LlamaGen.Ai LogoLlamaGen.Ai Brand
Vyzkoušejte zdarmaSpustit aplikaciVyzkoušejte zdarma

Comic API Documentation

Integrate professional comic generation capabilities into your applications.

How the Comic API works

Three quick onboarding illustrations: prepare the input, send the request, review the result.

Step 01
Prompt + token
Colorful hand-drawn comic API onboarding illustration showing a prompt sheet and API token note.
Step 02
Send request
Colorful hand-drawn comic API onboarding illustration showing a request moving into the API workspace.
Step 03
Review panels
Colorful hand-drawn comic API onboarding illustration showing finished panels being reviewed before export.
APIEndpointsCreateStatusContinueUpdateMCPSDKTypesWebhooksLogsRate LimitsErrors

REST + MCP

Komiksové API

Use the Comic API to turn prompts, customer uploads, character references, and story briefs into structured comic pages with generated panel assets.

Autentizace

Send your API key with HTTP Bearer Auth on every request.

1Authorization: Bearer $LLAMAGEN_API_KEY

Base URL

All production REST endpoints use the same API host. MCP clients can connect from the docs workflow.

1https://api.llamagen.ai

Endpoints

Create generation

POST /v1/comics/generations

Submit prompt, role references, scene references, layout options, language, upscale, and attachments.

Get status

GET /v1/comics/generations/{generationId}

Poll status, generated assets, panel metadata, and failure details for a generation.

Continue write

PATCH /v1/comics/generations/{generationId}

Extend an existing story or multi-page generation while preserving established characters and settings.

Update panel

PATCH /v1/comics/generations/{generationId}

Regenerate one panel with revised direction while keeping the surrounding comic structure intact.

Create request

Provide prompt text, then layer in optional style, size, character, scene, attachment, language, and upscale fields.

1curl -X POST https://api.llamagen.ai/v1/comics/generations \2  -H "Authorization: Bearer $LLAMAGEN_API_KEY" \3  -H "Content-Type: application/json" \4  -d '{5    "prompt": "A 4-panel comic about Leo finding a glowing key in a quiet library.",6    "size": "1024x1024",7    "fixPanelNum": 48  }'
1{2  "id": "cmqyt8ea60003lb04uvxug8i9",3  "status": "LOADING",4  "prompt": "A 4-panel comic about Leo finding a glowing key in a quiet library.",5  "usage": {6    "total_comics": 4,7    "amount": 6008  },9  "model": "cyani-model"10}

Input schema

promptstring

Povinné: Yes

Story, scene, or script instructions for the comic generation. Use natural language and include panel-by-panel direction when you need tighter control.

Povinné
prompt

Příklad: A 4-panel comic strip about The Little Prince meeting the Fox.

promptUrlstring URL

Povinné: No

Optional source file URL or id stored with the generation. The request still needs prompt text.

Format
uri

Příklad: https://s.llamagen.ai/workspace/storyboard.png

modelstring

Povinné: No

Optional compatibility field returned in the create response. The backend selects the actual generation model from account configuration.

Default
"cyani-model"

Příklad: cyani-model

presetstring

Povinné: No

Style preset id. Use template ids from the preset catalog to control manga, comic, cartoon, or illustration style.

Default
"neutral"

Příklad: neutral

sizestring

Povinné: No

Output canvas size in WIDTHxHEIGHT format. Common presets are listed in this field metadata.

Default
"1024x1024"
Presets
1024x1024 (1:1), 512x768 (2:3), 512x1024 (1:2), 576x1024 (9:16), 768x1024 (3:4), 1024x768 (4:3), 768x512 (3:2), 1024x576 (16:9), 1024x512 (2:1)

Příklad: 1024x1024

fixPanelNuminteger

Povinné: No

Single-page panel count from 1 to 20. Do not send this field together with pagination.

