Customize your input with more control.
Customize your input with more control.
Hint: Upload a prepared .zip archive or provide a URL. See the field description for the required file layout.
Customize your input with more control.
The cost of training depends on the number of steps. The formula is: 0.0024 * steps. With 1000 steps, your request will cost $2.40.
Fine-tune your training parameters and start right now.
/i2v)The /i2v endpoint trains a LoRA adapter for the LTX 2.3 video model that learns to animate a starting image. The first frame is held as a condition during training, so at inference you supply a still image plus a prompt and the model generates a video that begins from that image.
Key features:
0.5, the LoRA is versatile and works for both image-to-video and plain text-to-video inference.Provide a single .zip archive (linked via training_data_url) containing your clips and optional captions:
.mp4, .mov, .avi, .mkv.png, .jpg, .jpeg.txt file with the same base name as each media file (e.g. clip01.mp4 + clip01.txt). Optional but recommended.The archive must contain only videos OR only images. Aim for at least 10 files. Files in subfolders are fine — clips with the same name in different subfolders are kept distinct automatically. The first frame of each clip is what the model learns to condition on, so make sure clips open on a representative, clean frame.
Minimum clip length: with auto_scale_input off (the default), each video must already have at least number_of_frames frames (default 89, ≈ 3.7 s at 24 fps). Shorter clips are silently skipped, and if every clip is too short the request fails (422, "All training videos are too short to be trainable"). Turn on auto_scale_input to resample shorter clips to the target frame count instead. This applies to video datasets only; images are single-frame and always usable.
training_data_url (required)Type: string
URL to the .zip archive of training clips (and optional captions).
trigger_phraseType: string
Default: ""
A phrase prepended to every caption during training and used at inference to activate the learned concept.
rankType: integer (one of 8, 16, 32, 64, 128)
Default: 32
LoRA capacity. Higher values capture more detail but use more memory and overfit more easily.
| Value | Use Case |
|---|---|
| 8–16 | Small datasets, subtle changes |
| 32 | Balanced default |
| 64–128 | Larger datasets or complex subjects |
number_of_stepsType: integer
Default: 2000 (range 100–20000)
Number of optimization steps.
learning_rateType: number
Default: 0.0002
Optimization step size. Start with the default; lower it if results are unstable.
first_frame_conditioning_pType: number
Default: 0.5 (range 0.0–1.0)
Probability that the first frame is held as a condition during training.
| Value | Behavior |
|---|---|
| 0.0 | Never conditions on the first frame (behaves like text-to-video) |
| 0.5 | Versatile — the resulting LoRA works for both image-to-video and text-to-video inference |
| 1.0 | Always conditions on the first frame (most specialized to image-to-video) |
Note: for single-image datasets the first frame is the entire sample, so first-frame conditioning is skipped automatically and training behaves like text-to-video on stills.
number_of_framesType: integer
Default: 89 (range 9–121)
Frames per training clip. Must satisfy frames % 8 == 1; other values snap down to the nearest valid count.
frame_rateType: integer
Default: 24 (range 8–60)
Target frames per second. LTX 2.3's native rate is 24.
resolutionType: string (low, medium, high)
Default: medium
| Resolution | 16:9 | 1:1 | 9:16 |
|---|---|---|---|
| low | 512×288 | 512×512 | 288×512 |
| medium | 768×448 | 768×768 | 448×768 |
| high | 960×544 | 960×960 | 544×960 |
aspect_ratioType: string (16:9, 1:1, 9:16)
Default: 1:1
auto_scale_inputType: boolean
Default: false
Fit videos to the target frame count and frame rate. No effect on images.
split_input_into_scenesType: boolean
Default: true
Split long videos into scenes before training.
split_input_duration_thresholdType: number
Default: 30.0 (range 1.0–60.0)
Duration above which a video is eligible for scene splitting.
with_audioType: boolean or null
Default: null (auto-detect)
Controls joint audio-video training, all-or-nothing across the dataset. With null (default, auto-detect), audio is enabled only when every clip has an audio track — if even one clip is silent, the whole run trains video-only. Set true to force audio training (the request fails with a 422 if any clip lacks an audio track), or false to ignore audio even when present.
audio_normalizeType: boolean
Default: true
Peak-normalize audio for consistent loudness.
audio_preserve_pitchType: boolean
Default: true
Preserve pitch when fitting audio to video duration.
validationType: array
Default: [] (max 2 entries)
Validation samples, each an object with:
prompt (string) — the text prompt.image_url (string, required) — the first-frame image to animate.validation_negative_promptType: string
Default: a built-in quality negative prompt.
validation_number_of_framesType: integer
Default: 89 (range 9–121)
validation_frame_rateType: integer
Default: 24 (range 8–60)
validation_resolutionType: string
Default: high
validation_aspect_ratioType: string
Default: 1:1
stg_scaleType: number
Default: 1.0 (range 0.0–3.0)
Spatio-Temporal Guidance scale for previews; 1.0 recommended, 0.0 disables.
debug_datasetType: boolean
Default: false
Return an archive of the preprocessed data for inspection.
lora_file — the trained LoRA weights (.safetensors).config_file — JSON describing the trigger phrase and training type.video — a combined validation reel (when validation samples were provided).debug_dataset — preprocessed-data archive, only when debug_dataset is enabled.A successful run is billed max(100, number_of_steps) billable units. Requests that fail before training completes (input-validation errors / HTTP 422, or dataset-download failures) are billed 0 units.
number_of_steps, conditioning on the first frame with probability first_frame_conditioning_p, running validation previews at intervals..zip is unpacked; macOS __MACOSX metadata folders are ignored..txt of the same base name.auto_scale_input, clips are resampled to the target frame rate and frame count.first_frame_conditioning_p) and the model learns to generate the rest.A LoRA is a compact set of adapter weights trained on top of the frozen base model. You load it alongside the base LTX 2.3 model at inference. Here it specializes the model at animating a supplied first frame.
If clips have sound and audio is enabled, the LoRA can produce a matching soundtrack. Audio is all-or-nothing across the dataset: if any clip is silent, the whole run trains video-only.
tronlog0 a glowing logo rotating slowly.Good caption: a golden retriever runs across a grassy field
Weak caption: dog
With scene splitting on, one long video produces several clips sharing one caption — which may not describe each split accurately. Pre-split and disable scene splitting for precise captions.
Prompt the LoRA the way you captioned it, including the trigger phrase. Supply a first-frame image consistent with your training footage. At first_frame_conditioning_p = 0.5 the LoRA also works without an image (text-to-video).
json{ "training_data_url": "https://example.com/my_dataset.zip", "trigger_phrase": "tronlog0", "rank": 32, "number_of_steps": 2000, "learning_rate": 0.0002, "first_frame_conditioning_p": 0.5, "number_of_frames": 89, "frame_rate": 24, "resolution": "medium", "aspect_ratio": "1:1", "validation": [ { "prompt": "tronlog0 rotating slowly", "image_url": "https://example.com/start.png" } ] }
rank, more varied data.rank, better captions.first_frame_conditioning_p toward 1.0.