> ## 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.

# Generate, Monitor, and Download Your Fashion Photoshoot

> Start a five-pose photoshoot, track live pose-by-pose progress, stop or retry a job, and download individual images or a full ZIP from History.

Once your references are uploaded, Gemini analysis is current, and all five poses have prompts, you're ready to generate. Studio queues the entire five-pose photoshoot as a single generation job and resets immediately so you can start your next product without waiting.

<Note>
  Generation defaults: aspect ratio 3:4, quality medium, 5 poses in sequential order, resolution 2K.
</Note>

## Starting generation

<Steps>
  <Step title="Verify all requirements are met">
    Before clicking **Generate**, confirm:

    * Required reference cards show a preview image (no empty "Required" badge remaining)
    * The **Scene & styling** panel shows **Locked** — not "Analyzing…" or "Stale"
    * The **Poses** panel shows **5/5 ready from Gemini**

    If any requirement is missing, the **Generate** button remains disabled with a gray appearance.
  </Step>

  <Step title="Click Generate">
    Click the **Generate** button in the top-right header. Studio submits the shoot as a generation job and immediately resets the form — your references, analysis, and pose plan are cleared so you can begin uploading the next product right away.

    A submission banner appears at the top of the page showing the SKU name and live job status.
  </Step>

  <Step title="Monitor live progress">
    The submission banner tracks your job in real time:

    * **Status text** — shows the current pose generating (e.g., *'Pose 3 is generating'*) and a count of images stored vs. total poses (e.g., *'2/5 images stored'*)
    * **Progress bar** — fills as each pose completes, displaying the delivery percentage
    * **Thumbnail grid** — shows all five pose slots; each thumbnail updates with the generated image as it arrives

    The progress bar and thumbnails update live without requiring a page refresh.
  </Step>

  <Step title="Understand the consistency anchor">
    Pose 1 — **Full Front Product View** — is always generated first. Once Pose 1 is complete and stored, it becomes the **consistency anchor** for the remaining four poses. Poses 2–5 receive Pose 1 as an approved reference so the model's face, hair, and overall appearance remain identical across the full set.

    Your uploaded product references (front, back, fabric detail) always remain the highest-priority source of truth. The anchor only governs model appearance and scene continuity — it cannot override product fidelity.
  </Step>

  <Step title="View the completed shoot in History">
    When all five poses are complete, navigate to **History** to see the full shoot. Click any job row to expand it and view all five images at full resolution. From the expanded view you can download individual poses or download a ZIP of the entire shoot.
  </Step>
</Steps>

## Queue position

If another job from your organization is already active, your submission is held in the queue. The submission banner shows:

> *Queued — N tasks ahead*

Only one generation job runs at a time per organization. Your job starts automatically as soon as the active job finishes or is stopped.

## Stopping a job

Click **Stop** in the submission banner while a job is queued or processing to cancel it. A confirmation dialog appears:

> *The queued or active generation job will be cancelled. Every image already completed remains saved in History.*

Confirm to cancel. Images that already completed are never deleted — they remain accessible in History regardless of whether the job was stopped before all five poses finished.

## Automatic retry on QA failure

When **Pose QA** is enabled in Output settings, each generated pose is validated by Gemini against the Product Identity Profile and the consistency anchor. If a pose fails QA:

1. The failed attempt is archived (visible in History for review and download)
2. The pose is automatically re-queued for retry
3. The retry uses a **minimum 30-second backoff** before the next attempt

The submission banner shows the retry state. The final delivered image for each pose is always the one that passed QA (or the last archived attempt if all retries are exhausted).

<Note>
  If a pose prompt exceeds the provider's character limit, the pose is blocked before image generation. The error message reads *'invalid 'prompt': string too long'* and is visible in the History expanded view. Shorten the generation direction for that pose and regenerate it from History.
</Note>

## Cost tracking

Every generated pose stores the token usage reported by OpenAI:

* Input text tokens
* Input image tokens
* Total input tokens
* Output tokens
* Total tokens
* OpenAI request ID

**Actual cost** is calculated from those provider-reported tokens using the public rate for the selected GPT Image model. The History expanded view shows both an estimated cost (pre-generation) and the actual cost per pose.

<Warning>
  If OpenAI omits usage data for a pose, the UI shows *'Provider token usage not returned'* and does not fabricate a token count or cost estimate. The original pre-generation estimate remains visible separately. No cost is invented.
</Warning>

## Downloading your images from History

Open **History** from the top navigation or from the **View in History** link in the submission banner.

<Steps>
  <Step title="Find your job">
    History shows all generation jobs newest-first, paginated at 10 per page. Use the **Search** field to find a job by SKU code or prompt text. Filter by **Studio** source type to see only Studio-originated jobs.
  </Step>

  <Step title="Expand the job">
    Click any job row to expand it and load the full five-pose view with individual image previews.
  </Step>

  <Step title="Download an individual pose">
    Hover over any pose card and click the **Download** button that appears in the bottom-right corner. The image downloads as a named file: `{skuId}_{poseNumber}_{poseTitle}.{ext}`.
  </Step>

  <Step title="Download all five poses as a ZIP">
    Click **Download ZIP** in the job detail header. All stored poses are packaged into a single archive named `Youthnic_{skuId}.zip`. Each file inside is named `{poseNumber}_{poseTitle}.{ext}`.
  </Step>
</Steps>

## Regenerating a single pose

If a completed pose doesn't meet your standards, you can regenerate it from History without re-running the full shoot:

1. Expand the job in History
2. Hover over the pose card and click **Regenerate** (available for 24 hours after generation)
3. In the dialog, add extra instructions — for example: *'Back side should not have hanging or latkan elements. Preserve the plain uploaded back construction exactly.'*
4. Optionally enable **Run Gemini consistency QA** for the regenerated pose
5. Click **Regenerate pose** to queue the single-pose job

The original product references, model identity, scene, and locked shoot plan remain authoritative. Only the specific correction you added in the instructions field is applied.

<Warning>
  Pose 1 is the face and shoot anchor for the entire set — Poses 2–5 were generated to match it. Regenerating Pose 1 can change the model's face. If it does, regenerate Poses 2–5 afterward so every image in the set still shows the same person.
</Warning>

## Cloning a session

In the History expanded view, click **Clone Session** to load all the settings, references, and pose plan from that job back into Studio as a new session. Use this to run the same product at a different aspect ratio, quality level, or with a revised pose direction without re-uploading references.