Which LoRAs work with LTX-2 19B LoRA?

Files trained for LTX-2 19B that are under 2 GiB, such as Lightricks' camera-control LoRAs (Dolly In, Dolly Out, Dolly Left, 327 MB each) and community style LoRAs. Pass a direct HTTPS link to the .safetensors file in the {path, scale} list, up to three per request.

Why was my LoRA request rejected?

Usually one of four reasons: the file is over 2 GiB (2,147,483,648 bytes), the link needs a login (HTTP 401), the link is a repository page instead of the file, or the list holds four LoRAs, which is rejected at submit. The first three come back as “Failed to load LoRA” and none of them is charged.

What happens with a LoRA for another model?

It loads without an error and does nothing visible. A Wan 2.1 LoRA changed neither composition, wardrobe nor motion in our test, so run a with-and-without pair before you tune.

What scale should I use?

Start at 1 for camera LoRAs. For the anime-style LoRA, 0.5 already gave a clearly hand-drawn look, and 1 was stronger, with the set and characters drifting further from the prompt. The allowed range is 0 to 4.

Can I use uncensored LoRAs?

Yes. The weights are not reviewed and nothing filters the output. Everyone on screen must be an adult, and rights and consent stay with you. In our tests an anime-style LoRA also toned down our strongest probe.

Do LoRAs cost extra, and is the audio included?

No separate charge for loading LoRAs or for how many you load, and every clip includes a stereo soundtrack generated with the picture. Current prices are on the model page.

LTX-2 19B LoRA or plain LTX-2 19B?

Pick the LoRA build when you have a camera or style file to apply. Pick LTX-2 19B when the base look is enough, since it is the simpler request and has no LoRA field. Both make 5 to 20 second clips with sound.

Does a LoRA change image-to-video?

Yes. In our test an anime-style LoRA at 1 turned a photo into hand-drawn animation within the first second. Camera LoRAs keep the photo's look and change the move.

LTX-2 19B LoRA API: Uncensored NSFW Video, Your LoRAs

lightricks/ltx-2-19b-lora/text-to-video

LTX-2 19B LoRA text-to-video loads up to three of your own LoRA files on Lightricks' audio-video model and returns a 5 to 20 second clip with stereo sound at 480p, 720p or 1080p. Camera moves, styles and characters by direct link, no content filter.

From $0.018/s

Run model View API

Input

Text to video

Prompt

Aspect ratio

Resolutionaffects price

Durationaffects price

Style LoRAs1/3

Camera Control Dolly In

Seed

Safety checkerOff

SpicyAPI does not run a safety checker on this model. Your prompt and media go to the model as written, and the model itself decides what comes back.

Checking your account before enabling submission and pricing.

Output

Ready for a new run

Example result

Extend

Video upscaling

Video editing

{
  "model": "lightricks/ltx-2-19b-lora/text-to-video",
  "input": {
    "aspect_ratio": "16:9",
    "resolution": "720p",
    "duration_seconds": 5,
    "loras": [
      {
        "path": "https://huggingface.co/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-In/resolve/main/ltx-2-19b-lora-camera-control-dolly-in.safetensors",
        "scale": 1
      }
    ],
    "seed": 520527,
    "prompt": "Night inside the round chamber of an old observatory: a giant white telescope points up through the open slit of the dome at a field of stars, red lamps glow low on the curved riveted walls. An adult South Asian woman in her mid-thirties with long black hair, in a black satin slip dress under a loose grey wool cardigan, leans against the telescope pier, lifts her eye from the eyepiece and turns to the camera with a slow, amused smile. The camera dollies in from the doorway to a close-up of her face. Deep red and starlight blue, haze, anamorphic, film grain. Audio: the dome motor rumbling as the dome turns, the steel shutter clanking, wind whistling through the slit."
  }
}

Introduction

LTX-2 19B LoRA Text to video generates a clip with its own soundtrack from a prompt, with up to three of your LoRA files loaded. It runs Lightricks' open 19-billion-parameter audio-video model; clips run 5 to 20 seconds at 480p, 720p or 1080p in 16:9 or 9:16. The platform adds no content filter to the prompt, the weights or the clip.

Choose a mode by starting material

What you have	Mode
a written description	LTX-2 19B LoRA Text to video
a photo to animate	LTX-2 19B LoRA Image to video
no LoRA at all	LTX-2 19B Text to video

Key features & highlights

Official camera LoRAs

Lightricks publishes Dolly In, Dolly Out and Dolly Left LoRAs for this model; each is a few hundred megabytes and loads by link.

Community styles

In our test an anime-style LoRA turned a photoreal prompt into cel animation at scale 0.5.

Three at once

Stack a style and a camera move; in our test the style held with the camera LoRA loaded.

Uncensored with custom weights

LoRAs are not reviewed and the output is not filtered.

