Blog
AI Image & Video API Guides
Hands-on model tests, production tutorials, prompt techniques, and current API pricing for generative media teams.
Image Inpainting and Object Removal via API: Image-to-Image Models on hiapi
hiapi's image-to-image endpoints (flux-2, seedream-4.5, seedream-5.0-pro, gpt-image-2) do object removal and inpainting without masks — you describe the edit in the prompt and the model localizes it. All four share the same async lifecycle (POST /v1/tasks, poll or callback, download output[0].url) but have strict, mutually incompatible input schemas, each validated against the live API in this guide.
How to Use happyhorse-1.1 via the hiapi API: Text-to-Video, Image-to-Video, and Reference-to-Video
happyhorse-1.1 exposes three endpoints on hiapi's unified async /v1/tasks API: text-to-video, image-to-video, and reference-to-video — same lifecycle, different required inputs. text-to-video needs only `prompt`; image-to-video needs `image_urls` (exactly one URL); reference-to-video needs `prompt` plus 1–9 `reference_image` URLs. Shared optionals: `resolution` (720p|1080p) and `duration` (integer 3–15 s). `aspect_ratio` (9 ratios) works on text-to-video and reference-to-video but is rejected by image-to-video. Schemas are strict — `size`, `seed`, `negative_prompt`, and `audio` all return 400. Webhooks support only `callback.when="final"`. Billing is per second of output, so `duration` is the cost lever: iterate at 3 s / 720p, render the final at full length.
How to Use the grok-imagine-quality Text-to-Image API: curl, Python, and a Working Request
A working recipe for the grok-imagine-quality/text-to-image API on hiapi: create a task on POST /v1/tasks, poll or use a callback, and download the image from output[0].url before it expires. Covers the exact verified input schema (prompt, 13 aspect ratios, lowercase 1k/2k resolution tiers, output_format), plus production patterns: callbacks, Idempotency-Key, and the three failure shapes.
How to Use the hailuo 2.3 Image to Video API: curl, Python, and a Working Request
A working recipe for the hailuo-2.3/image-to-video API on hiapi: create a task on /v1/tasks with a prompt plus one public image_url, poll for success, and download the clip. Includes the verified input schema (duration is a string "6"|"10", no motion or resolution params), callback setup, and the common errors with their exact messages.
Kling 3.0 Turbo Text-to-Video Prompt Recipes: Copy-Paste Prompts With Real Outputs
Five copy-paste kling-3.0-turbo text-to-video prompt recipes, each paired with the real clip it generated on the live hiapi API — raw first outputs, hosted permanently. Covers a continuous crane-up reveal, a fast tracking shot, Chinese ink-wash style locking, a vertical 9:16 emotion close-up, and a 12-second single-take sequence. Schema verified live before generating: duration is an integer 3-15s, resolution 720p/1080p, aspect_ratio 16:9/9:16/1:1, strict schema that rejects unknown fields. Pricing is per second: $0.13/s at 720p and $0.16/s at 1080p — the 5s clips here cost $0.65 each and a 3s probe is $0.39. Every clip shipped with a native ambient audio track, and all five came back in under ten minutes submitted in parallel.
How to Use flux-2/image-to-image via the hiapi API: curl, Python, and a Working Request
A working recipe for the flux-2/image-to-image API on hiapi: create a task on /v1/tasks, poll for success, download output[0].url. All four input fields are required: prompt, image_urls (1-8 public URLs, i.e. Multi-Reference), aspect_ratio, and resolution (1K/2K). There is no strength/fidelity knob - the schema rejects unknown fields; fidelity is controlled via prompt phrasing and resolution.
How to Use grok-imagine Image-to-Image via the hiapi API: curl, Python, and a Working Request
grok-imagine/image-to-image needs just two fields — prompt and image_urls — via POST /v1/tasks; poll the task ID or register a callback for the result. Multi-Reference accepts up to 3 public image URLs; the API hard-rejects a fourth (maxItems: want 3). The schema is strict: no size, seed, or n — control output with aspect_ratio (14 enums incl. auto), resolution (1k/2k lowercase), and output_format. Output URLs are signed and expire — download bytes immediately, never hotlink. Persist the taskId on create and only retry creates that returned no ID; re-submitting a returned ID bills a second task.
How to Use grok-imagine-quality/image-to-image via the hiapi API: curl, Python, and a Working Request
grok-imagine-quality/image-to-image runs on the async /v1/tasks API: POST to create, poll or use a callback, download output[0].url before it expires. The input schema is strict: prompt + image_urls (1-3 public URLs) required; aspect_ratio, resolution (lowercase 1k/2k), output_format optional; no seed/strength/n. There is no consistency or quality parameter -- the tier is the model id, and source fidelity is controlled via the prompt and reference images.
Kling 3.0 Turbo Image-to-Video API: Schema, Limits and Python Errors
This version-specific page documents only kling-3.0-turbo/image-to-video fields, validation limits, prices, and Python failure handling. Turbo requires a prompt and exactly one public image URL; duration is an integer from 3 to 15 and resolution is 720p or 1080p. HTTP 400 means the payload failed schema validation; TASK_FAILED after a 200 create usually points to an unreachable input asset or render failure.
How to Use the kling-3.0-turbo Text-to-Video API: curl, Python, and a Working Request
Call Kuaishou's Kling 3.0 Turbo text-to-video model through hiapi's unified /v1/tasks endpoint with a working curl request, a complete Python polling client, and the production callback pattern. The live-verified input schema — prompt, Flexible Duration 3-15s, three aspect ratios, 720p/1080p — plus the exact 400s the strict schema returns when you guess field names.
Build with the models you just read about.
Compare image, video, and audio models, then use the same API workflow in production.