# Create an image

Generate one to four images from a prompt, with an optional reference image. Every control on the Create Image page, what the price shows, and what to do when a run fails.

## Before you start

- Sign in. Without an account the button reads **Sign in to generate**.
- Check your credit balance. The price of the run shows on the **Generate** button. See [What actions cost](https://www.uova.com/help/billing/costs).

*Figure: A prompt and optional references go to the model you chose. Results land on cards and in your Library.*

## Generate an image

1. **Open Create image** Select **Create image** in the Create section of the Studio sidebar, or go to [/studio/create/image](https://www.uova.com/studio/create/image). On a phone the same page opens from the Create switcher at the top.
2. **Pick a model** Open the model picker. Models are grouped by maker family. The default is GPT Image 2.
3. **Write your prompt** Type in the **Describe your image** card. **Expand prompt** opens a larger editor with a **Done** button, and **Clear prompt** empties the field. If you select Generate with an empty prompt, the page says "Describe your image first".
4. **Add a reference image if you want one** Select **Add visual references** (tagged Optional) and choose a file. The tile shows **Reference added**, the file name and a 1/1 counter. The picture is scaled down to 2,048 px on its longest side before it is sent.
5. **Set the output** Choose an aspect ratio and, if the model offers one, a resolution. On desktop these sit in the Output card, on a phone in the chips above Generate. Then set **Count** from 1 to 4.
6. **Check the price, then select Generate** The price on the button is the price charged. A card with the model name and a running timer appears for each image, and fills in when the image is ready.

## Every control

| Control | What it does | Default and limits |
| --- | --- | --- |
| Model picker | Chooses the model. Each model has its own price and strengths. | GPT Image 2. 24 image models; see [Image models](https://www.uova.com/help/models/image). |
| Prompt | Describes the image. Sent to the model as you type it. | Required. |
| **Add visual references** | Adds one reference image to guide the result. | Optional. JPEG, PNG or WEBP, 20 MB at most, one image only. |
| Aspect ratio | Shape of the image, listed from tall to wide. | 4:3 when the model offers it. All 24 image models offer 16:9, 4:3, 1:1, 3:4 and 9:16. Shape changes the price on two models: Z-Image Turbo and GPT Image 1.5. |
| Resolution | Output size, where a model has more than one. | Nano Banana 2 offers 1K (147 credits), 2K (221) and 4K (294). Grok Imagine Image 2.0 offers 1K (111) and 2K (148). Seedream 5 and Seedream 5 Flash show a single 2K option. |
| **Count** | How many images one run makes. | 1 to 4. A run with a reference image makes 1. |
| **Generate** | Runs the generation. Shows the credit price. | Price is the model's quote multiplied by Count. Reads **Sign in to generate** when signed out. |

## Reference image rules

- 11 of the 24 image models take a reference image (text-to-image plus image-to-image). The other 13 are text only.
- If the model cannot use a reference, the Studio says "<model> can't use a reference image — remove it or pick another model".
- Select **Remove reference** on the tile to clear it.
- A reference image makes one result, not one to four.

## What the price shows

The credit price on **Generate** comes from the same calculation the server charges, so it is not an estimate. An option the model cannot do is refused before anything is charged. Prices differ widely by model: image generations run from 6 to 386 credits. Compare them in [Model pricing](https://www.uova.com/help/billing/model-pricing).

- The lowest price on any image setting is Z-Image Turbo at 6 credits, at 16:9 or 9:16. It is 8 at 4:3 or 3:4 and 11 at 1:1.
- FLUX Schnell is 7 credits at every shape. At the Studio's default 4:3, FLUX Schnell (7) is one credit below Z-Image Turbo (8).
- A reference image can change the price. GPT Image 2 is 98 credits at 4:3 without a reference and 113 with one. The button updates when you add the image.
- The most expensive image model is Recraft V4.1 Pro at 386 credits.

## Results, failures and downloads

- Results appear as cards and are saved to your [Library](https://www.uova.com/help/studio/library-and-downloads). Cards from this visit sit under **This session**, and the folder button at the top (**Your creations**) opens everything you made.
- Hover a finished card (or tap it once on a touch screen) to see its prompt and four actions: download, reuse the prompt, animate the image (opens Create video with the image as the start frame), and delete. Delete asks for a second press within 3 seconds. Selecting the picture opens a larger view.
- If a run fails, the card shows a short reason and then either **Refunded** with the amount or **No credits used**. The reason is **Not enough credits**, **Your session expired**, **Too many requests**, **Connection lost** or **Model unavailable**. Hover the card for the full message, and select **Use this prompt** to put the prompt back in the field.
- If the card says **Check your Library and credit balance**, the connection dropped after the request was sent. UOVA cannot tell from your browser whether the charge went through, so look in the Library before you run again.
- The Studio runs exactly the model and prompt you chose; if that model fails, nothing else is tried in its place and the charge is refunded.

## Limits

- One reference image per run. There is no multi-reference input.
- A file that is not JPEG, PNG or WEBP is refused with "Use a JPEG, PNG or WEBP image". A file over 20 MB is refused with "That image is over 20 MB".
- Image requests are limited per minute, and each image in a run counts as one request, so Count 4 uses four. The account limit is 10 a minute on Free and is multiplied by 3 on Basic, 6 on Pro, 12 on Studio and 24 on Creator. A separate limit of 10 a minute per network address does not scale with your plan, and whichever limit you reach first applies. The card says **Too many requests**.
- GPT Image 1.5 retires on 2026-12-01.
- Studio prompts go to the model as typed. The site editor's own image helper behaves differently: it adds text-free instructions to its prompts.

## Common questions

**Can I generate more than four images at once?**

No. **Count** goes from 1 to 4 per run. Run again for more.

**Do I need a paid plan for a particular model?**

No. Every image model is available on every plan, including Free. Credits are the limit.

**Can I edit the image afterwards?**

Yes. Use **Edit Image**, **Upscale** or **Remove Background**. See [Studio tools](https://www.uova.com/help/studio/tools).

Source: https://www.uova.com/help/studio/create-an-image