Skip to content

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.

refactor(panel): pass 7 — Extend uses Q8 Draft / Q8 Pro labels (one vocab)

mrbizarro5d7aea615b9f
diff --git a/mlx_ltx_panel.py b/mlx_ltx_panel.py
index b1243b0..6a7c8f3 100644
--- a/mlx_ltx_panel.py
+++ b/mlx_ltx_panel.py
@@ -15965,14 +15965,30 @@ HTML = r"""<!doctype html>
</select>
</div>
</div>
- <label class="lbl" style="margin-top:10px">Extend mode</label>
- <div class="pill-group cols-2" id="extendModeGroup">
- <button type="button" class="pill-btn active" data-extend-mode="fast"><span>Fast</span><span class="sub">~16 min · 12 steps · 64 GB safe</span></button>
- <button type="button" class="pill-btn" data-extend-mode="quality"><span>Quality</span><span class="sub">~38 min · 30 steps · 96+ GB</span></button>
+ <!-- Pass 7 (Codex C+ 2026-05-17): renamed from "Fast / Quality"
+ to "Q8 Draft / Q8 Pro" so the labels match the Character
+ mode's quality vocabulary. The underlying mechanism stays
+ the same — Extend uses its own sampler (extend_steps +
+ extend_cfg) that isn't compatible with the main HQ res_2s
+ sampler's stage_1/stage_2 step model. Same labels make it
+ visually clear both intents live in the Q8 family without
+ forcing the implementation onto the HQ sampler. -->
+ <label class="lbl" style="margin-top:10px">Quality</label>
+ <div class="pill-group cols-2 quality-strip" id="extendModeGroup">
+ <button type="button" class="pill-btn pill-quality active" data-extend-mode="fast">
+ <span class="ql-name">Q8 Draft</span>
+ <span class="q-spec ql-spec sub">12 steps · 64 GB safe</span>
+ <span class="ql-tier">Q8 extend · ~16 min</span>
+ </button>
+ <button type="button" class="pill-btn pill-quality" data-extend-mode="quality">
+ <span class="ql-name">Q8 Pro</span>
+ <span class="q-spec ql-spec sub">30 steps · 96+ GB</span>
+ <span class="ql-tier">Q8 extend · ~38 min · reference</span>
+ </button>
</div>
<input type="hidden" name="extend_steps" id="extend_steps" value="12">
<input type="hidden" name="extend_cfg" id="extend_cfg" value="1.0">
- <div class="hint">Each latent ≈ 8 frames (~0.33s). Quality mode runs the upstream defaults but pushes 1280×704 into swap on 64 GB Macs (~2 hr/render). Stick with Fast unless you've got more RAM.</div>
+ <div class="hint">Each latent ≈ 8 frames (~0.33s). Q8 Pro runs the upstream Lightricks defaults but pushes 1280×704 into swap on 64 GB Macs (~2 hr/render). Stick with Q8 Draft unless you've got more RAM.</div>
</div>
</div>