Large-Scale 3D Digital Twins with AI and Unreal Engine | Unreal Fest 2023

Creating 3D Digital Twins is a challenging and time-consuming task, requiring the combination of many technologies and in particular an extensive use of the latest AI technologies. Join us to find out how AI and Unreal Engine are being used to enhance large-scale 3D digital twins that are being built by Blackshark.AI leveraging UE and Quixel assets.
Hannes Walter, VP of Product Management at Blackshark.AI will demonstrate how Quixel and other Unreal Engine capabilities can further improve the quality of the results. In addition, he will present the most recent results obtained by Blackshark.
We’re excited to bring you sessions from Unreal Fest 2023, available to watch on demand: • Unreal Fest '23 Playlist

[@artiomarenkov99]

Yeah so i see that nothing changed from the last showcase you did in like 2021, still the plugin is not availble as well as the site that says that only New Orleans demo is availble😂

[@SuperMyckie]

Maybe when you try building something you'd understand

[@joseandreszuluaga]

Wonderful

[@1nsanetr]

Impressive!

[@jimj2683]

All you need to do is wait for compute power to get cheaper. Then you can simply train a giant neural net on all the photos, videos, text and maps on the internet. The ground truth data will be millions of photos and videos with recorded position, orientation and time/date. The AI will generate realistic guesses for places with missing data. Then you could add an AI model on top that changes things to meshed. Or just go full world-model end-to-end. In the mean time you can train a model on bing maps and use the best photogrammetry cities as ground truth. Also you should link place names in the maps with photos related to that name. That way you can get real detail at ground level. Alternatively synthetic data from computer games. The type of "manual" AI stuff I have seen in MSFS so far is not impressive at all once you get closer than 1km. Also, the ground textures just look like low res photos and the green color seems off.

[@roldanduarteholguin7102]

Export the Azure, Power Apps, Copilot, Chat GPT, Revit, Plant 3D, Civil 3D, Inventor, ENGI file of the Building or Refinery to Excel, prepare Budget 1 and export it to COBRA. Prepare Budget 2 and export it to Microsoft Project. Solve the problems of Overallocated Resources, Planning Problems, prepare the Budget 3 with which the construction of the Building or the Refinery is going to be quoted.

[@k4rl33]

Gonna write you ASAP.

[@petrbondar535]

👍

[@NUCLEARARMAMENT]

Imagine having a digital twin of yourself, or someone who's no longer with us.

[@jimj2683]

With all the photos on facebook/instagram they could copy most peoples looks with AI and place them in this digital twin world.

[@edmundbell-king9756]

Thats an avatar.

[@cg-school]

Kaliningrad, Moscow, Belgorod - this is Ukraine. Just waiting a few mo )

[@insanedevice]

Kaliningrad - Ukraine? You need to check twice lmao

[@igorewshik]

russia lose official border, so... I mean, they change their laws to include ukrainian regions and no one can tell where their borders now. Government told that they will discuss borders with people on that territories, but there ukrainian army and they didn't agree with new laws. And no country in entire world agreed. So, you can't say that some city inside russia or not.

[@gromislav]

@@igorewshik true story

[@zes7215]

wr

[@jimj2683]

First he talked badly about Google Earth photogrammetry, then he shows his own companies procedural buildings that look nothing like the real buildings (apart from the roof color and shape). The AI needs to relate place names on the map with photos of that place on the internet (e.g.: it links New York on the map with all the photos of New York in Google Images). That way you could have much better detail at ground level.