RawSnap v3 LoKr released: retrained on 749 images, stackable with v2 for realism
Wide_Director_8897 · reddit · 2026-11-07
RawSnap v3 is out, retrained from scratch as a LoKr (not a standard LoRA) on an expanded dataset of 749 images across 11,235 steps and 15 epochs. The recommended setup is stacking: RawSnap v3 (epoch 15 or 12) at weight 1.2 plus RawSnap v2 at 0.35. Comparison images show v3 producing more natural skin tones and night-photo realism than v2. Available on Civitai.
Related event: RawSnap v3 LoKr retrained on 749 images, stacks with v2(2 posts)→
More from Multimodal
- Short film 'They Already Saw It' made with Midjourney, H3, Suno and Claude — techhalla · 2026-11-08
- AI-generated vintage patent blueprints of biplanes and airships look stunningly detailed — azed_ai · 2026-11-08
- Creator recreates signature hyperlapse audiovisual technique with AI — uisato · 2026-11-08
- MER&KEL releases AI-generated music video 'Darlig, You an I' — merkelmusic · 2026-11-08
- Filmmaker shares full workflow for a 1:42 AI dialogue scene: master voice track + timestamps — DavidmComfort · 2026-11-08
- Werewolf Image Generated Locally with Krea2 and MiniMax — -Ghosa- · 2026-11-08