Unleash the power of 360 cameras with AI-assisted 3D scanning. (Luma AI)

Olli Huttunen
15 Jun 202311:50

TLDRExplore the innovative use of 360 cameras with AI-assisted 3D scanning through Luma AI's neural Radiance Fields (Nerf) technology. This method advances traditional photogrammetry by creating volume models from camera recordings, allowing for the depiction of reflections and transparent objects. Luma AI simplifies the process with a user-friendly app, enabling anyone to scan objects and generate 3D models with ease. Discover the potential of this technology for creating compelling retakes and camera movements, and its future possibilities in the field of AI and 3D modeling.

Takeaways

  • 📱 Phones are increasingly capable, but 360 cameras are now being used for 3D modeling with AI-assisted 3D scanning.
  • 🤖 NeRF (Neural Radiance Fields) is an advanced 3D modeling technique that uses AI to calculate environmental data from camera recordings and create a volume model.
  • 🌐 Luma AI is a user-friendly cloud service that allows users to create NeRF models through a mobile app by simply scanning objects from all sides.
  • 🔄 After scanning, the video is sent to Luma AI's cloud for processing, and within about 30 minutes, users can interact with the 3D model.
  • 🎥 Luma AI enables the creation of new camera movements within scanned environments, allowing for compelling re-takes without returning to the shooting location.
  • 🌟 NeRF can represent reflections and transparent objects, which have been challenging to capture with traditional photogrammetry.
  • 🖥️ Luma AI's web service supports uploading various materials, including videos and still photos, for processing into NeRF models.
  • 🔍 360 cameras are compatible with Luma AI, understanding the wide shots taken with double fisheye lenses and supporting the equirectangular image format.
  • 🤳 Using a 360 camera simplifies capturing subjects from all sides, especially with features like stabilization and horizon lock, which are crucial for NeRF scanning.
  • 🌞 The best conditions for shooting are overcast days with minimal shadows and even lighting, though NeRF can handle some shadows and remove the photographer from the final model.
  • 🚫 Shadows and the photographer's presence can affect the scan quality, so it's recommended to stay behind the camera and focus the lens on the object.
  • 🌐 Full 360-degree images are useful for capturing tight spaces where it's not possible to move around objects, like narrow alleys or inside cars.
  • 🛠️ NeRF models can be exported into 3D programs, but they may require cleaning and smoothing out due to their initial broken and loose vertex structure.
  • 🎮 The most promising use of NeRF models is opening them in Unreal Engine as volume models, offering new possibilities for lighting and camera effects.

Q & A

  • What is the main topic discussed in the video by Olithutunen?

    -The main topic discussed is the use of neural Radiance Fields (Nerf) for 3D modeling with 360 cameras, particularly using the Luma AI service.

  • What is a neural Radiance Field (Nerf) in the context of 3D modeling?

    -A neural Radiance Field, or Nerf, is an advanced method for 3D modeling that uses AI to calculate and create volume models from the environment recorded by a camera, allowing for the exploration of the model in three-dimensional space.

  • How does the Luma AI service simplify the process of creating Nerf models?

    -Luma AI provides an app that allows users to easily scan objects by videoing them from all sides, then sends the video to the cloud for processing, resulting in a 3D model that can be viewed from different angles within about 30 minutes.

  • What is special about using a 360 camera for Nerf scanning compared to a regular camera?

    -A 360 camera captures a wider area and supports the equirectangular image format, which is ideal for Nerf scanning as it allows for capturing subjects from all sides more easily, especially in tight spaces.

  • How can you create camera movements inside a scanned environment using Luma AI?

    -With Luma AI, you can create new camera movements within the scanned environment and render retakes without returning to the shooting location, providing more flexibility in post-production.

  • What are the benefits of using a 360 camera with features like stabilization and horizon lock for Nerf scanning?

    -Stabilization and horizon lock features ensure that the image remains level and the subject stays in focus during scanning, which is crucial for creating accurate Nerf models based on circular motion around the object.

  • How does Luma AI handle the presence of the photographer in the 360 camera images during scanning?

