Inside Unreal: Building Simulation Applications with Unreal Engine

The training and simulation domain is trending strongly these days. This week on Inside Unreal we’ll discuss what we mean when we talk about Simulations and the work done by creators and technology providers in the Unreal Engine ecosystem. We’ll also demonstrate how to integrate worldwide content in your applications with the use of the Cesium for Unreal plugin and Esri’s ArcGIS Maps SDK for Unreal Engine plugin, and we’ll follow up with Q&A!

Thursday, March 4 @ 2:00PM ET - Countdown


Alban Bergeret - Solution Architect, Simulation & Training - @Vertex85CL
Sébastien Lozé - Industry Manager, Simulations - @SLFeeding
Victor Brodin - Community Manager - @victor1erp

Cesium For Unreal
ArchGIS for Unreal
Simblock for Unreal
Virtuvrio from ESRI


Can’t wait to watch this…

64b heightmaps :frowning: geo-location :frowning:

how long is this seminar?

I am looking forwards to this!

wow seams to be very interesting as i m working on a similar project

Is this something that we are going to be able to watch at a later time or only a live-stream?
Living in Australia the start time is 6:00 AM and I am unable to watch it at that time.

Same! will have to watch the replay cause it will stream at 2am in where I am. :o

This is awesome! Esri’s ArcGIS is incredible powerful and useful when making real world environments. I used it to generate a splat map from a segmentation map and recreate a real world location similar to how they did it in Microsoft Flight Simulator 2020. Forum post HERE.

incredible, awesome, unbelievable

You can watch all of our livestreams on Twitch (perpetually) and on YouTube afterwards! YouTube playlist for Inside Unreal can be found here:

Hey I would love to chat about how you get that semantic segmentation going! thats really really interesting!
not sure how we can best get in touch but would love to discuss!

Are the slideshow going to be uploaded?


Is the problem with moving away from the center of the “world” giving jitter (floating point precision error) solved?
For example 200 km from the “center of the world”? Because being in an airplane and the camera is not tied to the center of the model coordinates, I would like the camera not to fly by itself, but to be tied to a specific point in the model.
P.S. Sorry for my bad english
P.S. The size of the map is about 400x400 km and this is a single “mesh”

Couple of questions-

Does the georeferencing plugin work with the ArcGIS Maps SDK? I noticed the georeferenced coordinates in the UI are quite off when in the ArcGIS Maps SDK (it has New York at 49ºN, 3ºE rather than 40ºN, 74ºW). I have been able to replicate this on my own with the plugin from GitHub. Any thoughts on what’s causing that?

Will the new georeferencing plugin with the flat earth/round earth toggle be available via GitHub?


1 Like

Thats a great question. Its a shame that the only thing they show at the livestreams is how to use X plugin. Information on how to avoid float precision limitations like the one you mention or how to create the spheric terrain will be much more usefull. Miss the in depth talks that really show how to solve problems.

Hi Unreal Engine Simulation team Alban, Sebastien and Victor

I am interested in using UE4 for Wildlife Conservation - specifically ecosystem-plant-animal interaction simulation for ecosystem monitoring and deeper understanding of interactions and effects of e.g. climate change.

I am thinking along the lines of spatially, temporally and nutritionally / energetically explicit Agent-based modelling. Plant types and animal types are needed as agents very much like AI entities in UE4, but that have to be parameterised with real-world parameters:

e.g. physiological parameter of metabolic rates, pregnancy status, birth events and young animal growth, and animal mortality parameters (e.g. linked to episodic events, disease/predators, lack of energy/key nutrient intake). They would have energy use values for different activities, daily energy needs and key nutrient needs and intake rates (which are plant-agent-part dependent and impacted by season and climate trends).

Predator animals would have their own parameters and behaviours, with prey providing the energy / key nutrients.

Animals would also have various biological urges and behaviour ranges (motility, aggression, predator avoidance or prey-seeking, mating, care and defence of offspring, need for shelter or water for thermo-regulation).

The plants would have their own parameter like water-use rates, growth rates (size), plant parts with values for digestibility and toxicity to herbivores, a way to enact parameter changes (eg height, impacts on growth rate when herbivores eat them, or from floods or fire, and their natural senescence patterns.

And and simulation should include all these interacting with the real-world environment characteristics of climate and geolocation such as rainfall patterns, daylight hours, temperatures, FIRE (a whole thing on its own).

Landscapes would be realistic based on actual digital terrain models, and plant distribution would depend on terrain, elevation, soil types(re soil characteristics to hold water deliver water & nutrients to plants) and aspect (re sun hours). Also ability so simulate flooding in lake and riparian (river) systems. Ideally data to inform and update the landscape and climate (eg daily weekly or monthly) should come from free GIS information provides - especially Google Earth Engine.

Important feature of this simulation process would be monitoring (output) of above parameters via sending data to in-simulation or external data analysis programs, and generating summary infographics etc. One could have ability for the “player” to Possess any AI entity to go along for the ride or tweak attributes to learn more. and ability to be an observer entity moving alongside an individual animal of choice.

If we really went Big-Picture, the impacts and effects of human activities and infrastructure could be added to the ecosystem simulation (e.g. increasing development encroaching on land, poaching, disturbance, canalisation of river courses etc).

This is a HUGE ask - would it ever be doable? The issue of reasonable simulation of plants (numerically huge) may be one of biggest challenges. Realisitc graphic representation would be nice but could (should, I guess) be simplified for performance gains,

It would be a godsend to assist / test ecological understanding e.g. of emergent features of interacting ecosystem parts, and to help decision-making in taking needed actions in natural area and endangered species management, and would also assist in Citizen and next-generation wildlife management staff education globally. Any suggestions on how to develop such an amazing tool?

Start with a small scope!

@VictorLerp Great stream! You said in it that links post in the live chat will be post here under the ressources section but there aren’t. Could you post them?
Thanks in advance