25-second renders, with and without temporal position interpolation

.gitattributes

@@ -118,3 +118,7 @@ samples/sharp/03-road.mp4 filter=lfs diff=lfs merge=lfs -text
 samples/sharp/04-smoke.mp4 filter=lfs diff=lfs merge=lfs -text
 samples/sharp/05-idol-15s.mp4 filter=lfs diff=lfs merge=lfs -text
 samples/sharp/sheet.png filter=lfs diff=lfs merge=lfs -text
+samples/long/ab.png filter=lfs diff=lfs merge=lfs -text
+samples/long/crop_ab.png filter=lfs diff=lfs merge=lfs -text
+samples/long/idol25-fit.mp4 filter=lfs diff=lfs merge=lfs -text
+samples/long/idol25-nofit.mp4 filter=lfs diff=lfs merge=lfs -text

samples/long/README.md

@@ -0,0 +1,66 @@
+# Past 20 seconds: what the temporal RoPE does, and what to do about it
+
+## There is a hard number, and it is in the config
+
+`LTXAVModel` carries `positional_embedding_max_pos=[20, 2048, 2048]`, and
+`get_fractional_positions` divides each coordinate by its entry. The temporal
+one is **seconds, not frames** — both `_prepare_positional_embeddings`
+(`model.py:841`) and the AV cross-attention path (`av_model.py:865`) multiply
+the temporal coordinate by `1 / frame_rate` first.
+
+So the temporal rotary embedding was built for **20 seconds**. Past that, the
+fraction exceeds 1.0 and the frequencies extrapolate off the end of anything the
+model saw in training. This is the failure
+[Train Short, Inference Long (arXiv 2602.14027)](https://arxiv.org/html/2602.14027v1)
+attributes long-horizon degradation to.
+
+`fit_temporal_positions` raises `max_pos[0]` to the clip's own duration, mapping
+it back onto [0, 1] — position interpolation, the same trick used to extend a
+language model's context, and training-free. It is safe to patch at runtime
+because `generate_freq_grid_np`'s cache is keyed on theta, dimension count and
+device, never on `max_pos`.
+
+## Measured at 25 seconds
+
+601 frames, 1024x640, two-pass, same prompt and seed, on the Space's H200.
+
+| | segment sharpness (5 equal parts) | last/first |
+|---|---|---:|
+| untreated | 0.00086 0.00130 0.00143 0.00139 0.00096 | 1.12 |
+| **interpolated** | **0.00247 0.00302 0.00286 0.00318 0.00198** | 0.80 |
+
+The interpolated render is **2.2-2.9x sharper in every segment**.
+`crop_ab.png` at 1:1: hair strands, eyelashes and background props are resolved
+on the treated side and waxy on the untreated one.
+
+## What this did not find
+
+**No progressive collapse.** The premise being tested was that quality decays as
+the clip runs on. It does not, in either version — the last-fifth to first-fifth
+sharpness ratio is 1.12 untreated and 0.80 treated, and neither drifts in
+character, wardrobe or room across 25 seconds. Most of the long-video literature
+found in this search — compounding error, history forgetting, attribute drift —
+describes **autoregressive or sliding-window** generation, where each window
+conditions on the last. LTX-2.5 as driven here samples the whole clip at once,
+so that class of failure does not apply and searching for it would have been
+looking in the wrong place.
+
+What extrapolation costs here is **detail across the whole clip**, not stability
+at the end.
+
+## Limits
+
+* The two renders are different trajectories, not the same trajectory rendered
+  twice, so the sharpness gap is what changed and not a controlled isolation of
+  cause. Same prompt, seed, size, sampler and refine strength; only `max_pos[0]`
+  differs.
+* One duration (25 s), one prompt, one seed. 30 s, 40 s and beyond were not
+  tried, and the point where interpolation itself starts to hurt — it compresses
+  temporal detail by construction — was not located.
+* Below 20 s this changes nothing and is not applied.
+* Laplacian variance tracks contrast as well as sharpness. It is used here
+  between two renders of the same scene, which is the only place it is fair.
+
+## Files
+
+`idol25-fit.mp4`, `idol25-nofit.mp4`, `ab.png` (timeline), `crop_ab.png` (1:1).

samples/long/idol25-fit.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eb69e83909f4ef5b5308d13c4b431b001579d206b4dae902996d8931e605a18c
+size 16806661

samples/long/idol25-nofit.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a31a7b8539f709ef18cac15d518e90dc3de29cc3a8f908bc88c6cb380b550b3a
+size 16186038