LTX2 video LoRA training

Train a Lightricks LTX video LoRA on a small set of source video clips using AI Toolkit. The output LoRA is usable in LTX2 video generation.

ecosystem	Base	Default price	Notes
ltx2	Lightricks/LTX-2 (19B)	2750 Buzz	Original LTX2.
ltx23	Lightricks/LTX-2.3 (22B)	2750 Buzz	Newer LTX 2.3.

The base checkpoint is fixed by ecosystem; there's no model field on the input.

Video training is the slowest training mode on the platform — video needs a longer run, so the LTX default is 3000 steps (vs 2000 for image ecosystems). Submit with wait=0. The request fields, training data, continuing a run and live progress are shared by every ecosystem — see the LoRA training overview.

The request shape

{
  "$type": "training",
  "input": {
    "engine":    "ai-toolkit",
    "ecosystem": "ltx2"          // ltx2 | ltx23
  }
}

Prerequisites

• A training-data zip of images, video clips, or a mix — addressed by an AIR (accepted sources), with an accurate count of items

LTX2

Original 19B-parameter LTX video model. resolution: 768 is the typical training resolution.

POST https://orchestration.civitai.com/v2/consumer/workflows?wait=0
Authorization: Bearer <your-token>
Content-Type: application/json

{
  "tags": ["training", "video"],
  "steps": [{
    "$type": "training",
    "priority": "normal",
    "retries": 2,
    "input": {
      "engine": "ai-toolkit",
      "ecosystem": "ltx2",
      "steps": 3000,
      "resolution": 768,
      "lr": 0.0002,
      "trainTextEncoder": false,
      "lrScheduler": "cosine",
      "optimizerType": "adamw8bit",
      "networkDim": 32,
      "networkAlpha": 32,
      "trainingData": {
        "type": "zip",
        "sourceUrl": "urn:air:other:other:huggingface:datasets/Civitai/orchestration-samples@e86d025874700507615ae8ef74937c319ea41dfe/sample-training-dataset.zip",
        "count": 15
      },
      "samples": { "prompts": ["a video of TOK", "TOK moving in a garden"] }
    }
  }]
}

POST/v2/consumer/workflows

Set your Civitai API token via the Token button in the navbar to enable Try It.

Request body — edit to customize (e.g. swap the image URL or prompt)

{ "tags": [ "training", "video" ], "steps": [ { "$type": "training", "priority": "normal", "retries": 2, "input": { "engine": "ai-toolkit", "ecosystem": "ltx2", "steps": 3000, "resolution": 768, "lr": 0.0002, "trainTextEncoder": false, "lrScheduler": "cosine", "optimizerType": "adamw8bit", "networkDim": 32, "networkAlpha": 32, "flipAugmentation": false, "trainingData": { "type": "zip", "sourceUrl": "urn:air:other:other:huggingface:datasets/Civitai/orchestration-samples@e86d025874700507615ae8ef74937c319ea41dfe/sample-training-dataset.zip", "count": 15 }, "samples": { "prompts": [ "a video of TOK", "TOK moving in a garden" ] } } } ] }

Valid JSON

LTX 2.3

Newer 22B model. Same shape as LTX2; lr is typically lower. Both LTX ecosystems share the same rates and 3000-step default, so the default price matches (2750 Buzz).

POST https://orchestration.civitai.com/v2/consumer/workflows?wait=0
Authorization: Bearer <your-token>
Content-Type: application/json

{
  "tags": ["training", "video"],
  "steps": [{
    "$type": "training",
    "priority": "normal",
    "retries": 2,
    "input": {
      "engine": "ai-toolkit",
      "ecosystem": "ltx23",
      "steps": 3000,
      "lr": 0.0001,
      "trainTextEncoder": false,
      "lrScheduler": "cosine",
      "optimizerType": "adamw8bit",
      "networkDim": 32,
      "networkAlpha": 32,
      "trainingData": {
        "type": "zip",
        "sourceUrl": "urn:air:other:other:huggingface:datasets/Civitai/orchestration-samples@e86d025874700507615ae8ef74937c319ea41dfe/sample-training-dataset.zip",
        "count": 15
      },
      "samples": { "prompts": ["a video of TOK", "TOK moving in a garden"] }
    }
  }]
}

POST/v2/consumer/workflows

Set your Civitai API token via the Token button in the navbar to enable Try It.

