Nodes/LTExperiments/LTEx LTXVAddGuide (Cond Only)
ComfyUI Node

LTEx LTXVAddGuide (Cond Only)

Proof of concept: carry an LTXV / LTX-2 guide on the conditioning instead of appending it to the working latent (the LTXVAddGuide approach). The guide is injected as extra tokens inside the model forward and stripped before output, so no LTXVCropGuides node is needed and the latent is never modified. Works for LTXV and LTXAV (video stream). Installs a gated monkeypatch of the LTX forward that is a no-op unless this node is used. Drop-in alternative to LTXVAddGuide for A/B testing the conditioning-carried path.

By drozbay·Created 2 months ago·Updated 2 months ago· 5
LTEx LTXVAddGuide (Cond Only)
  • positive
  • negative
  • vae
  • latent
  • image
  • attention_mask
  • iclora_parameters
  • positive
  • negative
  • latent_passthrough
frame_idx0
strength1.00
CategoryLTExperiments

Inputs (9)

NameTypeDefaultDescription
positiveCONDITIONING
negativeCONDITIONING
vaeVAE
latentLATENTThe latent being sampled. Used only to size the guide encode; not modified.
imageIMAGEGuide image or video. Must be 8*n + 1 frames or it is cropped to the nearest.
frame_idxINT0-16384–16384Pixel-frame index to place the guide at. For 9+ frame guides, frame_idx must be divisible by 8 (rounded down). Negative values count from the end.
strengthFLOAT1.000–1How strongly the guide pins its frames. 1.0 locks them (clean), lower loosens; 0 ignores the guide.
attention_maskoptMASKOptional pixel-space spatial mask. Controls per-region guide influence via self-attention, multiplied by strength. Same semantics as LTXVAddGuide.
iclora_parametersoptIC_LORA_PARAMETERSOptional IC-LoRA parameters (e.g. reference_downscale_factor > 1) from a Get IC-LoRA Parameters node. The low-res reference is spread across the full-res grid via dilated RoPE coordinates; the latent is still untouched.

Outputs (3)

NameTypeDescription
positiveCONDITIONING
negativeCONDITIONING
latent_passthroughLATENTInput latent, returned unmodified