🚫[DEPR] Motion Model Settings (Simple) 🎭🅐🅓①

Another one marked 🚫[DEPR] right in its display name - Motion Model Settings (Simple) is a leftover from before AnimateDiff-Evolved built out its current PE Adjust / Weight Adjust node families. If you've opened an older workflow and found this attached to a motion model loader, it still works; it's just not what the pack wants you building new graphs around.

What it does

Before positional-encoding remapping (ADE_AdjustPEFullStretch and friends) and per-category weight scaling (ADE_AdjustWeightAllMult and friends) existed as their own dedicated node family, this single node was the entry point for tweaking a motion model's behavior:

• motion_pe_stretch (int, default 0) - a single-number version of the positional-encoding stretching that the current PE Adjust nodes now handle with far more control (three separate strategies: Full Stretch, Manual, Sweetspot).
• mask_motion_scale (optional MASK) - lets you scale the motion module's influence spatially, so different regions of the frame get more or less motion. This is the one piece of functionality here that isn't obviously superseded by a newer node in this brief's set - masking motion scale by region is a genuinely useful, distinct capability.
• min_motion_scale / max_motion_scale (floats, default 1 each) - the range the mask value gets scaled into.

Output is AD_SETTINGS, wired into your motion model loader the same way PE_ADJUST and WEIGHT_ADJUST wire into the newer settings system.

What to use instead

For PE stretching specifically, the current pack gives you three purpose-built nodes instead of one blunt number: ADE_AdjustPEFullStretch, ADE_AdjustPEManual, and ADE_AdjustPESweetspotStretch, each a different strategy for remapping positional encoding rather than one undifferentiated stretch value. For anything about scaling the motion module's overall strength rather than masking it spatially, the scale_multival/effect_multival inputs on the Gen2 Apply nodes are the current, more flexible path - those accept a float, a list of floats, or a mask, per the README's description of multival inputs, which covers the spatial-masking use case this node's mask_motion_scale was built for.

Installing the pack

If this node showed up in a workflow you're trying to run, you still need the pack: ComfyUI Manager - search AnimateDiff Evolved, confirm the author is Kosinkadink, install, restart. Manually:

cd ComfyUI/custom_nodes
git clone https://github.com/Kosinkadink/ComfyUI-AnimateDiff-Evolved

restart. You'll need a motion module downloaded - mm_sd_v15_v2, v3_sd15_mm, and the rest listed in the README - into ComfyUI/models/animatediff_models for anything to attach to.

Common issues

It still works, but building new workflows around it means missing out. The PE Adjust family gives you real control over quality at extended context lengths that a single motion_pe_stretch number can't match. If you're actively tuning long-clip quality, migrating is worth the effort.

The mask-based motion scaling is genuinely still useful - it's just this specific node doing it. If all you want from this node is spatial motion masking (more motion in one part of the frame, less elsewhere), the concept carries over cleanly to scale_multival/effect_multival on the current Apply nodes, which accept a MASK input the same way.

Old workflow, unexpected results after a pack update. Deprecated nodes get less testing attention than the current node family, so if behavior seems to have drifted since you last used it, that's a real possibility - worth rebuilding the settings on the current PE Adjust / multival system rather than debugging the legacy path.