# Troubleshooting Guide

<details>

<summary><strong>Which mode should I use?</strong></summary>

Before diving into specific issues, use this to find your starting point:

* I want to generate a video from a text prompt → **Text-to-video** (any model)
* I want to use a specific image as the starting frame → **Image-to-video**
* I want to recreate the motion from an existing video → **Motion Control** (Kling 3.0)
* I want to use a character or background as reference → **Image/Video Reference** (Kling O3 or Seedance 2.0)
* I want to set both the first and last frame → **First/Last Frame** (Veo 3)
* I want to edit an existing video → **Video Edit** (Kling O3)

</details>

<details>

<summary><strong>I started a blank project and don't know what to upload</strong></summary>

Blank projects don't guide you through asset uploads the way templates do — you'll need to come prepared.

1. Find an inspiration image or video that matches the vibe you're going for.
2. Screenshot or save the key frame you want to reference.
3. Open **Studio Agent Chat** and attach both the inspiration image and your character. Ask the agent to generate a starting image using **GPT-2 Image** or **Nano Banana 2**.
4. Once you're happy with the starting frame, use it as your image input in your chosen video model.
5. Build scene by scene — you don't need all frames ready before you start.

</details>

<details>

<summary><strong>I don't know which video model to pick</strong></summary>

**Kling 3.0** — Best overall quality-to-cost ratio. Recommended as your default. Modes: Text-to-video, Image-to-video, Motion Control

**Kling O3** — More advanced reference capabilities. Modes: Text-to-video, Image-to-video, Reference-to-video, Video Edit, Video Reference

**Seedance 2.0** — Strong results with limitations. Modes: Text-to-video, Image-to-video, Reference-to-video Note: Does not support real human faces. Max output is 720p — a separate video upscaler may be needed for higher resolution.

**Veo 3** — Highest realism, higher cost. Modes: Text-to-video, Image-to-video, Reference-to-video, First/Last Frame Note: Stricter content filters than other models.

**Sora 2** — Realistic but not the sharpest output. Modes: Text-to-video, Image-to-video

**Hailuo** — Lightweight, good for quick tests. Modes: Text-to-video, Image-to-video

</details>

<details>

<summary><strong>I want to recreate the exact motion from another video</strong></summary>

Use **Kling 3.0 — Motion Control /** **Seedance 2.0** **— Reference to Video**

1. Upload the video whose motion you want to replicate.
2. Upload an image of your actor or character you'd like to have replicate the motion.
3. Write a prompt if there are any finer detailed you'd like to control.
4. Hit Generate.

</details>

<details>

<summary><strong>I like a clip but it's too short</strong></summary>

Once a video has been generated, look for the **extend icon** on the clip and click it to continue the video from where it left off. Repeat across multiple scenes to build a longer sequence.

</details>

<details>

<summary><strong>I want to build better prompts</strong></summary>

The **Studio Agent Chat** can create prompts for you. \
Describe what you want in the Agent chat and it will map out ingredients, scene breakdowns, and suggested prompts.

For more control, you can always fine tune the prompts before generating, or ask the agent to regenerate prompts for you.&#x20;

</details>

***

### Model Quick Reference

| Model        | Cost   | Human Faces | Max Resolution | Best For                     |
| ------------ | ------ | ----------- | -------------- | ---------------------------- |
| Kling 3.0    | Low    | ✅           | High           | Most use cases               |
| Kling O3     | Medium | ✅           | High           | Advanced reference workflows |
| Seedance 2.0 | Medium | ❌           | 720p           | Stylised / non-human content |
| Veo 3        | High   | ✅           | High           | Realism, cinematic quality   |
| Sora 2       | High   | ✅           | Medium         | Realistic scenes             |
| Hailuo       | Low    | ✅           | Medium         | Quick tests                  |


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.avastudio.com/start-here/ava-studio/troubleshooting-guide.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.