Which one generates a better 3D model from video? Luma AI or 3DPresso?
TLDRThis video compares Luma AI and 3DPresso, two AI applications for creating 3D models from video. Hosted by olioutron, the video demonstrates the process of 3D scanning using a 360 camera and evaluates the output quality of both services. While Luma AI excels in capturing environments and larger objects, 3DPresso produces more accurate and usable polygon-based models, especially suitable for 3D printing and gaming assets. The video concludes that both applications have their unique strengths and are continually evolving in the field of AI-assisted 3D modeling.
Takeaways
- ๐ Luma AI and 3DPresso are two AI-assisted applications that can create 3D models from video images.
- ๐ Luma AI is developed by Luma Labs, a Silicon Valley startup, while 3DPresso is a newer Korean alternative.
- ๐ Both applications utilize the neural Radiance field (NeRF) technology for 3D modeling from video.
- ๐ฅ The video demonstrates using a 360-degree camera to capture video of objects for 3D scanning.
- ๐น The video script explains the process of recording and editing 360-degree videos for optimal 3D model creation.
- โฑ Luma AI is faster in processing 3D models compared to 3DPresso, which has a new credit-based billing structure.
- ๐ Luma AI is currently free to use, with some limitations on rendering camera animations.
- ๐ฐ 3DPresso offers a free initial credit to new users and charges for additional usage.
- ๐ฟ The video compares the models produced by both applications for a statue, finding 3DPresso's model to be more accurate and complete.
- ๐ฆ For a lamppost with transparent surfaces, both applications struggled, but 3DPresso's model had slightly better textures.
- ๐ For a large trailer, Luma AI produced a usable model, whereas 3DPresso did not, indicating Luma's strength in larger objects.
- ๐ฎ The script concludes that both applications have their own strengths and are suited for different purposes, with 3DPresso being more accurate for polygon-based models and Luma AI better for capturing environments and larger entities.
Q & A
What are the two AI-assisted applications mentioned for generating 3D models from video images?
-The two AI-assisted applications mentioned are Luma AI, developed by Luma Labs, and 3DPresso, a Korean alternative.
What technology do Luma AI and 3DPresso use for creating 3D models from video?
-Both Luma AI and 3DPresso use a technology called Neural Radiance Fields, also known as NeRF.
What is the advantage of using a 360 camera for 3D scanning as described in the script?
-The advantage of using a 360 camera for 3D scanning is that it can capture larger areas at once, making it suitable for subjects of your own size or larger, and it simplifies the process of recording laps around the objects.
How does the Horizon lock feature on the 360 camera assist in the scanning process?
-The Horizon lock feature keeps the camera leveled even when the selfie stick is turned, ensuring that the person recording does not appear in the final image when the video is processed.
What are the challenges faced when scanning objects with transparent surfaces like the lamp post in the script?
-Scanning objects with transparent surfaces is challenging because these surfaces often cause problems in photogrammetry modeling techniques, leading to difficulties in creating accurate 3D models.
What is the significance of the statue model in comparing the capabilities of Luma AI and 3DPresso?
-The statue model is significant because it tests the ability of both Luma AI and 3DPresso to accurately capture and render the details and textures of an object, including challenging areas like the background and the object's surface.
How does Luma AI handle the rendering of the background in its models?
-Luma AI has the capability to record the background around the object, which can be seen in the pre-calculated NeRF rendering, although the accuracy may decrease when viewed in 3D mode.
What is the difference between the Nerf model and the regular polygon model in terms of accuracy?
-The Nerf model is a pre-calculated rendering that may look different from the regular polygon model when translated into a 3D mesh. The regular polygon model tends to be less accurate, with lumpy surfaces and holes, especially in the case of Luma AI's output.
How does 3DPresso perform in comparison to Luma AI when processing the statue model?
-3DPresso produces a more accurate model with better textures compared to Luma AI. It also manages to create a cleaner 3D mesh and fix holes in areas where Luma AI's model was less accurate.
What is the importance of examining the models in Blender after processing with AI applications?
-Examining the models in Blender is important because it allows for a more detailed assessment of the 3D mesh's accuracy,ๅฎๆดๆง, and the quality of the textures, which is crucial for determining the model's usability for purposes like 3D printing or game development.
What are the different outcomes when scanning the lamp post and the trailer with Luma AI and 3DPresso?
-When scanning the lamp post, both Luma AI and 3DPresso capture the surroundings well but struggle with the narrow forms of the lamp post itself. For the trailer, Luma AI produces a usable model, while 3DPresso fails to create a usable model, indicating that each application has its strengths and limitations depending on the object's size and complexity.
