Blog
AI Image & Video API Guides
Hands-on model tests, production tutorials, prompt techniques, and current API pricing for generative media teams.

Seedance 2.0@ext Short-Form Video: A Practical API Workflow for Social Clips
seedance-2.0@ext generates vertical (9:16) short-form video with native audio in a single async task call The full /v1/tasks submit-and-poll workflow, with the verified input schema (prompt and aspect_ratio required) Real pricing by resolution tier, including a confirmed quirk where 4K is cheaper than 2K The actual cost of an 8-second 720p demo clip: $1.19 Why writing camera movement, timing, and audio cues directly into the prompt produces better results than a one-line description

Grok Imagine 1.5 API: Turn Images Into Short-Form Video
Grok Imagine 1.5's image-to-video endpoint turns a single still into a short, motion-consistent clip through hiapi's async /v1/tasks API — no separate video model account needed. Pricing is metered per second of output: $0.0114/s at 480p and $0.0214/s at 720p, with clips from 1 to 15 seconds. The request schema is deliberately small — just a public image URL, an optional prompt, and a duration integer — so batching multiple clips from one still is straightforward. This walkthrough uses two real clips generated from the same source photo to show how prompt wording changes camera motion and pacing.

flux-2/image-to-image for E-Commerce Product Images: A hiapi API Guide
flux-2/image-to-image edits existing product photos in place — background, lighting and scene change, product identity stays intact. aspect_ratio: "auto" tracks the input image's own shape (verified: 9:16 in, 9:16 out); an explicit value like 1:1 forces a different shape instead. Pricing is resolution-tiered: $0.035/image at 1K, $0.05/image at 2K, verified against hiapi's live pricing page. The same /v1/tasks endpoint handles single-image background swaps and multi-image lifestyle-scene composites. A plain Python loop over create-task/poll/download is enough to batch-process a full product catalog.

How to Choose an AI Image Generation API for Professional Work: A Buyer's Guide
Four dimensions actually decide this for professional work: image quality for your use case, commercial licensing, integration effort, and price at real volume — not demo quality. We ran one identical prompt through FLUX.2, Nano Banana 2, and Seedream 5.0 Lite ourselves to compare real output side by side. Only the Nano Banana family (nano-banana, nano-banana-pro, nano-banana-2) carries an explicit Commercial Use tag on hiapi's live pricing data as of August 2026 — check the tag per model before client work, don't assume. Every model runs through the same create-task/poll/download pattern, so swapping or A/B-testing models is a parameter change, not a rewrite. Real per-image prices range from $0.03 (GPT Image 2, 1K) to $0.2992 (Nano Banana Pro, 4K) — do the math against your actual monthly volume, not one test render.

Ideogram API: Typography-First Image Generation With Ideogram V4
ideogram-v4 is a text-to-image model built for accurate typography — reach for it when legible text is the deliverable (posters, logos, signage, labels), not a general-purpose model. The input schema is strict: prompt and rendering_speed (TURBO/BALANCED/QUALITY) are required; aspect_ratio and seed are optional. No negative_prompt, style_type, num_images, or image-editing input. Quoting the exact copy in the prompt and describing its letter style next to it is the main lever for correct typography — there's no dedicated text parameter. Iterate cheaply on TURBO, then re-render the winning composition on QUALITY with the same seed for the final asset. Pricing scales with rendering_speed tier; check the live pricing page for current per-image rates before budgeting a batch.

Using seedream-4.5/text-to-image for E-Commerce Product Images via the hiapi API
seedream-4.5/text-to-image generates e-commerce product and promo images from a text description alone, with no reference photo required. Two real generations shown: a studio product shot built entirely from a prompt, and two promotional graphics with correctly rendered in-image headline text. The API takes prompt, aspect_ratio (8 values), and resolution (2K or 4K only) through the standard /v1/tasks endpoint, at a flat $0.045 per image. A list of prompts loops through the same task-submit-and-poll call to batch-generate a full seasonal set of hero shots, banners, and story graphics. Positioned against seedream-4.5/image-to-image (editing a real photo) and seedream-5.0-pro (a separate, newer tier) so you can pick the right model for the job.

Using grok-imagine/image-to-image for E-Commerce Product Images via the hiapi API
grok-imagine/image-to-image composites product photos onto one reusable scene for consistent e-commerce catalogs, billed flat at $0.035/image. The live schema accepts only prompt + image_urls (max 3, not the 6 the pricing page claims) — verified directly against the task-creation API. Output aspect ratio follows whichever reference image is listed first in image_urls, giving you a simple lever to target different marketplace crops.

Grok Imagine Quality Image-to-Image for E-Commerce Product Images
grok-imagine-quality/image-to-image keeps a product's shape, color, and proportions intact while placing it in new scenes — one clean studio shot becomes an entire catalog. The model takes prompt + up to 3 image_urls + optional resolution (1k/2k) and aspect_ratio; there's no strength or seed param, so consistency comes entirely from prompt wording. At $0.09/image (1K) and $0.11/image (2K) as of 2026-08, a 4-shot listing set costs $0.30 total, including the base studio photo. Covers the full /v1/tasks lifecycle, a batch script for looping scene prompts across a whole SKU catalog, and where this model differs from the cheaper base-tier grok-imagine i2i.

Kling 3.0 Turbo Image-to-Video: Make Short-Form Video with the hiapi API
Kling 3.0 Turbo image-to-video turns a single first-frame photo into an 8-second, 1080p clip in about two and a half minutes — no rigging, no timeline editor. The hiapi /v1/tasks endpoint handles the job asynchronously: POST to create, poll for status, then download the finished .mp4 from a signed URL. Pricing is tiered by resolution: $0.13 per second at 720p, $0.16 per second at 1080p — the 8-second 1080p clip in this article cost $1.28. There's no aspect_ratio parameter — output framing always inherits the input image's composition, so compose the first frame correctly upfront. The same task call works as a batch pipeline: loop it over a folder of stills and walk away with a stack of ready-to-post short-form clips.

Using seedream-4.5/image-to-image for E-Commerce Product Images via the hiapi API
seedream-4.5/image-to-image edits a real product photo instead of generating a new one, so shape, materials, and branding stay consistent across every background or context change. Two real edits shown: a pour-over dripper moved from a plain studio shot into a warm lifestyle scene, and a sneaker unified onto a matching branded studio backdrop. The API takes prompt, image_urls (1-14 public URLs), aspect_ratio, and resolution (2K or 4K only) through the standard /v1/tasks endpoint, at $0.045 per image. The same edit instruction applied across a SKU list batch-processes an entire catalog with one consistent backdrop treatment — about $2.25 for 50 SKUs. Tasks are independent and can be parallelized, making large-catalog batch edits fast even at hundreds of SKUs.
Build with the models you just read about.
Compare image, video, and audio models, then use the same API workflow in production.