    -Luma AI is capable of removing the photographer from the final model as they are constantly moving in relation to the background, leaving only the stationary objects in the model.

  • What is the best condition for shooting Nerf models with a 360 camera according to the video?

    -The best condition for shooting is overcast weather with few shadows and even lighting, as direct sunlight can create unwanted shadows and affect the scan quality.

  • How can the Nerf models created with Luma AI be used in other 3D programs?

    -Nerf models can be exported and opened in other 3D programs, although they may not be as accurate as those created with photogrammetry. They can also be downloaded as a volume model into Unreal Engine for further development.

  • What are the limitations of using Nerf models in their surface model format for 3D programs?

    -The surface models derived from Nerf scans can be quite rough and may have many loose vertices, making them prone to fraying and requiring significant cleaning and smoothing in 3D programs.

  • What does the future hold for neural Radiance Fields in the field of AI and 3D modeling?

    -Neural Radiance Fields are a developing technology with a lot of potential. While their real-world applications are still being explored, they offer new perspectives in 3D modeling and are expected to evolve rapidly alongside advancements in AI.

Outlines

00:00

📱 Exploring 360 Cameras and Neural Radiance Fields in 3D Modeling

Oli Thutunen introduces the concept of Neural Radiance Fields (Nerf), a cutting-edge technology for 3D modeling that uses AI to create volume models from photos or videos. Two primary methods for creating Nerf models are discussed: a complex, technical approach requiring Python and terminal commands, and a user-friendly cloud service called Luma AI. Luma AI allows users to create 3D models by simply scanning objects with their smartphones. The video explains how to use Luma AI's app, the process of sending videos to the cloud for processing, and the ability to create camera movements and render retakes within the scanned environment. The advantages of Nerf over traditional photogrammetry, such as the ability to represent reflections and transparent objects, are highlighted. Additionally, the video explores the potential of using 360 cameras for scanning, which can capture wider areas and is particularly useful for reaching high or low angles. The Insta 360 camera's stabilization and horizon lock features are emphasized for effective Nerf scanning.

05:02

🌐 Utilizing Luma AI for 360 Camera Scanning and Post-Processing

The script delves into the editing process using the Insta 360 Studio application, where users can adjust shots to keep subjects centered in the image. After editing, the video is rendered as HD and uploaded to Luma AI for processing. It addresses the challenges of including the photographer in the scan, which Luma AI can remove, but advises staying behind the camera to avoid distortions in the final model. The video also discusses the best conditions for scanning, such as overcast weather for even lighting and minimal shadows. Specific situations where full 360-degree images are beneficial, like narrow alleys or interiors of a car, are explored. The script notes that while Nerf models can be translated into surface models, they may appear uneven and lumpy in the browser, but these are low-poly versions, and the actual Nerf model contains more data for smoother rendering in videos.

10:02

🚀 The Future of Neural Radiance Fields and 360 Cameras in 3D Modeling

The final paragraph discusses the potential and current limitations of Nerf models when exported into 3D programs, noting that they are not as accurate as traditional photogrammetry but offer a unique perspective in 3D modeling. The script highlights the exciting possibilities of using Nerf models in Unreal Engine as volume models, allowing for advanced lighting and camera effects. It acknowledges that while the technology is still maturing, it presents a fascinating new direction in 3D modeling. The video concludes with a recommendation to experiment with 360 cameras for Nerf scanning, a call to like and subscribe to the channel, and a sign-off until the next video.

Mindmap

Keywords

💡360 cameras

360 cameras are devices capable of capturing panoramic images or videos, recording a full 360-degree field of view. In the context of the video, these cameras are used for 3D modeling by capturing the environment from all angles, which is essential for creating comprehensive 3D scans. An example from the script mentions using an Insta 360 camera for its stabilization and horizon lock features, which are beneficial for scanning 3D Nerf models.

