Logo of sharetextures

Single Image AI vs. Photogrammetry

4 months ago
Single Image AI vs. Photogrammetry

Welcome to a detailed comparison between two powerful methods for creating PBR (Physically Based Rendering) materials: the single image AI-based approach using Substance Alchemist and the traditional photogrammetry technique. In this post, we’ll dive into the workflow and results of each method, using a brick wall surface as our test subject.


Photogrammetry Workflow

The photogrammetry process is well-known for its accuracy and detail. Here’s a breakdown of how it was used to create a PBR material from a brick wall:


  1. Image Capture: 164 perpendicular images were taken, including two from a longer distance to cover a broader area.
  2. Image Processing: RAW images were color-calibrated and processed using photo editing software.
  3. 3D Reconstruction: The images were fed into 3DF Zephyr to generate a high-poly 3D model.
  4. Model Preparation: A low-poly model was created in ZBrush for baking.
  5. Texture Baking: Albedo, normal, height, and ambient occlusion maps were baked using Substance Designer.
  6. Manual Adjustments: Textures were refined in Substance Painter, including seam removal and tiling adjustments.


This process took approximately 3.5 hours, with various steps contributing to the overall time, including processing and manual adjustments.


Single Image AI Approach with Substance Alchemist

Substance Alchemist offers a more streamlined alternative using AI to generate PBR materials from a single image. Here’s how it works:


  1. Image Selection: A high-resolution image of the brick wall was chosen. Two images were captured from a distance to fit the required coverage.
  2. Cropping and Adjustment: The image was cropped and adjusted to match the coverage of the photogrammetry model.
  3. AI Processing: The image was processed using Alchemist’s AI-based Image to Material algorithm, generating PBR maps.
  4. Manual Refinement: Adjustments were made to improve texture quality, including tiling, seam removal, and color equalization.


This process took around 15 minutes, showcasing a significant time-saving advantage over photogrammetry.


Comparing Results


Photogrammetry Results:
The textures generated through photogrammetry were highly detailed and accurately reflected the original brick wall. The high definition and precision of the textures resulted in a clean and well-defined appearance.

Single Image AI Results:
The textures produced by Substance Alchemist were impressive for a single image approach, with good overall quality. However, there were some noticeable inconsistencies and blurriness compared to the photogrammetry results.


Key Differences


  • Detail and Accuracy: Photogrammetry offers superior detail and accuracy, closely matching the original surface. The AI-based method, while fast, may not capture the same level of detail and can produce some fuzziness.
  • Processing Time: Alchemist provides a much quicker solution, reducing the time from hours to minutes.
  • Ease of Use: The AI-based approach simplifies the workflow but may require additional manual adjustments for optimal results.


Conclusion


Both techniques have their strengths and applications. Photogrammetry is ideal for high-precision needs and complex surfaces, whereas Substance Alchemist is a valuable tool for rapid texture creation when time is limited. Each method has its place in a 3D artist’s toolkit, and the choice depends on the specific requirements of the project.

Substance Alchemist represents a significant advancement in AI-driven material creation, offering a practical solution for many scenarios. However, for the highest quality and accuracy, photogrammetry remains the gold standard.


We hope this comparison helps you understand the pros and cons of each approach and aids in selecting the right tool for your needs. 

Keep in touch:)