LTX-2.5-Diffusers

Quantised media packages of Lightricks/LTX-2.5-Diffusers.

MECHA Forge

MECHA — Model Engineering & Contextual Hearth Assembly

These media packages have been forged and optimized in the MECHA Forge.

Description

LTX 2.5: text/image-to-video generation. 40 steps, CFG 3.0. Elaborate English prompts.

Available quantizations

Quant	Denoiser	Text encoder	VAE
FP8	FP8	FP8	FP8
INT8	INT8	INT8	INT8
FP4	FP4	FP4	FP8
INT4	INT4	INT4	INT8

The VAE is shared between formats: FP4 reuses the FP8 file, INT4 reuses the INT8 file.

Components

In addition to the denoiser, text encoder, and VAE, the package includes:

• audio_vae, connectors, duration_head
• latent_upsampler, temporal_latent_upsampler
• prompt_enhancer, transformer_full
• vocoder, scheduler, processor, tokenizer

Recommended settings

From the source model card (adjust to taste):

Parameter	Value
Steps	30
CFG / Guidance scale	3.0
Resolution	under 720x1280, <=257 frames
Notes	Elaborate English prompts; negative prompt supported

Example prompts

Caught in the moment (candid, slice of life):

masterpiece, best quality, highres, absurdres, very aesthetic, safe, 1man, solo,
candid moment, natural lighting. A sunlit park bench in autumn, falling leaves,
a bicycle leaning nearby, warm afternoon light. A 29-year-old man with platinum
hair and blue eyes. He has an athletic build and a relaxed expression. He wears
a denim jacket and jeans, holding a camera, mid-smile, candid and relaxed.

Fashion photography (magazine cover style):

Fashion film still: a 29-year-old woman with platinum hair, sharp cheekbones,
and blue eyes; wearing a metallic silver gown, diamond earrings, stiletto heels;
studio spotlight, motion blur, cinematic, 24fps, 8K.

Scenery:

Scenery: a sweeping drone shot of a fjord at sunset, golden light on water,
mountains, clouds, boats, cinematic, 8K.