LTX-2.3 Surface Realism LoRA

Kills the plastic. Every surface earns its age.

LTX-2.3 (and the image models feeding it first frames) defaults to clean, smooth, showroom-new surfaces: pristine asphalt, unmarked walls, grounds that read as rendered geometry. This LoRA is an always-on materials prior: cracked wet asphalt stays cracked and wet, concrete gets pores and rust streaks, floors scuff, weeds mat, wood grains. It pushes every render toward surfaces that have existed in weather.

No trigger wordAudio-safe1,152 tensors video only24 fps recommended

No trigger word — by design

The training captions name only material, condition, framing, and light — they deliberately contain zero realism vocabulary ("photoreal", "detailed", "8k" never appear). The realism therefore trains as unconditional LoRA behavior instead of a phrase you have to remember. Load it at 1.0 and write your prompts normally; concrete material nouns ("scuffed linoleum", "rust-streaked concrete") still help composition, as they always did.

Voice / audio safe by construction

1,152 LoRA tensors, all in the video branches (attn1 / attn2 / ff). Zero audio-touching tensors: dialogue, voices, and accents render byte-identical with or without this LoRA. Verified alongside an audio-branch accent LoRA in the same stack with no interference.

Video-branch tensors1,152Audio-branch tensors0Cross-LoRA interferenceNone detected

Usage

• Strength 1.0 in any LTX-2.3 LoRA loader.

• Compatible with the LTX-2.3 22B family: stock distilled 1.1 and the JoyAI-Echo surgical merges.

• If you render dialogue: keep your workflow at 24 fps — not for this LoRA's sake, but because off-24 fps drifts every LTX voice toward British/Australian regardless of prompt.

Training

ai-toolkit on an RTX 3090, rank 32 / alpha 32, 3,000 steps, lr 1e-4, qfloat8. Base: stock ltx-2.3-22b-distilled-1.1. Data: 1,076 hand-curated surface plates (from a 1,900+ candidate pool) generated with Z-Image Turbo specifically for this training — grounds, walls, floors, and weathered materials under varied light. Every image is self-generated; no third-party photography. The module scope excludes every audio branch, which is what guarantees the audio lane stays untouched.

Rank / Alpha

32 / 32

Steps

3,000

LR

1e-4

Precision

qfloat8

Data Plates

1,076

GPU

RTX 3090

Also available on

  HuggingFace: joeygambino/ltx23-surfaces-realism-lora

Support this work

Everything I publish — workflows, merges, quants, patches — is free and stays free. The models are non-commercial by license and I have no plans to change that.

If something here saved you an evening:

• GitHub Sponsors — recurring, no platform cut

• Ko-fi — one-off, no account needed

• Liberapay — recurring, open-source and non-profit

Bug reports are worth as much as money. Most of the fixes in these packages exist because somebody posted a log.