Introduction: The Ideogram Editing Matrix
Ideogram has built a product line around character consistency and flexible image editing—from pure text-to-image to masked local edits, scene remixes, and identity-locked character workflows. On FuseAI Tools, six named modes cover the full creative path; the question is which lever to pull for each job.
This guide maps character-edit, character-remix, character, v3-text-to-image, v3-edit, and v3-remix to representative API fields and the exact routes where you can run them.
Ideogram hub: /home/ideogram
I. Snapshot: Six Models at a Glance
| Model | Core function | Inputs | Mask | Character ref | Positioning |
|---|---|---|---|---|---|
| character-edit | Masked character edit | image + mask + ref + text | ✅ | ✅ (1) | Local character fix |
| character-remix | Character scene remix | image + ref + text | ❌ | ✅ (1) | New background, same face |
| character | Character in new scene | ref + text | ❌ | ✅ (1) | Scene from reference only |
| v3-text-to-image | Text-to-image | text | ❌ | ❌ | Create from zero |
| v3-edit | Inpainting / local edit | image + mask + text | ✅ | ❌ | Generic masked fill |
| v3-remix | Global remix | image + text | ❌ | ❌ | Whole-image restyle |
II. Six Models Deep Dive
Model 1: character-edit
Edit a masked region on a base portrait while locking identity to a separate reference still—ideal for expression tweaks, wardrobe details, or localized refinements.
{
"model": "ideogram-character-edit",
"prompt": "Fill the masked region (max 5000 characters)",
"imageUrl": "Base image URL",
"maskUrl": "Mask URL (white = edit, black = keep)",
"referenceImageUrls": ["Character reference (1 image)"],
"renderingSpeed": "BALANCED",
"expandPrompt": true,
"numImages": "1",
"seed": 0
}
Mechanics: upload base image, mask (same dimensions), and one character reference. FuseAI Tools accepts JPG/PNG/WEBP up to 10MB per asset.
Best for: eye/expression changes, localized outfit edits, identity-safe inpainting.
Route: /home/ideogram/character-edit
Model 2: character-remix
Keep a character recognizable while remixing scene and context from an existing frame plus prompt. strength balances fidelity to the source image vs creative freedom.
{
"model": "ideogram-character-remix",
"prompt": "Scene remix brief (max 5000 characters)",
"imageUrl": "Base image URL",
"referenceImageUrls": ["Character reference (1 image)"],
"renderingSpeed": "BALANCED",
"style": "AUTO",
"expandPrompt": true,
"imageSize": "square_hd",
"numImages": "1",
"strength": 0.8,
"negativePrompt": "",
"seed": 0
}
Strength guide: higher (0.8–1.0) preserves more of the original frame; lower (0.1–0.5) follows the prompt more aggressively.
Best for: background swaps, environmental storytelling, “same hero, new location.”
Route: /home/ideogram/character-remix
Model 3: character
Place a character into a brand-new scene using only a reference still and text—no base image required.
{
"model": "ideogram-character",
"prompt": "Scene description (max 5000 characters)",
"referenceImageUrls": ["Character reference (1 image)"],
"renderingSpeed": "BALANCED",
"style": "AUTO",
"expandPrompt": true,
"imageSize": "square_hd",
"numImages": "1",
"negativePrompt": "",
"seed": 0
}
vs character-remix:
| Contrast | character | character-remix |
|---|---|---|
| Base image | Not required | Required |
| Scene source | Prompt only | Original frame + prompt |
| strength | N/A | ✅ |
Best for: character posters, multi-scene campaigns, net-new compositions anchored to one face.
Route: /home/ideogram/character
Model 4: v3-text-to-image
General-purpose generation from text with style, speed, size, MagicPrompt expansion, and optional negative prompt.
{
"model": "ideogram-v3-text-to-image",
"prompt": "Image description (max 5000 characters)",
"renderingSpeed": "BALANCED",
"style": "AUTO",
"expandPrompt": true,
"imageSize": "square_hd",
"negativePrompt": "",
"seed": 0
}
Style options: AUTO, GENERAL, REALISTIC, DESIGN.
Rendering speed: TURBO (fastest), BALANCED (default), QUALITY (slowest, highest fidelity).
Image size examples: square, square_hd, portrait_4_3, portrait_16_9, landscape_4_3, landscape_16_9.
Best for: concept art, product sketches, marketing stills without a reference photo.
Route: /home/ideogram/v3-text-to-image
Model 5: v3-edit
Mask-driven inpainting without a character reference—swap objects, fix backgrounds, or remove artifacts in a selected region.
