AdsTurbo AI 图片创作 · AdsTurbo AI Image Creation
文生图与图片编辑,外加电商场景全套:抠图、商品主图/场景图/细节图、活动海报、去水印消除、高清放大。
AdsTurbo AI Image Creation
Text-to-image, image editing, and full image sets for e-commerce and marketing scenarios.
When to use this skill
Use it when the user says "generate an image," "AI drawing," "text-to-image," "edit image," "cutout," "remove background," "product hero image," "e-commerce image," "promo poster," "remove watermark from image," or "upscale image."
For video generation, use adsturbo-video-generation.
Prerequisites
- Python 3.8+ (the scripts use the standard library only; nothing to install)
- An AdsTurbo account. No account yet is fine: the user can sign up or log in right in the chat with their email and a verification code, and the scripts get an API key automatically (see below)
- Optional
ADSTURBO_BASE_URL, defaults tohttps://adsturbo.ai/klian/novartapi
Logging in from the chat (signs up new users too)
On first use, or when a script prints "error": "api_key_required" or hits a 401, log the user in right here instead of sending them to the website:
- Ask for the user's email and run
python3 scripts/login.py start --email <email> - If it returns
"next": "enter_captcha": show the user the image atcaptcha_url(or the local filecaptcha_file), ask them to type the characters, then runpython3 scripts/login.py captcha --code <what they typed>. The user must read and type the captcha themselves; never solve it on their behalf - Once it returns
"next": "enter_email_code", ask the user for the code AdsTurbo emailed them (check spam too) and runpython3 scripts/login.py verify --code <code> - On success the API key is saved on this machine in
~/.config/adsturbo/credentials.jsonand every script picks it up; just re-run the command that needed it
A new email is signed up automatically; an existing user is simply logged in and no duplicate key is created. login.py status shows whether a key is available; run login.py logout before switching accounts. The old way still works too: sign up at https://adsturbo.ai?channel=clawhub, create a key at https://adsturbo.ai/iris/api and export ADSTURBO_API_KEY=<key>; the environment variable takes precedence.
Every plan can call the API with an API key, and usage is billed in credits; if you run out, top up or upgrade at https://adsturbo.ai/pricing. 1080p+ video and 4K images need a subscription (credit packs alone don't count), and creating a custom digital human needs a Pro plan.
If the user is not logged in yet: do not keep retrying. First finish everything that doesn't need the API (clarify the brief, draft the script or prompt, settle aspect ratio, duration, model and other parameters, prepare the exact command), then log them in with the steps above and run the prepared command. If the captcha answer is wrong, run start again for a fresh image. Never echo the key back in a reply; script output only shows its last four characters.
Choosing a command
What image does the user want?
├─ Generate from scratch / edit an image → create (pass --image-urls to edit)
├─ Product image set (hero/scene/detail) → ecommerce
├─ Promo poster → poster
├─ Background removal / cutout → cutout
├─ Remove watermark / clean up clutter → erase
└─ Upscale to high resolution → upscale
python3 scripts/image.py create --prompt "极简风格护肤品摆拍,柔和自然光"
python3 scripts/image.py create --prompt "把背景换成大理石台面" --image-urls https://.../product.jpg
python3 scripts/image.py ecommerce --reference-image-urls https://.../product.jpg \
--methods hero_banner lifestyle_scene close_detail
python3 scripts/image.py poster --event-info "双11 全场五折,11月1日开抢"
python3 scripts/image.py cutout --image-url https://.../product.jpg --output-format png
python3 scripts/image.py erase --image-url https://.../photo.jpg --instruction "去掉右下角水印"
python3 scripts/image.py upscale --image-url https://.../small.jpg --scale 4
Key point: the model determines valid parameter values, and differs by text-to-image vs. editing
For create, the valid ratio / resolution value sets differ by model, and some models are text-to-image only, others editing only — grok-2-image cannot edit, grok-imagine-image cannot generate from text.
Before building the request, check the model capability table in references/image.md to confirm whether the user wants text-to-image or editing, then pick the model. Omitting --model falls back to the server default nanobanana-pro, which supports both text-to-image and editing — the simplest choice.
ecommerce --methods accepts multiple values at once to produce a full set of images; omit it to get only the default hero image.
Hard constraint: assets must be public URLs only
--image-url, --image-urls, --reference-image-urls, and --mask-url do not accept local file paths:
python3 scripts/upload.py image ./product.jpg
See references/upload.md for details.
Sync vs. async
create returns result_url synchronously by default — take the result directly; add --async-mode to get a workspace_id instead. Each call produces exactly one image; call it again for more.
All other commands are async. The scripts poll automatically until a result is ready by default. On timeout, keep waiting with query --workspace-id <id> — do not resubmit. See references/work.md for details.
