Reddit: chasing natural AI selfies with Krea 2 Turbo + LoRAs in Forge NEO
Dandypleasure · reddit · 2026-10-09
A Reddit user shares their local image workflow and asks for advice: the goal is daily selfie-like images with natural lighting, posing and surroundings. Current setup is Krea 2 Turbo with two LoRAs in Forge NEO.
- Asks whether upscaling, ADetailers or double sampling are needed for more realism;
- Wants to know if Krea 2 Turbo is still the best option vs Z Image Turbo or Qwen 2.1 for photorealistic people;
- Looking for sites to track new local models and LoRA releases.
More from Multimodal
- Creator Makes Animation With Claude Motion That Would Have Cost Tens of Thousands a Year Ago — socialwithaayan · 2026-10-09
- Claude Motion's First Night: 9 Wild Motion-Design Examples Built With Code — socialwithaayan · 2026-10-09
- Nano Banana 2.1 lands in Leonardo with better prompt adherence — aziz4ai · 2026-10-09
- ORCA Paper: Single Training-Only Loss Fixes Text-to-Image Composition Failures — mo_lotfollahi · 2026-10-09
- Creator lists 10 video editing tasks AI did this week, cutting 34 min of footage to 16 min — toolstelegraph · 2026-10-09
- Creator ships product video with 4 prompts in 2 hours by reusing old projects in HyperFrames — toolstelegraph · 2026-10-09