WanImageMotionPro Legacy

IAMCCS_WanImageMotionProLegacy ("WanImageMotionPro Legacy") is the earlier generation of IAMCCS's end-frame-lock motion node, kept in the pack so older workflows keep loading. If you're starting fresh today, you almost certainly want the newer WanImageMotionPro or the trimmed Plus Simple instead - this one exists for compatibility, not because it's better.

That's not to say it's broken. The Legacy variant already has the two headline features of the family: motion amplitude boost on the prev-sample tail, and FLF-style end-frame locking via use_end_frame, end_lock_slots, and end_overshoot_slots with a trim_slots output. The shared core is all here: motion (1.0–2.0), motion_mode, motion_latent_count, safety_preset (base/safe/safer/legacy), lock_start_slots, vram_profile, and the diagnostic_log toggle.

What it's missing

Compare the schema with the current Pro and the differences are obvious. There's no clip_vision_mode routing (the fix for SVI-Pro crossfade bleed when anchor and end are the same image), no latent_refresh/delta_max AdaIN stabilization, no prev_samples_profile/prev_tail_pick_mode/prev_anchor_blend for handling ragged SVI tails, and no raw_mode parity switch. It's the 1.3.5-era feature set - good enough for straightforward A→B chunks, but it can't handle the messy continuation cases the modern node was built for.

The end_transition_frames input is also here, deprecated with the same warning as everywhere else in this family: keep it at 0, because blending latents toward the end frame in latent space produces frozen, static output. The VAE latent space just doesn't interpolate linearly.

When to use it (honestly)

Three scenarios. You loaded an old workflow and don't want to rewire - fine, it'll run. You need exact bit-for-bit behavior of an older run for comparison. Or you're deliberately avoiding the newer node's added complexity and want the small, predictable surface. Otherwise, pick the current node.

Outputs are the family standard: positive, negative, latent, and trim_slots for cutting the overshoot tail. Same end_overshoot_slots warning applies - if you have KJNodes FETA/wrapped-attention WAN optimizations active, overshoot > 0 can crash with an einops shape mismatch; set it to 0 or update KJNodes.

Install

Same pack, same routine:

cd ComfyUI/custom_nodes
git clone https://github.com/IAMCCS/IAMCCS-nodes.git

or ComfyUI Manager → search "IAMCCS" → restart. No models, no extra deps. Compatibility floor: ComfyUI ≥ 0.3.0, Python ≥ 3.12, Torch ≥ 2.8.

The practical takeaway: keep the pack updated, let legacy nodes retire in your saved graphs, and only reach for this one when a workflow literally references it.