LTX 2.3 multi-image first/last frame at 20 images, plus IC-LoRA Union video control
Two graphs, one clear verdict. The headline is the multi-image graph: LTX 2.3 now accepts up to 20 guide images, and the jump in consistency and motion smoothness over the 2.0 generation is large enough that the author calls it a different class. The second graph, IC-LoRA Union control, works but does not impress him the same way. Evidence note: all three graphs ship as JSON and are parsed below.…
The multi-image graph for LTX 2.3 now supports up to 20 guide images, resulting in a significant improvement in consistency and motion smoothness compared to the 2.0 generation. The author calls this a distinct class. The IC-LoRA Union control graph works but does not impress as much as the multi-image graph. The author provides several key findings, including the frame-preparation path with four ImageScaleBy nodes at 0.5 and one at 0.67, and the use of a 0.67 first-stage scale.
The video control graph covers three control types, including depth, Canny, and DW Pose, and advises against large visual differences between the control subject and the target. The LTX 2 distilled lip-sync graph is recommended for digital humans, as the 2.3 digital-human graph does not match its results.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.