When replying to the user
- Give the image link directly — don't mention script names, command lines, JSON, or exit codes
createusually takes 30 seconds to 1 minute; e-commerce sets and posters take 1–3 minutes, longer for multiple images- A rejected parameter is usually a ratio/resolution mismatch with the model, or a text-to-image request sent to an editing-only model — check the table, retry with a valid value, and don't dump the raw error on the user
Capability boundaries
| User wants | Use |
|---|---|
| Generate video | adsturbo-video-generation |
| Digital human voiceover | adsturbo-digital-human |
| Remove video watermark / upscale video | adsturbo-video-enhance |
| Swap character in video | adsturbo-video-transform |
| Recreate from a reference video | adsturbo-ad-clone |
AdsTurbo AI 图片创作
文生图、图片编辑,以及电商与营销场景的成套出图。
什么时候用这个 skill
用户说「生成图片」「AI 画图」「文生图」「改图」「抠图」「去背景」「商品主图」「电商图」「活动海报」「图片去水印」「图片高清放大」时用它。
要生成视频用 adsturbo-video-generation。
前置条件
- Python 3.8+(脚本只用标准库,不需要安装依赖)
- AdsTurbo 账号。没有也行:在对话里用邮箱 + 验证码就能注册或登录,脚本自动拿到 API Key,见下
- 可选
ADSTURBO_BASE_URL,默认https://adsturbo.ai/klian/novartapi
在对话里登录(没有账号就顺便注册)
用户第一次用、或脚本输出 "error": "api_key_required" / 报 401 时,就在对话里帮用户登录,不用让他跳去网页:
- 问用户的邮箱,执行
python3 scripts/login.py start --email <邮箱> - 如果返回
"next": "enter_captcha":把captcha_url这张图(或本地文件captcha_file)给用户看,请他照图输入字符,再执行python3 scripts/login.py captcha --code <用户输入>。图形验证码必须由用户本人辨认输入,不要自己识别代填 - 返回
"next": "enter_email_code"后,请用户去邮箱(也看看垃圾箱)找 AdsTurbo 发来的验证码,执行python3 scripts/login.py verify --code <验证码> - 成功后 API Key 保存在本机
~/.config/adsturbo/credentials.json,之后所有脚本自动使用,直接重跑刚才的命令
新邮箱自动注册,老用户就是登录,不会重复创建 Key。login.py status 查看是否已登录,换账号先 login.py logout。也可以照旧在 https://adsturbo.ai?channel=clawhub 注册后到 https://adsturbo.ai/iris/api 创建 Key 并 export ADSTURBO_API_KEY=<Key>,环境变量优先。
任何套餐都能用 API Key 调用,按用量扣积分;积分不够时到 https://adsturbo.ai/pricing 充值或升级。1080p 及以上视频、4K 图片需要订阅(只买积分包不算),创建自定义数字人需要 Pro 套餐。
用户还没登录时:不要反复重试。先把不需要调接口就能做的部分做完(理清需求,写好文案或提示词,定好画幅、时长、模型等参数,拼好要执行的完整命令),再按上面的步骤在对话里登录,成功后直接执行准备好的命令。图形验证码输错时按提示重新 start 换一张图。Key 不要在回复里回显,脚本输出里只有末 4 位。
命令怎么选
用户要什么图?
├─ 凭空生成 / 改一张图 → create(传 --image-urls 就是改图)
├─ 商品套图(主图/场景/细节)→ ecommerce
├─ 活动海报 → poster
├─ 去背景抠图 → cutout
├─ 去水印 / 消除杂物 → erase
└─ 放大到高清 → upscale
python3 scripts/image.py create --prompt "极简风格护肤品摆拍,柔和自然光"
python3 scripts/image.py create --prompt "把背景换成大理石台面" --image-urls https://.../product.jpg
python3 scripts/image.py ecommerce --reference-image-urls https://.../product.jpg \
--methods hero_banner lifestyle_scene close_detail
python3 scripts/image.py poster --event-info "双11 全场五折,11月1日开抢"
python3 scripts/image.py cutout --image-url https://.../product.jpg --output-format png
python3 scripts/image.py erase --image-url https://.../photo.jpg --instruction "去掉右下角水印"
python3 scripts/image.py upscale --image-url https://.../small.jpg --scale 4
关键:模型决定参数取值,且分文生/编辑
create 的 ratio / resolution 取值集每个模型不同,而且有的模型只能文生、有的只能编辑——grok-2-image 不能编辑,grok-imagine-image 不能文生。
构造请求前先查 references/image.md 里的模型能力表,确认用户要的是文生还是编辑,再挑模型。不传 --model 走服务端默认 nanobanana-pro,文生编辑都支持,最省事。
ecommerce --methods 一次可传多个,出一整套图;不传只出默认主图。
硬约束:素材只收公网 URL
--image-url --image-urls --reference-image-urls --mask-url 都不接受本地文件路径:
python3 scripts/upload.py image ./product.jpg
同步还是异步
create 默认同步返回 result_url,直接拿结果;加 --async-mode 才返回 workspace_id。每次固定出 1 张,要多张就多次调用。
其余命令都是异步,脚本默认自动轮询到出结果。超时用 query --workspace-id <id> 接着等,不要重新提交。详见 references/work.md。
回复用户时
- 直接给图片链接,不要提脚本名、命令行、JSON 或退出码
create通常 30 秒 – 1 分钟;电商套图和海报 1–3 分钟,出多张更久- 参数被拒多半是模型与 ratio/resolution 不匹配,或用只能编辑的模型做了文生——查表换一个合法值重试,别把原始报错甩给用户
能力边界
| 用户想要 | 该用 |
|---|---|
| 生成视频 | adsturbo-video-generation |
| 数字人口播 | adsturbo-digital-human |
| 视频去水印 / 视频高清放大 | adsturbo-video-enhance |
| 视频换角色 | adsturbo-video-transform |
| 照着参考视频复刻 | adsturbo-ad-clone |
- Downloads
- 928All time
- Security audit
- Review
- Last updated
- 4d ago
- Current version
- v1.4.0
- License
- MIT-0
