> For the complete documentation index, see [llms.txt](https://docs.atlas.design/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.atlas.design/atlas-ai-studio-overview/release-notes/0.53.0.md).

# Release 0.53.0

## Review game audio and video with the assistant

Ask a Gemini-powered assistant to listen to an audio asset and discuss its sound, or compare two clips. The assistant can also inspect three sampled frames from a video to help review character appearance, framing, and visual changes. Video inspection samples still frames; it does not hear the soundtrack. Audio and video inspection supports clips up to two minutes; trim longer media first.

![The assistant hears Captain Mira's dialogue and reviews the welcoming delivery](https://3654894688-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2FlNm6uP6UQSegMG59mAzs%2Fcaptain-mira-audio-feedback.jpg?alt=media)

## More control over Omni video

Choose Gemini Omni Flash 1.1 or 1.2 in Text + image to video. Version 1.1 offers 3–10-second clips; version 1.2 offers 3–15 seconds. Use the dedicated **Start Image** input to set the opening frame while keeping character and style references separate.

Advanced controls include **Thinking** choices for Google-served generation and Omni 1.2 editing: Provider default, Low, Medium, or High.

<figure><img src="https://3654894688-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2F7qOthAjvwajfZ3i1fGop%2Fimage.png?alt=media&amp;token=39d9aa32-a81f-4123-be23-45980a7cdaf9" alt=""><figcaption></figcaption></figure>

{% embed url="<https://files.gitbook.com/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2FKGBdbesHRBdmaTQGGGio%2Fcaptain-mira-omni12-opening.mp4?alt=media>" %}

## Bring NPC dialogue to life

The Lipsync node now includes **P-Video-2 Pro Avatar**. Animate a portrait with recorded audio, or enter a script and choose from 30 voices and 10 languages. Control character motion, prompt expansion, and output resolution from 480p through 4K.

<figure><img src="https://3654894688-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2F9Cz8D8BFHo2WQUhN1zQz%2Fimage.png?alt=media&amp;token=8320a781-4d53-4697-ad2c-62266b8f88a9" alt=""><figcaption></figcaption></figure>

{% embed url="<https://files.gitbook.com/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2FlLjYNgclZ8NPDFUEau9h%2Fcaptain-mira-avatar.mp4?alt=media>" %}

## Create music with Mureka 9.5

**Create Music** now offers Mureka 9.5 for songs and instrumentals. Describe the song you want, supply your own lyrics with optional musical direction, or generate an instrumental for a scene or game environment. Each run produces one audio track.

Song duration is an approximate 120–300-second target or Auto. Instrumental duration is chosen by the model; trim the result with Audio Edit when a scene needs an exact length.

![Mureka 9.5 instrumental settings and generated exploration track](https://3654894688-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2FtV1BfjEmey0PyZZvD5yD%2Fmureka-instrumental.jpg?alt=media)

![Mureka 9.5 song-from-description settings and generated tavern song](https://3654894688-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2FlOOekbONbPbfrhoPLyDF%2Fmureka-description-song.jpg?alt=media)

![Mureka 9.5 own-lyrics settings and generated quest song](https://3654894688-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2FWh3LspPR0gCWUspnljKA%2Fmureka-own-lyrics.jpg?alt=media)

## More control over Krea 3 reference edits

Krea 3 now offers **High** and **Ultra** reasoning in Text + image to image and Text + image list to image. These modes edit at least one reference image.

Use **Rect mask** and **Create image list** to guide edits with up to 24 rectangular regions, then refer to them as `@region-1`, `@region-2`, and so on in the prompt. **Outside Regions** lets you choose Auto, Preserve, or Regenerate. Each mask must match the chosen reference image’s dimensions; irregular mask shapes are treated as their bounding rectangles.

