Legible text on a sign
Five words of neon have to stay readable through a push-in while something moves in front of them.
2 models received different prompt text — show each
Each model is given the prompt its input contract calls for: a multi-reference model is told which reference is which character, a segmented model gets shot segments. The frame, the references and the requested length are identical throughout. This is the honest version of "the same prompt".
H3 Max, H3 ref2v, Wan 2.7, Seedance 2.0, LTX-2
Slow push in toward the diner. The neon sign reading OPEN ALL NIGHT stays sharp and legible the whole time. A cyclist rides through the foreground from right to left. Rain keeps falling and the reflections shimmer. Ambient rain and distant traffic, no music.
Kling v3 Pro
[{"prompt":"Slow push in toward the diner. The neon sign reading OPEN ALL NIGHT stays sharp and legible the whole time. A cyclist rides through the foreground from right to left. Rain keeps falling and the reflections shimmer. Ambient rain and distant traffic, no music.","duration":"5"}]| H3 Max | H3 ref2v | Kling v3 Pro | Wan 2.7 | Seedance 2.0 | LTX-2 | |
|---|---|---|---|---|---|---|
| Lines heard | — | — | — | — | — | — |
| Voice | model's own | model's own | model's own | model's own | model's own | model's own |
| Motion | 3.44 | 2.04 | 2.37 | 1.62 | 2.58 | 3.33 |
| Scene changes | 0 | 0 | 0 | 0 | 0 | 0 |
| Loudness / peak | -23.8 / -12.8 | -24.9 / -13.2 | -25.4 / -3.9 | -43.2 / -23.6 | -30.9 / -14.3 | -26.9 / -4.8 |
| Length | 5.18s | 5.18s | 5.04s | 5.04s | 5.06s | 4.92s |
| Frame | 768×1344 | 768×1344 | 1072×1928 | 720×1280 | 720×1280 | 576×1024 |
| Wait | 6s | 3:34 | 3:16 | 2:19 | 3:06 | 47s |
| Attempts | 1 | 1 | 1 | 1 | 1 | 1 |
| Cost | $0.40 | $0.30 | $0.98 | $0.50 | $0.90 | $0.22 |
The read
Every model kept OPEN ALL NIGHT legible for the full push-in, which is better than the folk wisdom about text in video models. The differences were in the cyclist: H3 Max, H3 reference-to-video, Kling and Seedance all put a rider through the foreground on cue; Wan's arrived late and dim; LTX's was a dark smear at the bottom of the frame. H3 Max was again the most directed take — lamp glow, window detail, rain — in 6 seconds. Seedance's mix sits at −31 LUFS, quiet enough to read as silence next to the others.
One person's judgment from the takes above, 2026-09-03. The table is the evidence; this is the opinion.
what whisper heard
- H3 Max: [BLANK_AUDIO]
- H3 ref2v: [BLANK_AUDIO]
- Kling v3 Pro: (air whooshing)
- Wan 2.7: [BLANK_AUDIO]
- Seedance 2.0: [BLANK_AUDIO]
- LTX-2: (fire crackling)