Free, open-source AI video generator for Mac. Run LTX-2.5 and MiniMax H3 (Hailuo) locally on Apple Silicon: text-to-video and image-to-video with sound, lip-sync, character LoRA training, songs from lyrics (YuE2) and image editing (Qwen-Image-Edit). One-click install via Pinokio. Offline: no cloud, no API key, no NVIDIA GPU.
Python ai-filmmakingai-videoai-video-generatorapple-siliconhailuoimage-to-videolip-synclocal-ailora-trainingltx-2ltx-2-5ltx-videomacosminimax-h3mlxmusic-generationoffline-aipinokiotext-to-videovideo-generation
fix(load-params): restore seed_used in Character-mode loadParams
mrbizarro5831417467d9
diff --git a/mlx_ltx_panel.py b/mlx_ltx_panel.pyindex 498b172..9c90ac2 100644--- a/mlx_ltx_panel.py+++ b/mlx_ltx_panel.py@@ -20746,10 +20746,19 @@ async function charactersLoadParams(p) {if (durInp) durInp.value = durSec;}- // Restore seed if not -1 (random).- if (typeof p.seed !== 'undefined' && String(p.seed) !== '-1') {+ // Restore seed — prefer seed_used (the integer the helper actually+ // picked at generation time) over seed (often '-1' when the user let+ // the panel randomize). Without this preference, Load Params on a+ // -1 submission restored -1 and the next render got a fresh random+ // seed instead of reproducing the source clip. Mirrors the Manual+ // loadParams fix (commit b024bb5, 2026-05-18); the Character branch+ // was missed in that pass and reported by Mr Bizarro on the panel.+ const _seedRaw = (p.seed_used != null && String(p.seed_used) !== '' && String(p.seed_used) !== '-1')+ ? p.seed_used+ : p.seed;+ if (typeof _seedRaw !== 'undefined' && _seedRaw !== null && String(_seedRaw) !== '-1') {const seedInp = document.getElementById('seed');- if (seedInp) seedInp.value = String(p.seed);+ if (seedInp) seedInp.value = String(_seedRaw);}// Prompt textarea — verbatim. Use prompt_body if present (legacy