<figure><img src="https://3654894688-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2FCj6QCJRzAgCeyY7jxA5k%2Fimage.png?alt=media&amp;token=ebbd86c0-efe9-4ee8-a3cd-3e55e9c909dc" alt=""><figcaption></figcaption></figure>

<figure><img src="https://3654894688-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2FwawJW52hiOcp9T5c1l0G%2Fimage.png?alt=media&amp;token=53cbc91a-5743-4320-8f03-f6257bea6c4e" alt=""><figcaption></figcaption></figure>

![Krea High gives Captain Mira a winter outfit and snowy harbor setting](https://3654894688-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2FTk5Rh16SeZd02duJd8ed%2Frelease053-krea-high-winter-viewer.jpg?alt=media)

<figure><img src="https://3654894688-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2FxpcEYLpuit08VQZiIygH%2Fimage.png?alt=media&amp;token=831bce89-b686-43aa-9031-d536aedaa5f7" alt=""><figcaption></figcaption></figure>

![Captain Mira’s festival jacket after the regional edit](https://3654894688-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2FkLD1lA7HkNyabLcyxvTy%2Frelease053-krea-ultra-region-viewer.jpg?alt=media)

## HY Image 3.5 Preview

**Text to image** now offers HY Image 3.5 Preview, with 1K, 1.5K, 2K, and 4K output tiers and 13 aspect ratios. Choose the combined aspect/resolution setting for a new character, prop, or environment concept.

<figure><img src="https://3654894688-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2FQ813BKVX8PRPOMGRRWJS%2Fimage.png?alt=media&amp;token=d0028d28-bbb7-4a15-b0b6-bd4e7bb4e68d" alt=""><figcaption></figcaption></figure>

![Clockwork Harbor at dawn, generated with HY Image 3.5 Preview](https://3654894688-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2FjBsIMwbh8l04MoD1CAX7%2Frelease053-hy35-harbor-viewer.jpg?alt=media)

## Meshy T2 Smart Topology for image-to-3D

Choose **Meshy T2 Smart Topology** separately from T2.5 in **Image to 3D**. Use Auto, or set a custom triangle target from 100 to 15,000. T2.5 keeps its separate controls and 100–30,000 target range.

<figure><img src="https://3654894688-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2FhpR5hsBsLDvCNk76TgcJ%2Fimage.png?alt=media&amp;token=ff65f74c-4592-4f2d-88d6-1c36d257c5d5" alt=""><figcaption></figcaption></figure>

<figure><img src="https://3654894688-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2FM1sKqeQvXlb5tZy51YLR%2Fimage.png?alt=media&amp;token=d26d48b2-4ffa-4f8e-abc0-1fd77fb89616" alt=""><figcaption></figcaption></figure>

## Baidu ERNIE Light DT 1.0

Choose **Baidu ERNIE Light DT 1.0** in **Text generation (LLM)**, **Structured output**, or **Procedural image builder** for text-based game-content workflows. Light offers Off, Low, High, and Max reasoning, with Max as its default.

![ERNIE Light generates harbor keeper Mira’s greeting](https://3654894688-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2FNWKMdnkiON4drt6m0foj%2Frelease053-ernie-light-greeting-ui.jpg?alt=media)

## Prepare dialogue tags for Drama3

**Add audio tags** now includes **Fish Audio :: Drama3** as a target. Prepare expressive dialogue for the existing Drama3 text-to-speech engine while retaining the words of your script.

Our quiet lighthouse quest connects an original NPC line through Add audio tags to Drama3 speech.

![The original quest line connects through Drama3 audio tags to generated speech](https://3654894688-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2FK4R7PNZgmuvB7NFaSfGW%2Fdrama3-quiet-harbor-workflow.jpg?alt=media)

## Sourceful Riverflow v2.6 Pro

**Text to image** now includes **Sourceful Riverflow v2.6 Pro**, with 1K, 2K, and 4K output tiers, ten fixed aspect ratios, and Auto. Choose its combined aspect/resolution setting; the default is Auto / 2K.

