Yesterday's update was built from one player's review. This one is built from their follow-ups — four detailed notes sent within hours of v1.56's predecessor going live. That kind of feedback loop is a gift, so we turned it around in a day.
NEW: Line-break detection now understands your selected region. When you translate a chosen area — a chat box, typically — the app also treats the region's right edge as evidence: a left-aligned line that stops well short of it, where the next word would have fit, is a deliberate break and stays its own line. We calibrated this on real screenshots so ordinary wrapped paragraphs never get chopped up. One tip: draw the region reasonably close to the text column for best accuracy. The same on-off switch under Settings, Display applies.
NEW: The translation panel can run chat-style. A new option under Settings, Display appends fresh translations at the bottom — matching how most chat boxes flow — and auto-scrolls to follow. Off by default; the classic newest-on-top order is unchanged.
NEW: The overlay nudge now goes much further, and you can see it before you commit. The range grew from 200 to 2000 pixels in every direction, enough to park translations on the far side of the screen, and a small live preview next to the X and Y fields shows exactly where the translation box will land relative to the original.
NEW: Your glossary now reaches inside sentences. Until now, glossary entries only replaced short exact matches like button labels. When you translate with Ollama or LM Studio, your entries are now fed into the per-game sheet and win inside full sentences too. The glossary editor also gained a scrollbar and a note explaining exactly where entries apply. Honest caveat: we tested this for the high-quality AI engine as well, but its translation-only model kept leaking the term list into the output, so it stays off there for now — the editor says so.