I’ve just seen a video on YouTube about this cool AI tool located at: https://skybox.blockadelabs.com/ which allows you to generate skyboxes (not the regular ones used in BabylonJS), a simple jpeg per skybox. Code to visualize the skybox is included below. It’s just a huge sphere with an emissiveTexture and the camera is inside the sphere.
Only a few steps are needed:
- Go to https://skybox.blockadelabs.com/ and describe your skybox in the Dream up your world input box.
- Hit Generate and wait for the resulting jpeg.
- Download the JPEG.
- Upload it to your server.
- Change the path at line 10 to your jpeg:
const texture = new BABYLON.Texture('https://playgrounds.babylonjs.xyz/scifi_a_dying_star.jpeg')
in this playground code:
My description was “A dying star”. Cool! Isn’t it?!
Enjoy and don’t forget to share your creations with us here!
Credits for the YouTuber: Immersive AI Worlds 🌎 on your Website in Seconds - YouTube
I am listening to two steps before hell and just imagined an army of angels and devils fighting each other in hell/heaven.
Strength of a Thousand Men - Two Steps from Hell - YouTube
So my prompt was: fighting angles and devils in heaven
I did get something different from what was asked for, but a cool landscape. Awesome!
PS: No need to upload the image to your own server. The image is already on the amazon server
vr_cinematography_micro_detailed_vr_fighting_angles_devils_heaven__e57bbfda5a86d489__1674465_e5.jpg (6144×3072) (blockade-platform-production.s3.amazonaws.com)
Hey, cool song!
Yeah, the tool seems to be more focused on the world and the landscape and not on the details. But you can try to generate in two steps. The world and after that you can Remix it and just ask to “add angels”. I got this bad boy/girl
Good to know, thx!
Just keep it calm and relaxed buddy
@mawa I get it and I mainly agree with you, it’s not for pros but how on Earth could there be an example of usage for ThreeJS and not for our beloved BabylonJS? now we have too and I posted it into the comments of the youtube video as well…
HOWEVER: Stable Diffusion version?
I feel this goes to show how public Midjourney has such a narrow spectrum of CGI derivative output - doesn’t really respond to the text prompts and then gets boring quick. The good bit of AI image gen is liminal and hallucinatory - and also something for us humans to think on: get weird, but in ways the machines can’t.
im trying :
“10 million baobab trees with ring tailed lemurs and a few anthills with aardvarks”
I didnt get any aardvarks … or lemurs … I got … bambi and some other things
“sunny on one side, rainy on the other, covered in grapevines. A few steam punk robotic harvesters floating about picking the grapes”
it got the grapevines , and a normal harvester … nothing else
ok one more haha :
“planet completely under water , with megalodon sharks , sperm whales , and giant squid swimming around”
this one … has this one totally random fish … haha go look for it
This one is a fucking scary megalodon LOL
But I could squeeze the juice a bit more out of it:
Even more, look at that killer gangsta shark
car race on the moon of saturn
Someone did a good job with this tool
Yeah, this is all very fun and sometimes nice (from a distance) but I’m keeping with my say. Too many issues to be able to use this thing in productivity. At the best, you can use it as a tool/sider for creative thinking. But that’s about it, for me.
Here is my prompt for sharks. The dark fantasy style (color palette) is not respected. Obviously the AI doesn’t know how the mouth of a shark works. It also makes for some hallucinating distorsions and fancy integration of objects. Whereas, the lighting and color integration is quite convincing. But that’s not enough for me, I’m afraid to say
@mawa this could be your logo!