{
"model": "ideogram-v3-edit",
"prompt": "What to paint inside the mask (max 5000 characters)",
"imageUrl": "Source image URL",
"maskUrl": "Mask URL",
"renderingSpeed": "BALANCED",
"expandPrompt": true,
"seed": 0
}
vs character-edit: v3-edit is identity-agnostic; character-edit adds referenceImageUrls so faces stay on-model.
Best for: prop swaps, watermark cleanup, localized retouching on non-character assets.
Route: /home/ideogram/v3-edit
Model 6: v3-remix
Transform an entire image toward a new style or content direction—no mask, no character reference, with adjustable strength.
{
"model": "ideogram-v3-remix",
"prompt": "Remix direction (max 5000 characters)",
"imageUrl": "Source image URL",
"renderingSpeed": "BALANCED",
"style": "AUTO",
"expandPrompt": true,
"imageSize": "square_hd",
"numImages": "1",
"strength": 0.8,
"negativePrompt": "",
"seed": 0
}
vs character-remix: v3-remix does not take a character reference; use character-remix when facial consistency matters.
Best for: material changes, illustration styles, holistic art-direction shifts.
Route: /home/ideogram/v3-remix
III. Cross-Model Comparison
3.1 Capability matrix
| Feature | char-edit | char-remix | char | v3 T2I | v3-edit | v3-remix |
|---|---|---|---|---|---|---|
| Text-to-image | ❌ | ❌ | ❌ | ✅ | ❌ | ❌ |
| Mask edit | ✅ | ❌ | ❌ | ❌ | ✅ | ❌ |
| Character reference | ✅ | ✅ | ✅ | ❌ | ❌ | ❌ |
| Global remix | ❌ | ✅ | ✅ | ❌ | ❌ | ✅ |
| strength | ❌ | ✅ | ❌ | ❌ | ❌ | ✅ |
| negativePrompt | ❌ | ✅ | ✅ | ✅ | ❌ | ✅ |
| imageSize | ❌ | ✅ | ✅ | ✅ | ❌ | ✅ |
3.2 Character line vs V3 line
| Product line | Models | Core idea |
|---|---|---|
| Character | character-edit, character-remix, character | Identity-locked edits and scene work |
| V3 create | v3-text-to-image | Prompt-only generation |
| V3 edit | v3-edit, v3-remix | Generic masked or global edits |
IV. Selection Decision Trees
Character workflows
Character consistency required?
|
|-- Local change inside a region (eyes, outfit detail)
| `-- character-edit (+ mask)
|
|-- New environment but same person
| |-- Have a frame to remix -> character-remix
| `-- Start from reference only -> character
|
`-- Campaign with many scenes
`-- character for new shots; character-remix to adapt existing frames
General image workflows
No character reference?
|
|-- Create from scratch -> v3-text-to-image
|
`-- Edit existing photo
|-- Local swap / cleanup -> v3-edit (+ mask)
`-- Whole-image restyle -> v3-remix
V. Practical Payload Examples
Expression tweak (character-edit)
{
"model": "ideogram-character-edit",
"prompt": "Open the eyes and add a gentle forward-looking smile",
"imageUrl": "https://example.com/portrait.jpg",
"maskUrl": "https://example.com/eyes-mask.png",
"referenceImageUrls": ["https://example.com/character-ref.jpg"]
}
Background swap (character-remix)
{
"model": "ideogram-character-remix",
"prompt": "Place the subject at Shibuya crossing at night, neon reflections",
"imageUrl": "https://example.com/character.jpg",
"referenceImageUrls": ["https://example.com/character-ref.jpg"],
"strength": 0.7
}
Landscape text-to-image
{
"model": "ideogram-v3-text-to-image",
"prompt": "Cinematic dusk lakeside, reeds and lily pads with golden reflections",
"style": "REALISTIC",
"renderingSpeed": "QUALITY",
"imageSize": "landscape_16_9"
}
Object swap (v3-edit)
{
"model": "ideogram-v3-edit",
"prompt": "Replace the coffee cup with a red mug",
"imageUrl": "https://example.com/room.jpg",
"maskUrl": "https://example.com/cup-mask.png"
}
VI. Final Recommendations
| Use case | Model | Why |
|---|---|---|
| Identity-safe local edit | character-edit | Mask + reference lock the face |
| Same character, new backdrop | character-remix | Remix with strength control |
| Character poster from ref | character | No base frame required |
| Fresh illustration / photo | v3-text-to-image | Full V3 style and size stack |
| Masked object replace | v3-edit | General inpainting |
| Global style shift | v3-remix | Whole-image transformation |
One-line playbook:
- Character consistency → character-edit, character-remix, or character.
- Create from scratch → v3-text-to-image.
- Masked local edit → v3-edit.
- Whole-image remix → v3-remix.
Open every Ideogram mode from /home/ideogram.