What conclusion can be drawn from the comparison of Luma AI and 3DPresso based on the script?
-The conclusion is that Luma AI and 3DPresso have their own purposes. Luma AI is better for saving environments and larger entities and is capable of producing real NeRF models, while 3DPresso is more useful for creating accurate and complete polygon-based mesh models suitable for 3D printing or game development.
Outlines
๐ Introduction to AI-Assisted 3D Modeling from Video
This paragraph introduces the topic of using AI to create 3D models from video images, mentioning two specific tools: Luma AI by Luma Labs and 3D Presser, a Korean alternative. Both utilize neural radiance fields, or NeRF technology. The speaker, Oliutron, sets the stage for a comparative exploration of these web services using video files of three different objects captured with a 360-degree camera. The process of 3D scanning with a 360 camera is briefly explained, highlighting the advantages of this method for larger subjects and the importance of recording from various angles to ensure complete scanning.
๐ฅ Post-Processing 360 Video for AI Analysis
The second paragraph details the process of editing the 360-degree video footage for optimal use with AI modeling software. The video material is uploaded to a computer, cropped into a 16x9 format, and keyframes are added to keep the object centered during the rotation. Specific settings for focal length and lens distortion are provided, along with instructions for exporting the video in mp4 format at a high resolution and bitrate. The goal is to prepare the video material for analysis by AI programs like Luma AI and 3D Presser, which will then generate 3D models.
๐ค Comparing AI-Generated 3D Models
This paragraph compares the results of using Luma AI and 3D Presser to create 3D models from the video files. Luma AI is noted for its speed and the inclusion of a pre-calculated NeRF animation, which provides an impressive visual effect but may not reflect the final model's accuracy. On the other hand, 3D Presser is praised for producing more accurate models with better textures, despite not including the background in its models. The comparison extends to examining the models within Blender, revealing that 3D Presser's model is more intact and has a cleaner mesh, making it more suitable for 3D printing or use in games.
๐ Final Verdict on AI 3D Modeling Services
The final paragraph summarizes the comparative analysis of Luma AI and 3D Presser, highlighting the strengths and weaknesses of each service. Luma AI excels at capturing environments and creating NeRF models suitable for use in Unreal Engine, while 3D Presser is better for producing clear polygon-based mesh models for 3D printing or game assets. The speaker acknowledges the early stages of development for these applications and the potential for future advancements. The paragraph concludes with an invitation for viewers to experiment with these services themselves and a reminder of the speaker's other related videos, encouraging engagement with the channel.
Mindmap
Keywords
๐กAI-assisted application
๐กNeural Radiance Fields (Nerf)
๐ก360 camera
๐ก3D scanning
๐กInsta360 RS1 camera
๐กInsta360 Studio
๐กLuma AI
๐ก3DPresso
๐กNerf rendering
๐กGLB format
Highlights
Luma AI and 3DPresso are two AI-assisted applications that generate 3D models from video images using neural Radiance field technology.
Luma AI is developed by Luma Labs, a Silicon Valley startup, while 3DPresso is a newer Korean alternative.
Olioutron compares Luma AI and 3DPresso by using a 360 camera to create videos of three different objects for 3D scanning.
The 360 camera is utilized to capture 180-degree areas, focusing on the front lens view for scanning.
A selfie stick is recommended for shooting 360 videos to maintain a proper distance from the camera.
The Insta360 RS1 camera, capable of recording up to 6K resolution, is used for capturing high-quality video.
The 360 camera is particularly advantageous for scanning larger subjects due to its wide-angle lens.
The process involves three rounds of recording at different heights to capture the object from various angles.
Editing 360 videos in Insta360 Studio involves cropping and rotating to keep the object centered during laps.
Exporting videos in 1920x1080 resolution with a high bitrate setting ensures better quality for AI processing.
Luma AI processes 3D models faster than 3DPresso but has limitations on rendering camera animations.
3DPresso uses a credit-based billing structure, offering new users free credits to start with.
Luma AI automatically creates a pre-calculated Nerf rendered animation showcasing the model from different angles.
3DPresso produces more accurate models with better textures compared to Luma AI in the web browser view.
Examining models in Blender reveals that 3DPresso's models are more intact and less fragmented than Luma AI's.
Luma AI excels in capturing environments and larger entities, making it suitable for Unreal Engine and similar programs.
3DPresso is better for creating usable polygon-based mesh models for 3D printing or game development.
Both applications have their own strengths and are in the early stages of development, with potential for future advancements.
Olioutron encourages viewers to test these services themselves, using any smartphone capable of video recording.