Request body — edit to customize (e.g. swap the image URL or prompt)

{ "tags": [ "training", "video" ], "steps": [ { "$type": "training", "priority": "normal", "retries": 2, "input": { "engine": "ai-toolkit", "ecosystem": "ltx23", "steps": 3000, "lr": 0.0001, "trainTextEncoder": false, "lrScheduler": "cosine", "optimizerType": "adamw8bit", "networkDim": 32, "networkAlpha": 32, "flipAugmentation": false, "trainingData": { "type": "zip", "sourceUrl": "urn:air:other:other:huggingface:datasets/Civitai/orchestration-samples@e86d025874700507615ae8ef74937c319ea41dfe/sample-training-dataset.zip", "count": 15 }, "samples": { "prompts": [ "a video of TOK", "TOK moving in a garden" ] } } } ] }

Valid JSON

LTX parameters

Defaults apply to both LTX ecosystems. Fields and defaults not listed here are as in the overview's common parameters.

Field	LTX default	Notes
steps	3000	Video needs a longer run, hence the higher default.
epochs	round(steps / 200)	15 for steps: 3000; 10 if steps is omitted too. Each adds 50 Buzz — LTX preview samples are videos and costly to compute.
batchSize	1	Fixed at 1 for this ecosystem.
continueFrom	(none)	A LoRA of the same ecosystem (urn:air:ltx2:lora:... / urn:air:ltx23:lora:...) to resume from.
lr	0.0001	LTX2 examples often use 0.0002; LTX 2.3 typically 0.0001.
trainTextEncoder	false	Leave off — LTX text encoder is not retrained by AI Toolkit.
trainingData	✅	Images, video clips, or a mix; the example uses an image dataset.
samples.prompts[]	[]	Preview videos rendered at each saved checkpoint.

Reading the result

The standard training output: each saved checkpoint yields a video LoRA .safetensors blob plus any sample .mp4 files. Use the trained LoRA in LTX2 video generation by referencing it in the workflow's loras field.

Runtime

Wall time, default settings on a 4-clip dataset:

Ecosystem	Per 100 steps	Typical full run
ltx2	~1–2 min	30–60 min for 3000 steps
ltx23	~2–3 min	60–90 min for 3000 steps

Cost

The training cost formula with LTX rates:

price = steps × costPerStep + epochs × 50        (rounded)
costPerStep = 0.75   (ltx2 and ltx23)
floor: never less than 80% of the default-configuration price

Each epoch adds 50 Buzz. With both steps and epochs omitted the run is 3000 steps / 10 epochs → 3000 × 0.75 + 10 × 50 = 2250 + 500 = 2750 Buzz. The floor is 80% of that price (2200 Buzz).

The examples at the top of the page send steps: 3000 with no epochs, so they get round(3000 / 200) = 15 epochs: 15 × (200 × 0.75 + 50) = 15 × 200 = 3000 Buzz. Send epochs: 10 alongside steps: 3000 to get the 2750 price.

Configuration	Buzz (training only)
LTX2 / LTX 2.3, default (steps: 3000, epochs: 10)	2750 + samples
LTX2 / LTX 2.3, steps: 3000, epochs: 20	3250 + samples (+500 for 10 more checkpoints)
LTX2 / LTX 2.3, steps: 2000, epochs: 10	2200 + samples (floor)

Sample-prompt rendering uses LTX2 video-generation rates and is billed separately. Preview the exact charge with whatif=true (Payments (Buzz)).

Troubleshooting

steps out of range, an unresolvable sourceUrl, a count mismatch and a rejected dataset are covered in the overview's troubleshooting.

Symptom	Likely cause	Fix
Step failed with VRAM-related error	Resolution × clip length too high	Lower resolution (e.g. to 512), shorten clips.
Training cost surprises you	Video defaults to 3000 steps, so the floor is higher than image ecosystems	Check whatif=true before submitting. Lowering steps/epochs saves at most 20% (the floor).
Trained LoRA produces no motion	Too few steps / static reference clips	Raise steps, ensure clips show the motion you want learned.

Related

• LoRA training overview — shared parameters, training data, live progress and cost formula
• Wan video LoRA training — Wan video LoRA training (preview)
• LTX2 video generation — use a trained LoRA in LTX2 inference
• Flux 2 Klein LoRA training — image-side counterpart
• Results & webhooks
• SubmitWorkflow · GetWorkflow
• training OpenAPI spec