Dreamzo 0.1.1

What's New

• Omni mode: reference uploads now support @-mentions with stable numbering (references never drift), and generation always uses the latest prompt
• Export is now a dropdown: full video, or all shots bundled as a zip
• Vendor connectivity test now uses structured error codes, with per-model probing and API key format validation
• "Contact Us" surfaced as a global floating button (bottom-right) for better discoverability

Fixes

• qtum video: reference media is uploaded and swapped for a file-id first, bypassing the /videos body-size 502 wall
• JSON endpoints are now no-store, fixing "deleted items only disappear after a full refresh"; delete robustness aligned
• Project list not refreshing in real time (silent delete failures + no refetch on entering the list)
• Streaming TTS voiceover duration halved (audio_duration was reporting half the real value)
• Omni shot planning now feeds the prop list to the LLM, fixing prop reference-image association failures
• No longer shows a false "unsaved changes" prompt after deleting a shot

Other

• Removed decorative diagonal stripes throughout

Dreamzo 0.1.0

docs: add bilingual provider setup guides

Add English (default) and Chinese setup docs for four providers —
Qtum.AI, Volcengine Ark, Doubao TTS, and DeepSeek — covering required
credentials, key-creation steps, and screenshots. Includes a README
index and shared images directory.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>