Default
4
Minimum
1
Maximum
20

Příklad: 4

paginationobject

Povinné: No

Multi-page mode. Use pagination.totalPages and pagination.panelsPerPage when you want several pages in one request. Do not use top-level totalPages or panelsPerPage.

Conflicts
fixPanelNum

Příklad: { "totalPages": 2, "panelsPerPage": 4 }

comicRolesArray<Role> | JSON string

Povinné: No

Character references used for identity consistency across panels and pages. Each role can include a name, demographics, clothing, and image URL.

Default
[]

Příklad: [{ "name": "Alice", "age": 12, "gender": "female" }]

comicLocationsArray<Location> | JSON string

Povinné: No

Scene or background references used to keep environments consistent. Each location should have a stable name and can include an image URL.

Default
[]

Příklad: [{ "name": "Dreamwood Forest", "image": "https://..." }]

attachmentsArray<Attachment> | JSON string

Povinné: No

Additional source files for product images, brand assets, rough sketches, or client references.

Default
[]

Příklad: [{ "type": "image", "url": "https://..." }]

languagestring

Povinné: No

Preferred language for generated captions, dialogue, and lettering.

Default
"auto"

Příklad: en

upscale"" | "2K" | "4K"

Povinné: No

Optional upscale target. 2K costs 2x credits and 4K costs 4x credits.

Default
""

Příklad: 2K

pagination.totalPagesinteger

Povinné: Required when pagination is used

Number of comic pages to generate.

Minimum
1
Maximum
20

Příklad: 2

pagination.panelsPerPageinteger

Povinné: Required when pagination is used

Panels per generated page.

Minimum
1
Maximum
20

Příklad: 4

comicRoles[].namestring

Povinné: Doporučené

Stable character name used by the model and returned panel metadata.

Příklad: Alice

comicRoles[].imagestring URL

Povinné: No

Reference image for character identity, costume, and visual continuity.

Příklad: https://cdn.example.com/alice.png

comicLocations[].namestring

Povinné: Doporučené

Reusable location name for visual continuity across pages and panels.

Příklad: Dreamwood Forest

attachments[].typestring

Povinné: No

Attachment role. When omitted, the API stores it as image.

Default
"image"

Příklad: product

attachments[].urlstring URL

Povinné: Required when attachments are used

Publicly accessible or signed file URL.

Příklad: https://cdn.example.com/reference.png

Output schema

idstring

Generation id. Store this value for polling, logs, and support.

Příklad: cmqyt8ea60003lb04uvxug8i9

statusstring enum

Queued status returned after the create request is accepted.

Příklad: LOADING

promptstring

Normalized prompt used by the generation.

usage.total_comicsinteger

Total number of comic panel outputs charged for the create request.

Příklad: 4

usage.amountinteger

Credits charged by the create request.

Příklad: 600

modelstring

Compatibility model field returned by the create response.

Příklad: cyani-model

Get status

Poll generation state until the job is processed, failed, or cancelled.

1curl https://api.llamagen.ai/v1/comics/generations/gen_123456789 \2  -H "Authorization: Bearer $LLAMAGEN_API_KEY"

Input schema

generationIdstring path parameter

Povinné: Yes

Generation id returned by create generation.

Příklad: gen_123456789

pageinteger query parameter

Povinné: No

Zero-based page index. Only used when requesting a single panel together with panel.

Default
0

Příklad: 0

panelinteger query parameter

Povinné: No

Zero-based panel index. When omitted, the endpoint returns generation status and comics[].

Příklad: 2

Output schema

idstring

Generation id.

Příklad: gen_123456789

statusstring enum

Current generation status.

Příklad: PROCESSED

promptstring

Original generation prompt.

comics[]array

Povinné: When not requesting a single panel

Normalized comic pages returned by the status endpoint, including panels with assetUrl, panel, caption, and image.

Continue write

Extend an existing story or multi-page generation while preserving established characters and settings.