Request parameters

Request parameters for LTX-2 19B LoRA · Text to Video
Parameter	Required	Description
`prompt`	Yes	Trigger words first if your LoRA has them, then the shot and where a camera LoRA should end up, then an "Audio:" line. at least 1 characters
`aspect_ratio`	No	16:9 for landscape (the default) or 9:16 for portrait. The shape does not change the price. `16:9` · `9:16` · Default: `16:9`
`resolution`	No	480p (832 x 448), 720p (1280 x 704, the default) or 1080p (1920 x 1024). Each step costs more per second. `480p` · `720p` · `1080p` · Default: `720p`
`duration_seconds`	No	Any whole second from 5 to 20, billed per second. `≥ 5` · `≤ 20` · Default: `5`
`loras`	No	Up to three {path, scale} entries trained for LTX-2 19B; each file under 2 GiB, scale 0 to 4 (1 for camera LoRAs, start style LoRAs at 0.5). Default: `[]` at most 3 items
`seed`	No	Keeps runs comparable while you change one LoRA scale at a time. It does not promise a frame-identical clip. `≥ 0` · `≤ 2147483647`

Pricing

Billed per second of output at the chosen resolution. Loading LoRAs adds no separate charge, and the aspect ratio does not change the price.

Pricing
Specification	Price
`480p`	$0.018 / second
`720p`	$0.024 / second
`1080p`	$0.036 / second

Use cases

Directed shots

Reveal, push-in and tracking moves you can rely on across a series.

Styled series

One style LoRA, many prompts, one consistent look.

Apps without a content filter

Load the LoRA your audience expects; adults only on screen.

Explore key use-case tutorials

Prompt guide & pro tips

Mind the 2 GiB limit

Larger files fail; many LTX-2 LoRAs on Hugging Face are 3.4 GB exports, so look for a smaller one.

Name the end of the move

Camera LoRAs work best when the prompt says where the camera ends up.

Compare against no LoRA

Same prompt, same seed, loras empty: if nothing changes, the LoRA was made for another base model.

Use direct, public links

The link must download the .safetensors file itself; a page link or a gated file fails and is not charged.

API quick start

# Generate the idempotency key once; reuse it when retrying after a lost response.
export SPICY_IDEMPOTENCY_KEY="${SPICY_IDEMPOTENCY_KEY:-$(uuidgen)}"
curl -X POST https://api.spicyapi.ai/api/v1/jobs/createTask \
  -H "Authorization: Bearer $SPICY_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: $SPICY_IDEMPOTENCY_KEY" \
  -d '{
  "model": "lightricks/ltx-2-19b-lora/text-to-video",
  "input": {
    "aspect_ratio": "16:9",
    "duration_seconds": 5,
    "prompt": "A slow aerial shot drifting over a misty pine forest at sunrise",
    "resolution": "720p"
  }
}'

# Copy data.taskId from the response into TASK_ID, then poll the status.
TASK_ID="TASK_ID_FROM_RESPONSE"
curl "https://api.spicyapi.ai/api/v1/jobs/recordInfo?taskId=$TASK_ID" \
  -H "Authorization: Bearer $SPICY_API_KEY"

See the API reference for authentication, the complete schema, polling, and error handling.

Related models and next tasks

Missing the model you need?

Tell us what to add and where you would use it. Requests go to the people who decide what gets onboarded next.

Direct the camera and set the style with your own LoRA files

LTX-2 19B LoRA loads up to three LoRA files per request onto Lightricks' open audio-video model, on text-to-video and image-to-video, so you can add a camera move or a style that the prompt alone does not deliver reliably. In our tests the official Dolly In, Dolly Out and Dolly Left camera LoRAs loaded at scale 1, a community anime-style LoRA turned a photoreal prompt into hand-drawn animation at scale 0.5, and the two stacked without losing the style. Clips run 5 to 20 seconds at 480p to 1080p with stereo sound, and nothing filters the prompt, the weights or the result.

Open the playground

Direct the camera and set the style with your own LoRA files

LoRAs per request: Up to 3, each {path, scale}; scale 0 to 4, default 1; a fourth is rejected at submit
File limit: Each LoRA file must be under 2 GiB (2,147,483,648 bytes) and a direct download link
Clip length: Any whole second from 5 to 20, at 480p, 720p or 1080p
Audio: Stereo AAC at 24 kHz on every clip; there is no switch to turn it off
Speed: 5 s at 720p took 14 to 25 seconds of model time in our runs on this model
Content filter: None; the LoRA files are not reviewed either

Camera: Make the move reliableOfficial camera-control LoRAs from Lightricks (Dolly In, Dolly Out, Dolly Left) are 327 MB each and loaded at scale 1 in our runs. Describe where the camera should end up, such as “dolly in until her face fills the frame”, and the LoRA does the move.
Style: One file restyles the whole clipWith the same prompt and seed, a community anime-style LoRA at scale 0.5 already gave a hand-drawn cel look, and at 1 the stylisation was stronger while the set and characters drifted further from the prompt. Without the file the same prompt stayed photoreal.
Limits: What fails, and whyFiles over 2 GiB fail with “Failed to load LoRA”, and so do gated links (HTTP 401) and repository page links. Many LTX-2 LoRAs on Hugging Face are 3.4 GB exports, and two official camera LoRAs (Jib Up and Static, 2.2 GB) are also over the limit. Failed requests are not charged.

