How To Use

1. Upload Your Image – Start by selecting a 2D concept image, sketch, or silhouette to use as a reference. This will serve as the foundation for the 3D rendering.
2. Describe Your Image (Optional) – You can provide a textual description of the image to guide the AI in rendering the desired style and details.
3. Run the Workflow – Invoke the workflow to generate an output that incorporates realistic textures, PBR materials, and high-poly details suitable for previs or game asset visualization.‍
4. Review & Save – The final rendered image will be saved to your selected board, ready for use in previs pipelines or further refinement in 3D modeling software.

‍

How It Works

1. This workflow takes a 2D image or sketch and progressively transforms it into a high-quality 3D render through a structured pipeline of AI-based processing. Below is a step-by-step breakdown of what happens under the hood:1. Input Processing
   • Image Input: The user provides a 2D sketch, silhouette, or concept image that serves as the foundation for the 3D transformation.
   • Optional Text Input: Users can add a brief description of the image to influence the rendering, specifying key details that should be highlighted.
   2. Preprocessing & Structure Extraction
   • Image Preprocessing: The image is resized and optimized to a standard resolution suitable for processing.
   • Edge & Structural Detection (ControlNet - SoftEdge Model):
     
     • The workflow extracts the major contours, outlines, and shapes from the input image. This step ensures the structure is preserved while allowing the model to reimagine the image in a 3D-rendered style.
   3. Generating the 3D Look
   • Style and Prompt Assembly: A structured positive prompt is created by combining user-provided input with predefined keywords that enhance the 3D realism (e.g., “game engine render, high-poly model, PBR materials”).
   • A negative prompt removes unwanted styles like “2D, flat, illustration” to steer the output towards a 3D aesthetic.
   • Noise & Latent Processing: The workflow generates an initial noisy image (latent space representation), which is progressively refined into a high-quality render.
   • The denoising process applies structured noise reduction while preserving the essential structure extracted from the original 2D image.
   4. Refinement & Enhancement
   • Control Influence & Weighting: The AI dynamically applies structural control settings to balance creativity and accuracy in the transformation.
   • Fine-tuned parameters adjust sketch strength, structure adherence, and artistic flexibility.
   • Final Rendering: The AI generates a detailed 3D-like render, complete with realistic textures, lighting, and materials that mimic game-ready assets.
   5. Output & Saving
   • Final Image Output: The processed image is saved to a selected board, making it easy to access for further use in previsualization or as a reference for 3D modeling.
   • Further Editing Options: Users can refine the output by adjusting input parameters and running additional passes to perfect the look.
   This structured workflow ensures that a simple 2D concept can be efficiently transformed into a detailed 3D previs asset, suitable for game development, previsualization, or digital art pipelines.