Krea users say some character LoRAs make image quality noticeably worse
corbzarim · reddit · 2026-07-27
A Krea user is asking whether character LoRAs are degrading output quality over time.
They report the same workflow producing progressively worse blotchy details, especially in hair and jewelry, and note that the artifacts are often blamed on the VAE. But in their case, some LoRAs seem to worsen the issue more than others, even without longer training runs. The question is whether this degradation is real, and whether there are known fixes.
More from Multimodal
- Third-party test: Claude Opus 5.5 renders finer 3D scenes but costs 13x more than GPT-6 Sol — testingcatalog · 2026-09-23
- ComfyUI trick: aux preprocessor + Qwen transfers poses across characters with one prompt — Acceptable-Work8202 · 2026-09-23
- Same portrait prompt across Midjourney V6.1, V7 and V8.2: do older models look better? — tisch_eins · 2026-09-23
- Testing AI character consistency across a 20-image travel sequence — SiennaVaire · 2026-09-23
- Midjourney v8.2 Faces: New Portrait Generation Samples Shared — azed_ai · 2026-09-23
- One-sentence prompt generates lifelike dog video, shown side-by-side with the real one — wgrathwohl · 2026-09-23