1curl -X PATCH https://api.llamagen.ai/v1/comics/generations/gen_123456789 \2  -H "Authorization: Bearer $LLAMAGEN_API_KEY" \3  -H "Content-Type: application/json" \4  -d '{5    "action": "continueWrite",6    "prompt": "The two friends walk deeper into the neon city and discover a hidden arcade.",7    "pagination": {8      "totalPages": 2,9      "panelsPerPage": 410    },11    "attachments": [12      {13        "type": "image",14        "url": "https://example.com/reference-1.png"15      }16    ]17  }'

Input schema

generationIdstring path parameter

Povinné: Yes

Existing comic generation id to extend.

Příklad: gen_123456789

actionstring

Povinné: Yes

Set to continueWrite so PATCH routes to the continue-write operation.

Value
"continueWrite"

Příklad: continueWrite

promptstring

Povinné: Yes

Story direction for the next page or pages.

Příklad: The two friends discover a hidden arcade.

paginationobject

Povinné: No

Use pagination.totalPages and pagination.panelsPerPage to append multiple pages. Do not send top-level totalPages or panelsPerPage.

Conflicts
fixPanelNum

Příklad: { "totalPages": 2, "panelsPerPage": 4 }

fixPanelNuminteger

Povinné: No

Panel count for one appended page when pagination is not used. Defaults to the previous page panel count.

Minimum
1
Maximum
20

Příklad: 4

attachmentsArray<Attachment> | JSON string

Povinné: No

Optional references attached to the first appended page. Fields such as size, promptUrl, comicRoles, comicLocations, language, and upscale are not supported by continue-write.

Příklad: [{ "type": "image", "url": "https://example.com/reference-1.png" }]

Output schema

idstring

Generation id being extended.

Příklad: gen_123456789

statusstring enum

Queued status after the continue-write request is accepted.

Příklad: LOADING

pageinteger

Zero-based index of the first appended page.

Příklad: 2

usage.total_comicsinteger

Number of new panel outputs charged.

Příklad: 8

usage.amountinteger

Credits charged by the request.

Příklad: 1200

paginationobject

Povinné: When pagination is provided

Returned pagination.totalPages and pagination.panelsPerPage.

Update a panel

Regenerate a single panel after review without rebuilding the full comic.

1curl -X PATCH https://api.llamagen.ai/v1/comics/generations/gen_123456789 \2  -H "Authorization: Bearer $LLAMAGEN_API_KEY" \3  -H "Content-Type: application/json" \4  -d '{5    "page": 0,6    "panel": 2,7    "panelPrompt": "Make Leo look more hopeful and keep the same orange wizard robe."8  }'

Input schema

generationIdstring path parameter

Povinné: Yes

Existing comic generation id containing the panel.

Příklad: gen_123456789

pageinteger

Povinné: No

Zero-based page index. pageIndex and page_index are accepted aliases.

Default
0

Příklad: 0

panelinteger

Povinné: Yes

Zero-based panel index. panelIndex and panel_index are accepted aliases.

Příklad: 2

panelPromptstring

Povinné: One of panelPrompt, prompt, images, or caption

Replacement prompt for the selected panel. prompt and panel_prompt are accepted aliases.

Příklad: Make Leo look more hopeful.

imagesstring | string[]

Povinné: One of panelPrompt, prompt, images, or caption

Optional reference image URL or URLs for the selected panel. images_url is accepted as an alias.

Příklad: ["https://example.com/reference.png"]

captionstring

Povinné: One of panelPrompt, prompt, images, or caption

Replacement caption for the selected panel.

Příklad: Leo whispers, I found it.

Output schema

idstring

Generation id containing the regenerated panel.

Příklad: gen_123456789

statusstring enum

Queued status after the panel regeneration is accepted.

Příklad: LOADING

panel.pageinteger

Zero-based page index for the regenerated panel.

Příklad: 0

panel.panelinteger

Zero-based panel index for the regenerated panel.

Příklad: 2

usage.redraw_panelsinteger

Number of panels queued for redraw.