Working with LoRAs on LTX-2 19B LoRA

Start with an official camera LoRA

Load a Dolly In LoRA at scale 1 and describe the destination of the move. The sheet shows the first and the last frame of one clip: the woman at the top of a marble staircase at the start, her face filling the frame at the end.

Pick the direction to match the story

Dolly In for a reveal, Dolly Out to show the place, Dolly Left to slide past a scene. The second sheet is a Dolly Out: a man at the helm of a sailboat at the start, the whole boat on the sea at the end.

Run a with-and-without pair
Same prompt, same seed, once with loras empty. If nothing changes, the file was trained for another base model: in our test a Wan 2.1 LoRA loaded without an error and changed nothing visible.
Stack gently, camera plus style
Up to three LoRAs can be combined. A style LoRA with the Dolly In camera LoRA, both at 1, worked together in our test and the style held. Raise the second file in small steps.
Keep trigger text in the prompt
Style LoRAs usually expect a trigger phrase. In our control run the trigger text alone did nothing without the file, so the file does the work and the phrase only helps it.

Who uses LTX-2 19B LoRA

Concert and stage videographers
A reveal from the back of the hall
Dolly In from the back of an empty concert hall to a man in a charcoal suit at a grand piano: the sheet shows the start and the end of one clip, the pianist small in the wide hall and large at the keys.
Photographers adding motion
A portrait that pushes in
On image-to-video, our stage-singer photo started wide and ended in a close-up at the microphone with the Dolly In LoRA loaded. The photo fixes the person and the wardrobe, the LoRA fixes the move.
Travel and lifestyle creators
Pull back to show where they are
A couple on a yellow scooter, then Dolly Out reveals the cliffs, the village and the sea. A pull-back is the classic way to end on the place instead of the people.
Uncensored platform builders
Community LoRAs, no review
Character and style LoRAs of any kind load like any other file: no allow-list, no review and no filter on the result. Everyone on screen must be an adult. In our tests the strongest probes still came back toned down, and an anime-style LoRA softened them further.

From first LoRA to finished batch with LTX-2 19B LoRA

Find a LoRA trained for LTX-2 19B
Check the model card for trigger text and a suggested strength, and check the file size is under 2 GiB.
Copy the direct download link
It must point at the .safetensors file itself and be public, not a repository page or a gated file.
Run it with and without the file
Same prompt, same seed, once with loras empty. A real LoRA is visible at a glance.
Ladder the scale
Try 0.5 and 1 and keep the value where the effect lands without taking over the subject.
Add a second file and run the batch
Bring in a camera or style LoRA at a modest scale, then send the settled loras array with every prompt.

LTX-2 19B LoRA prompts worth copying

A camera move with Dolly In

For text-to-video with the official Dolly In LoRA at scale 1.

The camera dollies in toward an adult woman in a forest-green dress standing at a candlelit dinner table on a terrace, until her face fills the frame as she looks up and smiles. Warm evening light, string lights. Audio: soft guitar, distant laughter, a glass being set down.

A pull-back reveal with Dolly Out

For image-to-video with Dolly Out. Describe what the camera should reveal.

The adult man leaning on the rooftop rail smiles as the evening wind moves his shirt, while the camera dollies out to reveal the whole skyline glittering behind him. Audio: wind, faint city traffic, a distant siren.

A style LoRA with a trigger phrase

For text-to-video with a style LoRA. Put the trigger words from your LoRA's model card at the very start of this prompt.

Hand-drawn animation style. An adult couple shares an umbrella at a seaside bus stop in light rain, she laughs and he points at the approaching bus, soft pastel colours. Audio: rain on the umbrella, gulls, a bus engine.

All LTX models6

LTX 2.5Text to video · Image to videoFrom $0.09per output second
LTX 2.3 SpicyImage to video · Video extensionSpicy list price: $0.02From $0.019per output second
LTX 2.3 Spicy LoRAImage to video · Video extensionSpicy list price: $0.03From $0.029per output second
LTX-2.3 LoRAText to video · Image to videoFrom $0.024per output second
LTX-2 19BText to video · Image to video · +1From $0.014per output second
LTX-2 19B LoRACurrentText to video · Image to videoFrom $0.018per output second

LoRA guide

Which LoRAs load on LTX-2 19B LoRA

Workbench