<figure><img src="https://3654894688-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2F9gIQnYfF8C5KPI2qdrY5%2Fimage.png?alt=media&amp;token=e509aa07-2ff7-41c0-b8e9-f03810625cde" alt=""><figcaption></figcaption></figure>

![A brass-and-crystal waypoint beacon generated with Riverflow v2.6 Pro](https://3654894688-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2FEAhPn0411DwcPoheU4jO%2Frelease053-riverflow-beacon-viewer.jpg?alt=media)

## Japanese dialogue with Kotoba voices

**Text to speech** now includes **Kotoba TTS**, with Male and Female Japanese preset voices. Choose a voice for Japanese NPC dialogue and receive a completed audio clip through the usual audio output.

![The same Japanese quest line connects to Kotoba’s Male and Female preset voices](https://3654894688-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2FRnMT8h9f52hEsNELFyjV%2Frelease053-kotoba-preset-voices-focus.jpg?alt=media)

## HeyGen Video

**Text + image to video** adds **HeyGen Video** with Text, Image, and Reference modes. Generate a 5–15-second clip from a prompt, use Start Frame in Image mode, or supply image, video, and audio references in Reference mode. Reference mode needs at least one image or video.

Choose output resolution from 480p through 2K. Prompt Enhancement offers Turbo, Quality, Ultra Quality, or Disabled. For 1080p and 2K text or reference generation, select 16:9 or 9:16.

![HeyGen text-mode settings and the generated Clockwork Harbor shot](https://3654894688-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2FsuEr4BBf6Vzugk438v3J%2Frelease053-heygen-controls.jpg?alt=media)

{% embed url="<https://files.gitbook.com/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2FkE2uorc7xDs3vkibF76M%2Frelease053-heygen-clockwork-harbor-original.mp4?alt=media>" %}

## Track subjects and edit masked video regions

**Track Video Masks** uses a text description of subjects to return up to four separate mask videos, a labelled preview, and track descriptions. Inspect the preview, then route a chosen mask with the new **Break Video List** utility. Start Index chooses where its four output slots begin; changing the selection can reuse existing masks.

**Masked Video Edit** combines a source clip, an aligned black-and-white mask, a prompt, and up to four appearance references. It edits the white mask region and composites the surrounding source context, retaining the source canvas, timing, and soundtrack through re-encoding. Choose 480p, 580p, or 720p processing quality; prompt rewriting is optional and off by default.

Tracking and masked editing accept matching constant-frame-rate clips within their supported frame and size limits. Inspect tracking across occlusions and cuts before using its masks in an edit.

<figure><img src="https://3654894688-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2F4gOqGFCbD7Wqk337EKuF%2Fimage.png?alt=media&amp;token=0589985e-9517-4b49-8d19-5e6d23f11ff3" alt=""><figcaption></figcaption></figure>

## Keep original frames when dubbing dialogue

**Edit video** adds **Keep Original Video** for VEED Standard and Pro. Enable it to replace a clip's speech while keeping the source video frames and frame rate. The dubbed audio is time-scaled to the input length; the original lip movements remain, so they will not follow the new speech. The option is off by default, which keeps the existing lip-synced render. It requires an MP4-compatible video codec.

<figure><img src="https://3654894688-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2F7yJGplvpxMHsOi0GSfJU%2Fimage.png?alt=media&amp;token=1a1e6241-f9ce-443c-af05-9824e3cb177c" alt=""><figcaption></figcaption></figure>

{% embed url="<https://files.gitbook.com/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2FN4vLi5dvQlbg95rjA5gX%2Fedited_video%20(11).mp4?alt=media&token=39f3bebe-5206-47c8-89cd-f3fde8641f42>" %}
&#x20;**Keep Original Video On**
{% endembed %}

{% embed url="<https://files.gitbook.com/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FR7boiMixMhR4q36Ns33Y%2Fuploads%2FmDT58HZZJipdYNyEAYxw%2Fedited_video%20(13).mp4?alt=media&token=a71c00dc-118d-41db-9f0d-73336ab1b13b>" %}
&#x20;**Keep Original Video Off**
{% endembed %}


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

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

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.atlas.design/atlas-ai-studio-overview/release-notes/0.53.0.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

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.
