10Eros i2v/v2v.
All nodes colored red (not vae) are adjustment points to finetune output. Extremely tempermental SDE sampling at highnoise. Produces good motion and anatomy with lots of ways to tweak the seed. Similiar to the triple sampling setups, LTX2.3 just needs CFG and more samples. Swap between er_sde and dpm_pp. Dpm is prone to audio problems with higher distilled strength and using the CFG on first steps. Needs to be balanced carefully.
Custom nodes you probably need: https://github.com/chrisgoringe/cg-sigmas
LTX2 simple i2v.
Uses res2s and usually lower distill strength with fp8 undistilled model, 0.75-0.9 distilled lora (rank175 seems best) strength depending on application for first pass. Stack as many loras as you can even if barely related to the concept, lora stack drowns out the base model noise and makes the output more stable.
Use 33-38 preprocess node compression strength to increase motion. Get the best motion and audio from the first pass, cancel and reroll seed if bad. If preview is kind of good stop it and refine with prompt and lora weight mixes.
Second pass uses audio directly from first seed to track and half strength distilled upscale pass based on the full size input image for max quality. Only way to get very good clear visuals is with the half distill, but passing audio latent into the half distilled sampler ruins it, so this is the neat trick.
If you wanna use this T2V, just click the bypass on the LTXVImgToVideoInplace node.