Příklad: 1

usage.amountinteger

Credits charged by the request.

Příklad: 150

MCP Usage

Connect via Streamable HTTP

MCP Endpoint
https://llamagen.ai/api/mcp
Authorization Header
Authorization: Bearer YOUR_API_TOKEN

Configure your MCP client to use Streamable HTTP transport with the endpoint above. Provide your API token via the Authorization header.

Available Tools

  • create_comic_generation— Create a generation job
  • get_comic_generation_status— Get status/result by id
  • get_api_usage— View current usage and quota

Client Configuration Example

Many MCP clients allow setting a remote HTTP endpoint with custom headers. Below is a generic example:

{
  "mcpServers": {
    "llamagen": {
      "url": "https://llamagen.ai/api/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_TOKEN"
      }
    }
  }
}

The exact configuration format varies by client. Ensure the Authorization header is set with your token.

Cursor Setup

  1. Open Cursor Settings and find the Model Context Protocol (MCP) section.
  2. Add a new server using Streamable HTTP.
  3. Set URL to https://llamagen.ai/api/mcp.
  4. Under Headers, add Authorization: Bearer YOUR_API_TOKEN.
  5. Save and test by listing tools; you should see create_comic_generation, get_comic_generation_status, get_api_usage.

Alternatively, you can configure via Cursor's MCP configuration file using the same JSON structure as above, if your version supports it.

Integration Demos

LangChain JS Agent (Streamable HTTP + Azure OpenAI)

import "dotenv/config";
import { Client } from "@modelcontextprotocol/sdk/client/index.js";
import { StreamableHTTPClientTransport } from "@modelcontextprotocol/sdk/client/streamableHttp.js";
import { createAgent } from "langchain";
import { DynamicStructuredTool } from "@langchain/core/tools";
import { AzureChatOpenAI } from "@langchain/openai";

async function main() {
  const mcpUrl = "https://llamagen.ai/api/mcp";
  const YOUR_API_TOKEN = process.env.LLAMAGENAI_API_TOKEN;
  const client = new Client(
    { name: "demo-agent", version: "1.0.0" },
    { capabilities: { tools: {} } }
  );

  let connected = false;
  let discoveredTools: any | null = null;
  if (mcpUrl && YOUR_API_TOKEN) {
    const headers = {
      Authorization: "Bearer " + YOUR_API_TOKEN,
      Accept: "application/json",
    };
    const transport = new StreamableHTTPClientTransport(new URL(mcpUrl), {
      requestInit: { headers },
    });
    await client.connect(transport);
    discoveredTools = await client.listTools();
    connected = true;
  }

  let tools: any[] = [];
  if (connected && discoveredTools) {
    tools = discoveredTools.tools.map((tool: any) => {
      return new DynamicStructuredTool({
        name: tool.name,
        description: tool.description ?? "",
        schema: tool.inputSchema,
        func: async (input: Record<string, any>) => {
          const result = await client.callTool({
            name: tool.name,
            arguments: input,
          });
          return JSON.stringify(result);
        },
      });
    });
  }
  console.log("Loaded " + tools.length + " tools");

  const AZURE_OPENAI_DEPLOYMENT_NAME = process.env.AZURE_OPENAI_DEPLOYMENT_NAME;
  const AZURE_OPENAI_API_KEY = process.env.AZURE_OPENAI_API_KEY;
  const AZURE_OPENAI_API_VERSION = process.env.AZURE_OPENAI_API_VERSION;
  const AZURE_OPENAI_ENDPOINT = process.env.AZURE_OPENAI_ENDPOINT;
  const llm = new AzureChatOpenAI({
    model: AZURE_OPENAI_DEPLOYMENT_NAME,
    azureOpenAIApiKey: AZURE_OPENAI_API_KEY,
    azureOpenAIApiInstanceName: AZURE_OPENAI_DEPLOYMENT_NAME,
    azureOpenAIApiDeploymentName: AZURE_OPENAI_DEPLOYMENT_NAME,
    azureOpenAIApiVersion: AZURE_OPENAI_API_VERSION,
    azureOpenAIEndpoint: AZURE_OPENAI_ENDPOINT,
  });

