站内工作流:快速试动效、在文本、图片和音频三种驱动方式之间切换,并在不暴露底层供应商细节的前提下完成更高质量的输出。
真实运行样例
这些案例展示了 P-Video 最擅长的几类输出:近景人物、音频驱动动作,以及从草稿到终稿的高频试片流程。
带吉他演唱的说话头像
当近景人物和清晰音轨一起输入时,适合做口型、表演和对镜头的短视频内容。
从单张图片延展动作镜头
先给一张静态图,再通过提示词去推动人物动作和镜头推进,适合高频试创意。
快速角色表演测试
草稿模式适合先确认角色节奏和动作,再在相同构图下切换更高质量输出。
P-Video workflow, step by step
Lock the look with a seed frame, test motion with quick 720p draft previews, then switch draft mode off for the final render at the same framing.
P-Video workflow at a glance
P-Video on this page runs Text to Video, Image to Video, and Audio to Video in one Studio. Duration can be set from 1 to 20 seconds. Draft Mode is the preview pass. Turning it off is the final pass, up to 1080p and 48 FPS. Keep the same start image and aspect ratio when you move from draft to final, so the shot you approved is the shot you export.
| Step | Setting in Studio | Goal |
|---|---|---|
| 1. Input | Image to Video (or Audio to Video) | Fix subject and look |
| 2. Seed frame | Upload a still; optional last frame | Consistency, planned ending |
| 3. Draft | Draft Mode on, 720p, about 5s | Check timing and motion |
| 4. Iterate | One change per draft | Isolate what works |
| 5. Final | Draft off, up to 1080p / 48 FPS | Cleaner deliverable |
| 6. Variants | Aspect ratio 9:16 / 1:1 / 16:9 | Placements |
Step 1 — Choose your input mode
Studio offers three modes: Text to Video, Image to Video, and Audio to Video. Text to Video is the start when you only have a description. Image to Video is the usual path for product motion and character clips, because one still fixes the subject and the opening frame before any motion is generated. Audio to Video is the path for a talking avatar or a music-led edit, where the file should set the timing.
Choose the mode before you stretch the prompt. An image keeps identity and framing close to that still. Audio keeps rhythm close to the upload. Text leaves both of those open.
Step 2 — Prepare a seed frame
Upload one still as the starting frame. Use a clear subject, a simple background, and the crop you want in the first second. If the clip needs a planned ending, add a last-frame reference. Studio can use that last frame for a landing shot, such as a product finishing in a set pose or a face landing on a set expression.
Match the still to the aspect ratio you will render. A vertical image in a widescreen frame asks the model to invent the side edges.
Step 3 — Draft pass
Turn Draft Mode on. Keep resolution at 720p and duration short, about 5 seconds. This page describes a typical 5-second 720p clip as rendering in about 10 seconds. Use the draft to judge timing and motion, not as the file you publish.
Draft previews cost much less than a full render. The Studio shows the credit cost for the settings you have selected.
Step 4 — Iterate
Change one thing per draft: prompt wording, camera direction, or pacing. If you change the still, the prompt, and the duration together, you cannot tell which change fixed the shot.
Watch the first second, the middle, and the last second. Faces, hands, and product edges are the usual places a short draft shows a problem.
Step 5 — Final render
Turn Draft Mode off once pacing, camera direction, and framing are settled. Raise resolution up to 1080p. When the placement needs smoother motion, use up to 48 FPS. Keep the same start image, aspect ratio, and core prompt so the final clip follows the draft you approved.
Leave on the option to keep generated or conditioned audio when the export should include that sound.
Step 6 — Variants
From the same direction, change the aspect ratio for each placement. Use 9:16 for vertical, 1:1 for square, and 16:9 for widescreen. Studio also offers 4:3, 3:4, 3:2, and 2:3.
Check the first and last frames after a ratio change. A tighter crop can cut off the subject the seed frame was holding. If the crop changes a lot, run that variant as a draft first.
Audio-led workflow
For a talking avatar or a music clip, switch to Audio to Video. Upload a clear speech or music file, and add an optional start frame when the face or product should stay close to a chosen still.
Keep generated or conditioned audio in the export if that track should ship with the video. Draft the timing at 720p, then turn Draft Mode off for the final render at the same framing.
Credits: how cost changes
Cost depends on resolution, duration, and draft mode. Drafts cost much less. The Studio shows the credit cost for your current settings before you render (160 credits at the default settings when we checked). Confirm the figure on the page before you run.
New accounts receive 4 trial credits, and those credits work only on some models, so do not treat P-Video as a free generator. Credit packs are listed on the pricing page.
When to use another model
Stay with P-Video when a still or an audio file should lead a short clip and you want a draft before the final render. Use another model when the job is different.
Use Seedance 2 Fast when the motion should follow a reference video or a music track. Use Grok Imagine Video when you want text, image, or source-video guidance with native audio. Use LTX-2 on the AI Video Generator for a simple text-to-video or image-to-video pass on the default video page.
FAQ
How long can a P-Video clip be?
Studio duration runs from 1 to 20 seconds.
What resolution and frame rate can I use?
Draft the shot at 720p. The final render can go up to 1080p and 48 FPS.
How many credits does a clip cost?
The Studio shows the cost for your settings. Draft previews cost much less than a full render. Confirm the number before you run. The pricing page lists the Studio default.
When should I turn draft mode off?
Turn it off once pacing, camera direction, and framing are settled and you want the cleaner export.
Can I keep the audio?
Yes. Studio includes an option to keep generated or conditioned audio in the export.
What setup works for a talking avatar?
Use Audio to Video with a clear audio file, and add an optional start frame when the face should match a chosen still.
为什么 P-Video 适合高频试片的视频工作流
P-Video 并不是那种偏慢、偏实验性的电影感视频模型。它更适合需要快速试想法、稳定近景主体、并在草稿与终稿之间高频切换的生产环境。
一个模型页覆盖三种工作流
在同一个页面内处理文生视频、图生视频和音频驱动视频。
草稿优先的创作逻辑
先快速摸清节奏和构图,再决定是否进入更高质量输出。
音频直接参与镜头生成
上传对白或音乐后,镜头节奏不必完全依赖纯文本描述。
同一模型支持多种投放比例
横版、竖版、方版都能覆盖,更适合内容分发与投放复用。
P-Video 在生产里的强项
它最适合短时长、主体明确、需要高频重渲染的镜头任务,而不是复杂多场景的电影式叙事。
常见问题
如果你在评估 P-Video 作为生产型视频模型,下面这些问题最值得先看。
用 P-Video 开始做下一条短视频
如果你需要更快的创意循环、更稳定的近景动态,以及从预览到终稿更顺手的过渡,P-Video 会是更实用的一类模型页。
