Source: https://docs.sogni.ai/sogni-create/studio/advanced/preview-count/

# Preview Count

While a generation is running, Sogni Studio can render intermediate "preview" frames so you can watch the image take shape. The **Preview Count** setting controls how many of those previews are produced per job.

## [#](https://docs.sogni.ai/sogni-create/studio/advanced/preview-count/#where-to-find-it)Where to find it

In the right-hand **Manual Controls** sidebar, expand the **Advanced** section. **Preview Count** sits there alongside Sampler, Processing, and Sensitive Content Filter.

![Preview Count control in Sogni Studio's Advanced settings](https://docs.sogni.ai/assets/remote/2485d465f0d8-spaces-2FjGm8Y44m4zCkKjnnr9Na-2Fuploads-2FVX3rSF.png)

The control is a numeric slider — set how many intermediate preview images you want per generation.

## [#](https://docs.sogni.ai/sogni-create/studio/advanced/preview-count/#what-it-changes)What it changes

-   **Higher count** — more checkpoints during denoising. You see the image evolve and can cancel early if it's clearly going off-track. Costs a small amount of extra render time per preview.
-   **Lower count** — fewer interruptions, faster total render. Best when you trust your prompt and just want the result.
-   **Zero** — no intermediate previews. Only the final image renders.

## [#](https://docs.sogni.ai/sogni-create/studio/advanced/preview-count/#when-to-raise-it)When to raise it

-   **Tuning a new prompt.** Watch how composition emerges so you can stop and rewrite if the early latents look wrong.
-   **Expensive jobs.** High-resolution image or video jobs cost more — earlier cancellation saves Spark.
-   **LCM / fast models.** With 2–8 step models you only get a couple of previews regardless; one or two is plenty.

## [#](https://docs.sogni.ai/sogni-create/studio/advanced/preview-count/#when-to-lower-it)When to lower it

-   **Batch generation.** Per-preview overhead multiplies across a batch.
-   **Production renders.** When you've locked in settings and just want the result.
-   **Slow on-device generation.** Each preview adds a CoreML pass — on smaller Macs the overhead is noticeable.

## [#](https://docs.sogni.ai/sogni-create/studio/advanced/preview-count/#see-also)See also

[Inference Steps— how many denoising steps the whole job runs.](https://docs.sogni.ai/sogni-create/studio/creating-images/generation-settings/inference-steps/)[🏁Activity Window & Background Queue— where queued jobs show their preview state.](https://docs.sogni.ai/sogni-create/studio/activity-window-and-background-queue/)[Generating Batches— preview behavior across multiple jobs.](https://docs.sogni.ai/sogni-create/studio/creating-images/generating-batches/)