💡3D modeling

3D modeling refers to the process of creating a representation of a three-dimensional object or environment in a virtual space. The video discusses how 360 cameras can be utilized for 3D modeling, specifically through a method called neural Radiance Fields (Nerf), which allows for the creation of volume models that can be explored in three-dimensional space.

💡Neural Radiance Fields (Nerf)

Neural Radiance Fields, or Nerf, is an advanced 3D modeling technique that uses artificial intelligence to calculate and generate a volume model from the environment recorded by a camera. Unlike traditional photogrammetry, Nerf can handle reflections and transparent objects more effectively. The script explains how this technology revolutionizes 3D scanning and model creation.

💡Photogrammetry

Photogrammetry is the science of making measurements from photographs, especially for recovering the exact positions of surface points. In 3D modeling, it traditionally involves creating polygon surfaces from a series of photos. The video script contrasts photogrammetry with Nerf, highlighting the latter's ability to produce more accurate and detailed 3D models, including those of reflective and transparent objects.

💡Luma AI

Luma AI is a cloud service mentioned in the script that offers an easy-to-use app for creating Nerf models. Users can scan an object by videoing it from all sides and then send the video to Luma AI's cloud for processing. The service is user-friendly and eliminates the need for complex programming skills, making 3D scanning more accessible.

💡Equirectangular image format

The equirectangular image format is a method of representing panoramic images where the entire 360-degree view is mapped onto a single rectangular image. The script explains that Luma AI supports this format, which is typical for 360 cameras, allowing for the creation of 3D models from wide-angle shots.

💡Insta 360 camera

Insta 360 is a brand of 360-degree cameras that are highlighted in the video for their stabilization and horizon lock features. These features are crucial for 3D scanning with Nerf models because they ensure that the camera remains level during the scanning process, which involves moving around the object in a circular motion.

💡Nerf models

Nerf models refer to the 3D volume models created using Neural Radiance Fields. These models can be manipulated and viewed from different angles, offering a more dynamic and interactive way to explore 3D environments compared to traditional polygon-based models. The script describes how these models can be created using Luma AI and 360 cameras.

💡Unreal Engine

Unreal Engine is a powerful game engine used for creating interactive experiences and visuals. The video script mentions the possibility of importing Nerf models into Unreal Engine as volume models, which opens up new possibilities for lighting, camera movements, and other visual effects that can enhance the 3D modeling experience.

💡Surface model

A surface model is a 3D representation made up of polygons that form the surface of an object or environment. The script discusses the option to translate Nerf models into surface models, which can then be downloaded in low, medium, or high poly versions. However, it notes that these models may not be as accurate or clean as those created through traditional photogrammetry.

Highlights

360 cameras can be used for 3D modeling with AI-assisted 3D scanning technology.

Neural Radiance Fields (Nerf) is an advanced 3D modeling technique using AI.

Nerf produces a volume model from camera recordings, allowing exploration in 3D space.

There are two ways to create Nerf models: a complex programming route and a user-friendly cloud service like Luma AI.

Luma AI offers an app for creating Nerf models by simply scanning objects with a phone.

Luma AI processes videos in the cloud, enabling users to view 3D models within 30 minutes.

Nerf models can simulate camera movements and render retakes without returning to the shooting location.

Nerf technology can represent reflections and transparent objects, which are difficult in traditional photogrammetry.

Nerf models can be trained with various cameras, including uploading materials through a web browser.

360 cameras with double fisheye lenses and equirectangular image format are supported by Luma AI.

360 cameras provide wider area coverage and easier positioning for scanning objects from all sides.

Insta 360 cameras offer stabilization and horizon lock features beneficial for Nerf scanning.

Insta 360 Studio allows editing of 360-degree videos to keep subjects centered for Nerf scanning.

Luma AI can remove the photographer from the final Nerf model during post-processing.

Overcast weather is ideal for Nerf scanning to minimize shadows and ensure even lighting.

Nerf models can be exported into Unreal Engine as volume models for advanced lighting and camera effects.

Nerf technology is a developing field with potential for significant impact in 3D modeling.

The video encourages viewers to try Nerf modeling with their 360 cameras for fun and exploration.