  const agent = createAgent({
    model: llm,
    tools,
  });

  const result = await agent.invoke({
    messages: [
      {
        role: "user",
        content: "Query the current usage and quota for the llamagen project",
      },
    ],
  });

  console.log(result);

  const result1 = await agent.invoke({
    messages: [
      {
        role: "user",
        content: "Create a comic of a little girl running in a forest",
      },
    ],
  });

  console.log(result1);
}

main().catch((err) => {
  console.error(err);
  process.exit(1);
});

SDK Quick Start

Install the SDK first, then create a generation job and wait for the final result in two clear steps.

Install

 npm i comic

Step 1: Create generation

1import { LlamaGenClient } from 'comic';2 3const llamagen = new LlamaGenClient({4  apiKey: process.env.LLAMAGEN_API_KEY!,5});6 7const created = await llamagen.comic.create({8  prompt: 'A detective fox in Tokyo',9  size: '1024x1024'10});

Step 2: Wait for completion

1const result = await llamagen.comic.waitForCompletion(created.id);2 3console.log(result);

Supported size values: 1024x1024 (1:1), 512x768 (2:3), 512x1024 (1:2), 576x1024 (9:16), 768x1024 (3:4), 1024x768 (4:3), 768x512 (3:2), 1024x576 (16:9), 1024x512 (2:1)

TypeScript Types

The SDK ships with first-class TypeScript types for request payloads and responses.

1import type {2  CreateComicParams,3  ComicArtworkResponse,4  ComicGenerationStatus5} from 'comic';6 7const payload: CreateComicParams = {8  prompt: 'A superhero cat saving a city',9  preset: 'render',10  size: '1024x1024'11};

Valid size values: 1024x1024 (1:1), 512x768 (2:3), 512x1024 (1:2), 576x1024 (9:16), 768x1024 (3:4), 1024x768 (4:3), 768x512 (3:2), 1024x576 (16:9), 1024x512 (2:1)

Status Lifecycle

StatusMeaning
LOADINGRequest accepted or generation is still running.
PROCESSEDGeneration completed successfully.
FAILEDGeneration failed. Inspect error details.

Recommended: poll every 3-5 seconds with timeout protection and exponential retry for transient failures.

Webhooks

Manage webhook endpoints in Settings → API → Webhooks. You can subscribe to create, update, completed, and failed events for comic generations.

Supported Events

  • comic.generation.created — Fires after a generation request is accepted.
  • comic.generation.updated — Fires after continue-write or single-panel regenerate is queued.
  • comic.generation.completed — Fires when the generation reaches a processed state.
  • comic.generation.failed — Fires when the generation ends in a failed state.

Signing Headers

X-Llama-Webhook-Id: evt_...
X-Llama-Webhook-Timestamp: 1715510400
X-Llama-Webhook-Signature: v1=...
X-Llama-Webhook-Request-Id: req_...

Compute the signature from ${timestamp}.${rawBody} using your webhook secret. The dashboard only shows the secret once when creating or rotating it.

Request Logs

In dashboard settings, use Request Logs to trace endpoint usage, status codes, latency, and credit changes for each API call.

Keep a request ID in your app logs and correlate it with dashboard request logs for faster debugging.

Rate Limits

Demo Users

  • 4 requests per minute
  • 15 requests per day
  • Watermarked outputs

Paid Plans

  • 10 requests per minute
  • Usage based on credits
  • High-resolution, no watermark

Errors

CodeDescription
401Unauthorized - Invalid API token.
402Payment Required - Insufficient credits.
403Forbidden - Access denied.
429Too Many Requests - Rate limit exceeded.
500Internal Server Error.