Colonist! Time to check on your recruits, are they performing assigned tasks?
Everyone is working diligently except… The Baron of course… too good for your task?! to the Mines you Fool!!!
@ParallelColony
PATCH 1.0.4
Patch 1.0.4 is now available! No downtime needed for this one. See below for screenshots. These are posted directly in the Parallel Discord server. Enjoy the new features and have a great weekend, Colonists!
[notes included below]
Does anyone in my network have a contact at @AppLovin?
We’ve been having some issues with our account and haven’t been able to get support. An introduction to the right person would be greatly appreciated!
@OpenAI I expect this technology to improve dramatically, and I’ll keep testing it.
For now, show me the close-ups, the meshes, and the cleanup time alongside the five-minute demo.
That���s the part that tells me whether we’ve actually saved work.
I tested @OpenAI Astra’s 3D modeling skills. Absolutely impressive. And also, if this were a junior artist’s portfolio test, I wouldn’t hire them.
I recreated Turin’s Galleria Subalpina in Blender, aiming for an eventual Unreal Engine experience. Here’s what happened.
@OpenAI It reminded me of an eager artist more focused on impressing the team than delivering work the team can use.
Eventually, you hand the assignment to your trusted senior.
The first impression was extraordinary. The production handoff was another story.
@OpenAI Useful as a blockout? Potentially.
But someone still needs to reconcile the dimensions with the grid, rebuild the modules, and clean up the geometry.
That substantially reduces the time saved. I wouldn’t confidently hand these assets over to production.
@OpenAI Confident completion reports made unresolved mistakes sound settled.
I’m not claiming deliberate intent. But the effect was that errors became harder to catch, and every “fixed” needed another inspection.
Supervising that behavior is exhausting.
@OpenAI It became a familiar dialogue between a junior artist and a supervisor.
Except corrections often added overlapping geometry instead of fixing the construction. The assembled scene reached roughly 12 million triangles while basic issues remained.
@OpenAI The most frustrating part was the confidence.
Uncertain details arrived as finished answers. I had to discover the guesses myself.
A simple “I can’t resolve this from the reference; can you guide me?” would have saved considerable supervision.
@OpenAI The overall result appeared at remarkable speed. From a distance, impressive.
Up close: wrong mouldings, missing curves, invented details, stairs on the wrong side. Basic shapes piled through each other to approximate forms without understanding their construction.
@OpenAI Galleria Subalpina is a great environment art test: interpret references, build reusable modules, adapt real-world dimensions to a game-friendly grid, model cleanly, and spend detail where players will see it.
My feed is full of “production-ready in five minutes” AI demos. Everyone seems eager to declare that computers have won.
I love exploring new tech and I’m excited about AI. But knowingly overselling unfinished work for engagement is disappointing.
In this week’s new game digest we’ve got the surprise return of Saints Row, news on incoming Overwatch, Palworld and Bleach games and plenty more: https://t.co/9H53sXi6lD