Which one generates a better 3D model from video? Luma AI or 3DPresso?

In this video, we'll compare Luma AI and 3Dpresso, two of the leading 3D NeRF web services that can generate 3D model from video. We'll see which one makes better and most useable 3D model for asset that you can use in games or even in 3D printing. Or is the purpose to generate 3D surroundings fast. I also demonstrate on this video how you can scan object with 360 camera. It is fun and clever way to videoscan your subjects for these 3D services.
3Dpresso: 3dpresso.ai/
LumaAI: lumalabs.ai/
#lumaai #3dpresso #comparison

[@Hi_mynameis_Joshua]

Thanks for the comparisons. I am familiar with Luma, but not 3dPresso, so it'a always another adventure finding new products.

[@badxstudio]

What a fantastic comparison! Can’t wait to try this out!

[@OlliHuttunen78]

Thanks Guys! I'm glad that you whatched it!

[@paullee8209]

This comparative analysis was well shot, edited, and straight to the point. Thank you for covering all the major bases in a concise video!

[@mag_and]

Both Luma AI and 3Dpresso impress with "beautiful" results, but I still haven't come across anything that beats standard photogrammetry when it comes to accuracy.

[@johnw65uk]

yeah these are good for speed, but quality you still need either a laser scanner or high res dslr photos and lots of them.

[@benjamindameworth5351]

Photogrammetry only works for individual objects. What Luma and 3DPresso do is able to scan entire rooms, outdoor areas, homes, or even city blocks.

[@tjust88]

Very informative. Thank you.This field requires nearly constant learning as it is developing and expanding so rapidly. The tech we already have at hand is incredible and I am excited to see where it takes us. I've had limited success with my Mini 3 Pro, iPad Pro and Insta360 X3 used for photogrammetry. I've had better--but still not consistent--results when GPS is included in the metadata of photos rather than video. I've yet to nail down a reliable, repeatable workflow but I learn something new with each attempt.

[@marleneponciano7674]

Congratulations on the quality of the analysis

[@seksonexead]

Nice video! Very detailed comparison!!

[@isherwood9321]

thank you for the comparison. I eventually decided for a paid account at 3dpresso

[@paveldesort5657]

Great video and kudos for the hard work you put in these. After carefully comparing terms and conditions by both companies in regards to the uploaded materials, I decided no to go forward with uploading equirectangular images or videos. Both companies can use uploaded materials in any way they see fit including images and videos of the author being on the equirectangular photos and videos. Still great that these options are there. Just need to think through what and how a user records and shares. For example 360 video of my living room with myself recording it. Pretty sure, In few months it will be available like Stable diffusion is for offline home use.

[@E102Gaming]

To follow up on this (for anyone else interested), here's the specific section in their terms that say this:
Checking for "can use uploaded materials in any way they see fit"
Luma AI's ToS (Date of Last Revision: 03-28-2023)
"You hereby grant Luma AI and its affiliates, successors and assigns a non-exclusive, worldwide, royalty-free, fully paid-up, transferable, sublicensable (directly and indirectly through multiple tiers), perpetual, and irrevocable license to copy, display, upload, perform, distribute, store, modify, and otherwise use your User Content in connection with the operation of the Service and the promotion, advertising or marketing of the foregoing in any form, medium or technology now known or later developed."
3dpresso's ToS (ast Updated: [2023.09.14])
"② Notwithstanding the foregoing, the Company may use such content for the purpose of operation, display, transmission, distribution, promotion of the Services without a separate permission or consent from the author and without any compensation in accordance with applicable law and fair and common practice of transactions as follows: a. Use of the content for the purpose of operation, promotion, improvement of the Services, development of new services, collecting data for AI training datasets."

[@Lexikon69]

@@E102GamingSo both aren't compatible with european DSGVO law.

[@Vlow52]

The simple answer according to results is: wait till next year

[@Dronerangerspro]

Great job and thank you 🙏

[@nathanburley9122]

Great comparison. I hadn't heard of 3Dpresso. How hard is it to bring models / scenes into something like Quest 3 and walk around them?

[@antdx316]

good tests

[@tracingzimpossible]

can anybody tell me how to download full mesh with back ground from luma AI , it autimatically selets center arear when i try yo download the mesh, even if i crop the area i want to download it doesnt let me do so

[@OlliHuttunen78]

Yes I noticed same problem. Models are pre cropped if you try to download them in mesh mode. Luma AI gives currently only the whole environmen of the scan in volume model which you can open in Unreal Engine.

[@TrasThienTien]

Benign competition is necessary

[@michael_knight3457]

Hello! Can both programs generate a 3D image on a scale of 1 to 1? By scanning something, e.g. a part of a machine, will we be able to print it with the same size, width, etc.? And this is all from photos? Not having a 3D scanner.

[@OlliHuttunen78]

Well not directly. These models are not that accurate. You need to edit the scale to correct dimensions in 3D program. And you need to check the surface structure several times to get these in 3D printable condition.

[@ciprian683]

How to remove logo/watermark from lumalabs ai ?

[@jaakkotahtela123]

Saako noita valmiita 3d-malleja jotenkin esimerkiksi nettisivuilla pyöriteltäväksi, tai saako jopa Facebookiin?

[@OlliHuttunen78]

Kyllä. Luma AI:lla on ainakin palvelussaan upotuskoodit joilla mallin voi jakaa ja upottaa minne vain nettisivuille.

[@user-qv1gj4dq2f]

What if I use Insta360 X3 and take 5.7K video? I don't have an Insta 360 RS

[@OlliHuttunen78]

Sure. It suits fine. Point is not the resolution of the video after all.

[@jbgheel5028]

People have been doing it for years with LiDAR, for decades with Photogrammetry.
Its dumb when revolutionary tech is like “we did it first, well, we did it first this good!”
Nope, no you didn’t.
Easier, yeah.

[@johnw65uk]

well apart from this is faster, easier and cheaper than traditional photogrammetry the quality isnt too bad. and yes ive been doing photogrammetry since 1998.

[@BryanHoward]

To be honest they both look disappointing compared to conventional photogrammetry. The 3DPresso version seems to have some grid artifacting baked into the mesh.