Guides

Models and parameters

Each model says exactly what it accepts: aspect ratios, durations, frames, value ranges and its own custom options. Read it once, and your requests are valid by construction.

Suites

Models are grouped by what they do. A model can belong to more than one suite.

Suite	In Rogue Studio	Used by
`image`	Conjure	POST /v1/images/generations
`edit`	Transform and inpaint	POST /v1/images/edits
`video`	Motion	POST /v1/videos/generations
`extend`	Extend video	POST /v1/videos/extend

Listing models

GET/v1/modelsModels your plan can use. Filter with suite, kind (system, custom, character) and search.
GET/v1/models/{id}One model. URL-encode ids that contain a slash, such as character models.
GET/v1/stylesStyle ids for the styles field.
GET/v1/suggestionsPrompt suggestions for a suite (image, edit or video).

curl "https://rogue-api.sentryq-va.com/v1/models?suite=image" \
  -H "Authorization: Bearer $ROGUE_API_KEY"

The list reflects your Rogue plan and is refreshed every few minutes. Leave out model in a create request to use the model whose default_for includes that suite.

The model object

An image model

{
  "id": "xxxv2alt-sd-50prostandard",
  "object": "model",
  "name": "Rogue Image",
  "kind": "system",
  "suites": [
    "image",
    "edit"
  ],
  "default_for": [
    "image"
  ],
  "inpaint": true,
  "nsfw": true,
  "prompt_required": true,
  "tiers": [
    "pro"
  ],
  "batch_size": {
    "Conjure": 2,
    "Transform": 4
  },
  "parameters": {
    "aspect_ratios": [
      {
        "ratio": "1:1",
        "width": 1024,
        "height": 1024
      },
      {
        "ratio": "16:9",
        "width": 1344,
        "height": 768
      }
    ],
    "video_aspect_ratios": [],
    "durations": [],
    "cfg": {
      "min": 1,
      "max": 10,
      "step": 0.5,
      "default": 4
    },
    "weight": null,
    "clip_skip": null,
    "mask_blur": {
      "min": 0,
      "max": 64,
      "step": 1,
      "default": 8
    },
    "sampler": null,
    "max_reference_images": 1,
    "frames": {
      "start": false,
      "start_required": false,
      "end": false,
      "end_required": false
    },
    "audio": null,
    "source_video": null,
    "options": [
      {
        "key": "resolution_sd50",
        "name": "Resolution",
        "type": "select",
        "values": [
          {
            "value": "1k",
            "label": "1K"
          },
          {
            "value": "2k",
            "label": "2K"
          }
        ],
        "default": "1k"
      }
    ]
  }
}

idstring: Pass it as model.
kindstring: system (platform models), custom (your trained models) or character.
suites, default_forstring[]: Where the model can run, and where it is the default.
prompt_requiredboolean: Whether a prompt is mandatory (a suggestion can stand in for it).
batch_sizeobject: Outputs per run in each Rogue suite: one request can return several files.
parameters.aspect_ratiosarray: Image ratios with their pixel size. Send the ratio string as aspect_ratio.
parameters.video_aspect_ratios, durationsarray: Video ratios and durations. Send a duration value such as "5".
parameters.cfg, weight, clip_skip, mask_blurrange | null: Allowed min, max, step and default. null means the model has no such setting.
parameters.max_reference_imagesinteger: How many reference_images the model takes.
parameters.framesobject: Whether start_image and end_image are accepted or required.
parameters.audio, source_videoobject | null: Limits for reference_audios and source_video, when supported.
parameters.optionsarray: Custom options, below.

Values outside these lists and ranges answer 400 invalid_request before anything is sent to Rogue.

Custom options

Some models expose their own settings in parameters.options: a select with fixed values, a toggle, or a slider with min, max and step. Send them in options, keyed by key:

Request body

{
  "model": "xxxv2alt-sd-50prostandard",
  "prompt": "studio portrait, soft light",
  "aspect_ratio": "1:1",
  "options": {
    "resolution_sd50": "2k"
  }
}

Unknown keys are rejected; options you leave out take the model's default.

Character models

Characters trained in Rogue Studio appear with kind: "character" and ids such as character/aria. They render on top of a base model: base_models lists the choices per suite and the default, and base_model in a create request picks one.

Character ids contain a slash, so encode it when you fetch one: GET /v1/models/character%2Faria.

Styles

GET /v1/styles lists preset looks, yours included. Add up to three ids to styles. Image styles work on image and edit requests, video styles on video requests.

Response

{
  "object": "list",
  "data": [
    {
      "id": "cinematic",
      "object": "style",
      "name": "Cinematic",
      "type": "image",
      "custom": false
    }
  ],
  "page": 1,
  "limit": 1,
  "has_more": false
}

Suggestions

Suggestions are the prompt chips from Rogue Studio (outfits, locations, lighting and more), grouped by category. GET /v1/suggestions?suite=image lists them; send up to 10 ids in suggestions, at most one per category.

Response

{
  "object": "list",
  "data": [
    {
      "id": "17",
      "object": "suggestion",
      "name": "Golden hour",
      "category": "Lighting",
      "category_id": "3",
      "suites": [
        "image"
      ],
      "media_seconds": 0
    }
  ],
  "page": 1,
  "limit": 1,
  "has_more": false
}

Suggestions and cost

Some video suggestions add seconds to a clip (media_seconds) and change its price. POST /v1/estimate includes them.