D-IDD-ID
D-ID 是「照片开口说话」的数字人台:Talking Photo 招牌 + 实时交互 Agents 线 + API-first 集成面; 轻量照片驱动向,不是高保真克隆向。
D-ID makes photos talk: Talking Photo is the signature + real-time interactive Agents line + an API-first integration surface; lightweight photo-driven, not high-fidelity cloning.
适合与不适合When to use it, when not to
适合:照片/插画角色口播批量、API 集成数字人、实时客服试水。 不适合:长叙事、高保真真人克隆、影视级画面。
Fits: photo-to-talking-head batches (historical figures/illustrated characters), API-integrated digital humans, live interactive CS trials. Not a fit: long-form narrative, high-fidelity real-person cloning, film-grade visuals.
中文生态China availability 国产 / 国内可直接用 / 备案状态 · 快照domestic / direct access / CAC filing · snapshot 2026-10-22
- 出品属地Origin
- 国外出品foreign-made D-ID(以色列特拉维夫)。D-ID (Tel Aviv, Israel).
- 国内使用Direct access
- 大陆可直连direct OK d-id.com 大陆直连实测 200(2026-10-22)。d-id.com measured 200 direct from mainland (2026-10-22).
- 备案状态CAC filing
- 未见于备案名单not in filing list:未见于网信办公示名单(2026-10-22 检索);该产品未面向大陆公众提供服务,备案不适用。not in the CAC public registry (searched 2026-10-22); the product does not serve the mainland public, so filing does not apply.
备案信息以网信办《生成式人工智能服务已备案信息》公示名单为准(按批次滚动更新,本站只抄编号不解读);「大陆可直连」是本站从大陆网络直连实测的结果,不是厂商承诺。GitHub 类站点当日实测有波动,口径写在依据里。Filing status is quoted from the CAC public registry (updated batch by batch; this site copies the number and does not interpret it). "Direct access" is this site's own reachability measurement from a mainland network, not a vendor promise. GitHub-hosted sites fluctuated on the measurement day — the note says so.
固定坐标系Fixed coordinate system
8 个维度dimensions, 与同分区其他对象逐项可比each comparable item by item against others in the same section.
- 模型与开放条件Model & openness
照片驱动数字人线(Talking Photo:单张照片+音频→口播视频,官方招牌叙事)+ 实时交互数字人线 (Agents,官网导航所见);底层模型未逐条核验到官方声明。
Photo-driven digital-human line (Talking Photo: one photo + audio → talking-head video, the official signature narrative) + real-time interactive digital-human line (Agents, as seen in site navigation); the underlying models — no official statement verified item by item.
- 运行位置Where it runs
网页(d-id.com)+ API(官网有 API 入口,开发者向是传统强项);生成在厂商云端。
Web (d-id.com) + API (an API entry on the official site; developer-facing is its traditional strength); generation on the vendor's cloud.
- 本地文件Local files
作品存账户侧云端,生成历史在网页作品页可见;下载归本地。 未见官方声明的本机数据面。
Works stored account-side in the cloud, generation history visible on the web works page; downloads go local. No officially stated local data surface.
- 关机后的任务Tasks after shutdown
生成即时/排队(按档位波动,未逐条核验);实时 Agents 线是流式交互(官网叙事); 无「关机续跑」任务面(未核验到声明)。
Generation is instant/queued (varies by tier, not verified item by item); the real-time Agents line is streaming interaction (official narrative); no "keep-running-after-shutdown" task surface (no claim verified).
- 工具与扩展Tools & extensions
Talking Photo(照片口播)、实时交互数字人(Agents 线:客服/导购场景叙事)、 视频翻译线(官网所见);API-first 的集成面是它的差异化产品姿态。
Talking Photo (photo talking-head), real-time interactive digital humans (Agents line: customer-service/sales-guide scenario narratives), a video-translation line (seen on the official site); the API-first integration surface is its differentiating product posture.
- 上下文与记忆Context & memory
Agents 线有会话状态(交互数字人的对话记忆面存在,细节未逐条核验);创作面以项目组织。
The Agents line has session state (an interactive digital human's conversation-memory surface exists; details not verified item by item); the creation side is organized by projects.
