Models & Inputs
Upscales
Image and video upscale models and their inputs.
image-upscale
POST /api/v1/processes with "type": "image-upscale". Pick any model — the API does not plan-gate models. GET /api/v1/models returns the same capabilities at runtime.
Input fields
| Field | Type | Required | Description |
|---|---|---|---|
disableAutoRetries | boolean | no | Disable automatic same-model retries. Defaults to false, so automatic retries are enabled. |
upscaleFactor | number | no | Upscale multiplier, e.g. 2. |
imageAssetDocId | string | no | Asset id of the image to process (from /assets). Required unless imageUrl is provided. |
imageUrl | string | no | Public image URL — imported into your asset library automatically, then upscaled. Alternative to imageAssetDocId. |
modelOptions | object | no | Settings this model's provider accepts natively, under the provider's own names. Omitted options keep their default. |
outputName | string | no | — |
promptItemCount | number | no | — |
promptItemId | string | no | — |
promptItemIndex | number | no | — |
promptItemTitle | string | no | — |
Example
Add "dryRun": true to get the exact credit cost without starting, and always send an Idempotency-Key header on real starts.
POST /api/v1/processes
{
"type": "image-upscale",
"model": "topaz-image-upscale",
"input": {
"imageUrl": "https://example.com/photo.jpg"
}
}Models
Cost depends on the input — use dryRun: true for an exact quote.
| Model | Name | Description | modelOptions |
|---|---|---|---|
topaz-image-upscale | Topaz Standard V2 | Fast, faithful enlargement for already-decent photos. Preserves the source with less reconstruction of damaged detail. | — |
topaz-wonder-3.5-image-upscale | Topaz Wonder 3.5 | Best overall quality for low-resolution or compressed photos. Rebuilds realistic detail; slower and may reinterpret tiny details. | enhancement_strength: low, medium, high (default high) |
topaz-bloom-2-image-upscale | Topaz Bloom 2 | Best for AI-generated images. Adds realistic skin, hair, fabric, and texture; most creative and most expensive. | creativity: 1 to 9 (default 3) |
topaz-recover-3-image-upscale | Topaz Recover 3 | Best for old, tiny, blurry, or damaged photos—especially faces. Reconstructs missing detail; unnecessary for clean sources. | enhancement_strength: 0 to 10 (default 5) |
video-upscale
POST /api/v1/processes with "type": "video-upscale". Pick any model — the API does not plan-gate models. GET /api/v1/models returns the same capabilities at runtime.
Input fields
| Field | Type | Required | Description |
|---|---|---|---|
prompt | string | no | What to generate. |
disableAutoRetries | boolean | no | Disable automatic same-model retries. Defaults to false, so automatic retries are enabled. |
upscaleFactor | number | no | Upscale multiplier, e.g. 2. |
videoAssetId | string | no | Asset id of the video to process (from /assets). Required unless videoUrl is provided. |
videoUrl | string | no | Public video URL — imported into your asset library automatically, then upscaled. Alternative to videoAssetId. |
modelOptions | object | no | Settings this model's provider accepts natively, under the provider's own names. Omitted options keep their default. |
outputName | string | no | — |
promptItemCount | number | no | — |
promptItemId | string | no | — |
promptItemIndex | number | no | — |
promptItemTitle | string | no | — |
Example
Add "dryRun": true to get the exact credit cost without starting, and always send an Idempotency-Key header on real starts.
POST /api/v1/processes
{
"type": "video-upscale",
"model": "flow-upscaling",
"input": {
"videoUrl": "https://example.com/clip.mp4"
}
}Models
Cost depends on the input — use dryRun: true for an exact quote.
| Model | Name | Description | modelOptions |
|---|---|---|---|
flow-upscaling | HD Upscaling | Upscale your video to 1080p or 4k. | — |
mmaudio-2 | AI Audio | Add AI-generated audio to your video | — |
topaz-video-upscale | Proteus Natural | Best for a clean, affordable 2x. Preserves the source while reducing compression, halos, and jagged edges. | — |
topaz-rhea-upscale | Rhea 4x | Best-value 4x for stable footage with solid structure. Recovers fine detail in AI video, interviews, products, and textures. | — |
topaz-starlight-precise-2.6 | Starlight 2.6 | Best quality for AI video. Improves soft or plastic-looking faces, skin, textures, and flicker; slowest and most expensive. | sharpness: 1 to 5 (default 2) |
flux-video-upscale | FLUX Creative | Creative FLUX 3 super-resolution for generated footage. Rebuilds fine detail aggressively for the sharpest result; faces and products may drift. | — |
flux-video-upscale-precise | FLUX Precise | Source-faithful FLUX 3 super-resolution for faces, products, text, and brand assets. Sharpens and removes noise without creatively changing details. | — |