Training Krea 2 LoRAs to generate 360 panoramas took 16 hours and worked well
mickmumpitz · x · 2026-08-25
To generate 360 panoramas from text or single images, the author trained two LoRAs using Krea 2. One generates panoramas from text, while the other takes an image and outpaints the entire world around it. The training process took 10,500 steps and 16 hours on an RTX 6000 Pro, with excellent results reported.
Related event: Krea 2 LoRAs Generate 360 Panoramas Convertible to Gaussian Splat 3D Scenes(2 posts)→
More from Multimodal
- AI Series Mary Sue Episode 4 Made with CapCut and Seedance 2.5 — gen_ericai · 2026-08-26
- MAI-Image-2.6-Preview Hits Image Leaderboards, Neck-and-Neck with GPT Image 2 — ArtificialAnlys · 2026-08-26
- Microsoft MAI-Image-2.6 Tops Image Editing Leaderboard — ArtificialAnlys · 2026-08-26
- PatternEval and PatternRL: Aligning response patterns in hybrid-thinking MLLMs — YouJiacheng · 2026-08-26
- Higgsfield lands in Blender: prompt the scene, animate the camera, reblock in seconds — petewoodbridge · 2026-08-26
- Best current img2video solution for a 12GB VRAM rig — is WAN 2.2 still the answer? — FAUVEisEditing · 2026-08-25