- 权限与限制Permissions & limits
账号+订阅+企业体系(D9:价格不采);真人照片驱动的授权验证是关键合规面 (官方 consent 叙事,细节未逐条核验)。
Account + subscription + enterprise system (D9: prices not collected); consent verification for real-photo-driven generation is the key compliance surface (official consent narrative, details not verified line by line).
- 适合什么任务Best for
适合:照片→口播的轻量批量(历史人物/插画角色开口说话)、API 集成数字人、实时交互客服试水。 不适合:长视频叙事、高保真真人克隆(HeyGen 向)、要影视级画面生成。
Fits: light batch photo→talking-head (historical figures/illustrated characters speaking), API-integrated digital humans, testing real-time interactive customer service. Does not fit: long-form video narrative, high-fidelity real-person cloning (HeyGen territory), cinematic-grade visual generation.
头号误解Top misconception
- 「一张照片就能开口」是招牌也是风险面——真人照片的授权验证在条款上严格,别按演示理解边界'Any photo can talk' is both the signature and the risk surface — consent verification for real photos is strict under the terms
- API-first 是产品姿态——它的开发者集成面比同类重,按「网页工具」理解会低估API-first is the product posture — its developer surface is heavier than peers; reading it as a 'web tool' underestimates it
- 数字人视频与生成式视频是两个物种——D-ID 造「照片说话」,不造世界Digital-human video and generative video are two species — D-ID makes photos talk, it doesn't build worlds
未知项清单Known unknowns
- 底层生成模型(未核验到官方声明)
- Agents 线的会话记忆持久化(未逐条核验)
- 训练用途开关(未逐条核验)
- 大陆可达性之外的支付/账号面(未核验)
- The underlying generation model (no official statement verified)
- Session-memory persistence on the Agents line (not verified item by item)
- The training-use switch (not verified line by line)
- Payment/account surfaces beyond mainland reachability (unverified)
证据来源Evidence sources
判断可回到以下一手源复核。本站核验日Every judgement can be rechecked against the primary sources below. Verified on 2026-10-22,内容更新日content updated on 2026-10-22。
| 类型Type | 名称Name | 链接Link |
|---|---|---|
| official | D-ID · 官网(产品与能力叙事)D-ID · official site (product and capability narrative) | https://www.d-id.com |
| official | D-ID · Terms of Service(官方条款,2026-10-22 直抓核验)D-ID · Terms of Service (official terms, direct-fetched and verified 2026-10-22) | https://www.d-id.com/terms-of-service/ |
| changelog | D-ID · Blog(官方更新线,changelog)D-ID · Blog (official update line, changelog) | https://www.d-id.com/blog/ |
实测记录Measured runs
本档未跑生成质量与速度测试;结论只覆盖能力面与许可面。
实测建议(三条,都指向官方可读页没给的东西):
- Talking Photo 对非正面照/风格化插画的效果衰减
- 实时 Agents 的延迟与打断行为
- API 渲染队列与网页档的能力差异
This site has not completed measured runs. The coding-oriented test protocol is at protocols/coding/_protocol.md (the coding task suite does not apply to creative products).
Suggested measurements (three, all pointing at things the official readable pages do not give):
- Talking Photo's quality falloff on non-frontal photos / stylized illustrations
- The real-time Agents line's latency and interruption behavior
- The capability gap between the API rendering queue and the web tiers
相关条目Related entries
- HeyGenHeyGen同为数字人:HeyGen 克隆/翻译线重,D-ID 照片驱动/API 向轻。Both digital humans: HeyGen leans on cloning/translation lines, D-ID is lighter on photo-driven/API orientation.
- SynthesiaSynthesia同为企业向:Synthesia 培训工作区重,D-ID 集成面重。Both enterprise-oriented: Synthesia leans on the training workspace, D-ID on the integration surface.
- RunwayRunway视频生成的另一物种:Runway 造画面,D-ID 让照片说话。Another species of video generation: Runway builds visuals, D-ID makes photos talk.
- ElevenLabsElevenLabs口播轨上游:Talking Photo 的音频常由 ElevenLabs 类 TTS 供给,两条线常搭着用。Upstream of the talking track: Talking Photo's audio is often supplied by ElevenLabs-style TTS; the two lines are commonly paired.
本页由This page is generated from the ai-handbook 数据层生成(CC BY 4.0)。data layer (CC BY 4.0).
方法论与坐标系定义见仓库内For the methodology and coordinate definitions, see METHODOLOGY.md.