Uploading Reference Images

4.1 Uploading an Image

• Drag and drop

• Browse from system

• Maximum file size: 20 MB

4.2 Choosing from History

• Previously generated images

• Previously uploaded images

• Combined “All” view

4.3 Uploading Multiple Images (Model Dependent)

• Supported in models like Gemini Nano Banana and CDream

• Add multiple reference images

• Multi-image blending and inspiration

---

3.1 Available Models

The Image-to-Image workspace supports multiple model types, each optimized for different transformation needs.

Gemini Models

• Balanced image transformation

• Strong prompt alignment

• Multi-image support (model dependent)

• Good mix of structure + creativity

Best for: General-purpose edits and controlled transformations.

---

Nano Banana Models

• Lightweight and fast

• Cost-efficient generation

• Good for quick variations

Best for: Rapid experimentation and iteration.

---

Flux Models

• Highly creative

• Cinematic and artistic rendering

• Strong stylization capabilities

Best for: Marketing visuals, concept art, stylized transformations.

---

GPT Image Models

• Instruction-driven editing

• Smart contextual understanding

• Strong alignment with complex prompts

Best for: Precise edits based on detailed instructions.

---

CDream Models

• Diffusion-based creative transformations

• Texture and aesthetic enhancement

• Supports expressive reinterpretations

Best for: Artistic and imaginative transformations.

---

Canny ControlNet

• Edge-detection guided transformations

• Preserves outlines and structural boundaries

• Strong shape accuracy

Best for: Product edits, architecture, technical visuals.

---

Depth ControlNet

• Depth-map guided transformations

• Maintains foreground/background separation

• Preserves spatial structure

Best for: Interior scenes, environments, perspective-sensitive edits.

---

Image Variation Models

• Generates multiple alternative versions

• Maintains core subject

• Explores aesthetic diversity

Best for: Creative exploration and design iteration.

---

3.2 Choosing the Right Model

Model selection depends on your transformation goal.

---

Style Transformation

Use when you want:

• Anime, cinematic, oil painting, fantasy styles

• Strong artistic reinterpretation

Recommended:

• Flux

• CDream

• Image Variation models

---

Structural Preservation

Use when you need:

• Accurate shape retention

• Minimal distortion

• Controlled modifications

Recommended:

• Canny ControlNet

• Gemini (balanced modes)

---

Depth-Aware Modification

Use when:

• Scene geometry matters

• Foreground/background separation is important

Recommended:

• Depth ControlNet

---

Edge-Controlled Transformations

Use when:

• Line accuracy is critical

• Technical layouts must remain intact

Recommended:

• Canny ControlNet

---

Multi-Image Blending (Model Dependent)

Use when:

• Combining subject + style references

• Blending inspiration images

Supported in:

• Gemini models

• GPT image models

• Selected Flux variants

Availability depends on model capability.

---

3.3 Model Credit Cost

Understanding credit usage ensures cost transparency.

---

Credits Displayed Per Image

Each model shows:

• Credit cost per generation

• Resolution-based cost (if applicable)

Displayed before generating.

---

Varies by Model

Credit usage differs due to:

• Compute complexity

• Advanced controls (e.g., ControlNet)

• Image size and resolution

Lightweight models → Lower creditsAdvanced structure-controlled models → Higher credits

---

Multiplies Based on Number of Generations

Total Credits =Credits per image × Number of outputsExample:

• 10 credits per image

• 3 generations

• Total = 30 credits

---

4.1 Uploading an Image

To start transforming:

Drag and Drop

• Drag your image directly into the workspace.

Browse from System

• Click upload

• Select from your device

Maximum File Size

• Up to 20 MB per image

Ensure:

• Supported format (e.g., JPG, PNG)

• Clear, high-quality source image for best results

---

4.2 Choosing from History

Instead of uploading again, you can reuse past images.

---

Previously Generated Images

• Select outputs from earlier sessions

• Modify or refine them further

Previously Uploaded Images

• Access images you’ve uploaded before

• Save time during iteration

Combined “All” View

• Unified history view

• Shows both generated and uploaded images

• Enables seamless reuse

---

4.3 Uploading Multiple Images (Model Dependent)

Some models support multiple reference images.Supported in models like:

• Gemini

• Nano Banana

• CDream

---

Add Multiple Reference Images

You can:

• Upload a subject image

• Upload a style reference

• Upload environment inspiration

---

Multi-Image Blending and Inspiration

The model can:

• Combine structure from one image

• Apply style from another

• Merge visual characteristics

• Generate hybrid outputs

This enables:

• Advanced creative direction

• Design exploration

• Complex visual storytelling