Prompt patterns across LTX models
Describe the subject, action, camera movement and visual style. For image-to-video, explain how the scene should move from the opening image. Keep the same starting point and change one instruction at a time.
Explore current and earlier LTX versions. Open a video to read its original prompt and settings.
No videos are available in this gallery yet.
Each LTX video keeps its exact model, original prompt and settings. The links below open examples from current and earlier versions.
LTX 2.5 Pro and Fast lead this page. Older LTX 2.3 and LTX 2 examples keep their original model labels so you can compare generations without confusing their capabilities.
Describe the subject, action, camera movement and visual style. For image-to-video, explain how the scene should move from the opening image. Keep the same starting point and change one instruction at a time.
Compare the whole clip across Pro and Fast: does the motion follow the prompt, and does the subject remain consistent? Use the same source image for image-to-video tests. A successful example is a starting point, not a guarantee of identical results. Standard LTX 2.3 Reframe accepts source timing up to 60 seconds at 720p or 1080p without a prompt. This source operation does not transfer to LTX 2.3 Fast or LTX 2.5 Fast/Pro.
The price shown is a current estimate for a new generation, not the original spend. Open the example for its recorded settings; confirm your own settings and quote in the app.
Look for videos labelled LTX 2.3 Pro or LTX 2.3 Fast in this family gallery and use Next to browse older examples. Open a video for its original prompt and recorded settings, then consult the linked LTX 2.3 model page for that version’s supported modes. LTX 2.5 examples are labelled separately; a current price estimate is not the historical cost of the original render.
Start from a labelled LTX 2.3 example and inspect its original prompt. Identify the subject, main action, camera direction and visual style. For an image-to-video example, separate what the opening image already supplies from the movement described in the prompt. Keep the recorded duration, framing and model version alongside the text when comparing results; LTX 2.5 examples are a separate baseline.
The best starting point is a simple structure: subject, action, camera direction, and style goal. The strongest examples keep that structure stable while changing only one variable at a time.
Start with one clear subject, one main action, one camera instruction, and one visual style cue. LTX 2.5 prompts usually work better when the motion goal is explicit and the scene description stays tight.
The main settings to watch are duration, aspect ratio, source image choice for image-to-video, and how much motion complexity you ask for in a single prompt. Keeping those stable makes prompt testing much easier.
Start from a strong source image, then add one motion instruction, one camera movement, and one output goal. LTX 2.5 image-to-video works best when the prompt extends the source image instead of replacing it with a completely different scene.
Compare LTX 2.5 Pro and Fast with the same prompt and settings. Use the examples to judge the output you need and the current pricing page to compare costs. Check each model page for supported modes and limits; older LTX examples describe the version shown on their label.