> ## Documentation Index
> Fetch the complete documentation index at: https://docs.aistudio.youthnic.shop/llms.txt
> Use this file to discover all available pages before exploring further.

# Generation History: Browse, Filter, and Download Jobs

> Browse every generation job, inspect five-pose results, download images or ZIP bundles, and debug failures with per-pose token and cost data.

History is your permanent record of every generation job run in your organization — whether from Studio on-demand shoots or Catalog Planning batches. Use it to review completed photoshoots, download images, investigate failures, and audit token usage and cost per pose.

Access requires the **Studio view** permission. Members with this permission can see the full organization history, not just their own jobs. Jobs appear newest-first and are paginated at 10 per page.

## Filtering and searching

Use the controls at the top of the History page to narrow the list before it reaches you — filters run server-side so performance stays fast even with large archives.

<CardGroup cols={2}>
  <Card title="Search" icon="magnifying-glass">
    Type any part of a SKU name or SKU ID. Results update automatically after a short debounce.
  </Card>

  <Card title="Source filter" icon="filter">
    Choose between **All Sources**, **Studio** (on-demand sessions), and **Catalog Production** (batch jobs).
  </Card>
</CardGroup>

**Status filter options:**

| Option             | What it includes                         |
| ------------------ | ---------------------------------------- |
| All generations    | Every job regardless of state            |
| Active generations | Queued and processing jobs combined      |
| Queued             | Jobs waiting for a worker                |
| Processing         | Jobs currently generating                |
| Completed          | All five poses stored                    |
| Failed             | Generation ended with an error           |
| Cancelled          | Jobs stopped by a user before completion |

## Job list

Each row in the job list shows:

* A **thumbnail** of the first completed pose (if available)
* The **SKU ID** in monospace and the **SKU name** below it
* A colored **status chip** with an animated spinner for processing jobs
* A **progress bar** showing images stored vs. total poses expected, plus a percentage
* The **creator's name** and the **date and time** the job was created
* Hover-revealed action buttons on the right side of the row

Click any row to expand the job and see all five pose images.

## Job detail view

Expanding a job reveals the full five-pose grid along with metadata about the generation session:

* **Aspect ratio**, **resolution**, and **quality** setting used
* **Estimated cost** (pre-generation estimate) and **actual cost so far** (calculated from provider-reported tokens)
* Any error message if the job or a specific pose failed

Each pose card in the grid shows:

* The generated image as a thumbnail — click to open a large full-screen preview
* A **status badge** in the corner indicating the pose's state
* The AI QA estimate percentage and QA verdict label beneath the image
* Token usage summary: input tokens, output tokens, and actual cost for that pose
* A **Download** button (visible on hover) to save that individual image

### Downloading images

<Steps>
  <Step title="Download a single pose">
    Hover over any completed pose card and click the **Download** button. The file is saved as `{SKU_ID}_{pose_number}_{pose_title}.{ext}`.
  </Step>

  <Step title="Download all five poses as a ZIP">
    Click **Download ZIP** in the job metadata bar above the pose grid. All completed poses are bundled into a single archive named `Youthnic_{SKU_ID}.zip`. Images are fetched in a single optimized pass, making ZIP downloads significantly faster than downloading each pose individually.
  </Step>
</Steps>

## Stopping and deleting jobs

Each job has action buttons that appear when you hover over its row:

| Job state                       | Available action       | What happens                                                                                    |
| ------------------------------- | ---------------------- | ----------------------------------------------------------------------------------------------- |
| Queued or Processing            | **Stop**               | Cancels all remaining poses. Images that already completed are preserved in History.            |
| Completed, Failed, or Cancelled | **Delete**             | Permanently removes the database record and all stored generated images. This cannot be undone. |
| Failed                          | **Retry failed poses** | Re-queues every pose that did not complete. Poses already stored remain unchanged.              |

Both Stop and Delete show a confirmation dialog with a description of what will happen before any action is taken.

## Generation flow view

Click the **brain icon** (View flow) on any job row to open the Generation Flow page for that job. This is your primary debugging tool when a specific pose failed or took longer than expected.

The flow view shows:

* Each operational stage in the generation pipeline with **completion timestamps** and **duration**
* Full **regeneration history** — every attempt, including QA-rejected frames that were never delivered
* The **five-pose status** at a glance
* Stage-level details including the reference manifest sent to the image generator, Gemini QA scores, and error messages

## Cost and token data per pose

Expand any job and click a pose to open its detail panel. The **OpenAI usage** section shows:

| Field              | Description                                          |
| ------------------ | ---------------------------------------------------- |
| Input tokens       | Text tokens sent in the prompt                       |
| Output tokens      | Tokens in the response                               |
| Image input tokens | Tokens consumed by reference images                  |
| Text input tokens  | Text-only portion of input tokens                    |
| Total tokens       | Sum of all token types                               |
| Actual cost        | Dollar cost calculated from provider-reported tokens |

If OpenAI did not include usage data in the response, the panel shows **"This provider response did not include token usage, so no token cost was invented."** No values are fabricated.

## Regenerating a pose

Failed poses and poses completed within the last 24 hours can be regenerated individually. Click the **Regenerate** button on the pose card (visible on hover for completed poses, always visible for failed ones).

In the regeneration dialog you can:

* Add **extra instructions** (up to 1,000 characters) — for example: *"Back side should not have hanging/latkan elements. Preserve the plain uploaded back construction exactly."*
* Leave instructions blank to retry with the existing locked shoot plan
* Optionally enable **Gemini consistency QA** on the regenerated pose

<Warning>
  Pose 1 is the face and visual anchor for the entire shoot — poses 2–5 were generated to match it. Regenerating Pose 1 may change the model's face. If it does, regenerate the other poses afterward to keep all five images consistent.
</Warning>

## Cloning a session to Studio

Open any job detail and click **Clone Session** to copy the job's product references, settings, and shoot plan into a new Studio session. Use this to re-run a past shoot with different output settings or updated references without rebuilding the session from scratch.

## Viewing reference images

Click **View references** inside any expanded job to load the product and style reference images that were attached to that session. References are fetched on demand — browsing History does not trigger extra storage requests for images you have not asked to see.
