# SDXL Online - Resolution, Steps, Guidance, and Seed Controls

> Canonical: https://stable-diffusion-web.com/sdxl

## SDXL online - configurable text-to-image form

This SDXL route is text-only: a prompt, an optional negative prompt, seven framings, five sizes, step presets, guidance, a seed, and an output count.

## What the current SDXL form exposes

Prompt, exclusions, ratio, resolution, steps, guidance, seed, and count all live in the form; the page only describes them. Check each result against the exact setup you submitted.

### A 5,000-character prompt and a 2,000-character negative prompt

Write the subject in the required field and name concrete unwanted elements in the optional exclusion field.

### Seven framings from square to 2:3

1:1, 16:9, 9:16, 4:3, 3:4, 3:2, or 2:3 set the frame; they say nothing about anatomical or factual accuracy.

### Resolution from 512 to 2048

Pick 512, 768, 1024, 1536, or 2048, then check the delivered size and any artifacts once the run returns.

### Seven step presets

Steps are chosen from 1, 10, 20, 35, 50, 100, or 150; a higher count is not a universal quality upgrade.

### Guidance in half-point increments

Guidance moves from 1 to 10 in steps of 0.5. Change one variable per run to see what it did.

### Seed, count, and who may use the output

A seed anchors comparisons and a count returns several candidates. Free-tier output is for personal use; paid plans carry this app's commercial license.

## How to use the SDXL resolution, steps and guidance form

Lead with the subject and the treatment it needs; touch a control only to answer a specific question about the output.

### Name the subject, then the exclusions

Put subject, composition, and treatment in the prompt. Reserve the negative prompt for things you can see and do not want — not adjectives.

### Set the frame, then sample deliberately

Pick a ratio and one of the five sizes first. Move steps, guidance, seed, or count only when the defaults fail the comparison you are making.

### Inspect before you iterate

Check composition, anatomy, text, geometry, size, and artifacts on the returned file, then change a single control for the next run.

## Who uses SDXL's resolution and sampling controls, and for what?

Designers, illustrators, marketers, photographers, and concept teams can probe resolution, ratio, steps, guidance, seed, and count with the workflows below; none of them removes the need to review.

### Portrait lighting tests

Vary light, framing, and treatment, then look closely at faces, hands, skin, clothing, and what the background did.

### Scene and illustration directions

Compare scene ideas while checking that several subjects, depth, and the focal hierarchy still read correctly.

### Studio and lifestyle product concepts

Generate product scenes and reject any result that touches labels, proportions, colors, trademarks, or regulated claims.

### Layout drafts with placeholder marks

Use generated posters as drafts only; redraw marks and verify each word before anything is published.

### Spaces, stairs, and windows

Examine lines, perspective, stairs, windows, doors, and furniture for impossible geometry instead of assuming the structure is sound.

### One-variable experiments

Keep prompt and seed fixed while changing resolution, steps, guidance, or count, and record the delivered files and the debit.

## What a run costs and returns

| Measure | Detail |
| --- | --- |
| 1 model | Available in the form today, each priced in the same credits. |
| 4 credits | Per run, depending on the model and length chosen. |
| Every paid plan | Includes a commercial licence for what you make. |
| 5,000 characters of prompt | The longest prompt every model on this page accepts. |

## SDXL resolution, steps and guidance FAQ

Text-only input, prompt limits, framings, sizes, step presets, guidance, seed, count, credits, time, rights, and what to review.

### What is SDXL on this page?

SDXL is the text-to-image workflow this route pins. The live form, not the page, owns its prompt, ratio, resolution, step, guidance, seed, and count controls.

### Can SDXL take a reference image here?

No — the current SDXL schema is text-only. Switch to an image-capable model when a source image has to shape the result.

### What sizes and framings does the SDXL form offer?

Sizes of 512, 768, 1024, 1536, or 2048 and seven framings: 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, or 2:3.

### How are SDXL steps and guidance set?

Steps come from a preset list — 1, 10, 20, 35, 50, 100, or 150 — and guidance runs 1 to 10 in 0.5 increments, alongside a seed and an output count.

### What does an SDXL run cost, and who may use the output?

Metadata starts at 4 credits with an estimate near 30 seconds. Free-tier images are personal-use only; every paid plan includes this app's commercial license. Confirm the live debit and the final rights before use.

### How many credits does SDXL use?

SDXL draws 4 credits per run.

### What limits should I check before using SDXL?

Generated output from SDXL can contain visual, audio, or factual errors. Review each result before publishing, and upload only material you are authorized to use.

### What happens to what I generate with SDXL?

A result is stored against your own account, not pooled with other people’s. Deleting the account removes it: the control is in Settings under Security, the request is confirmed by email, and the erasure covers the generation records and the stored files it owns. How long anything else is kept, and which processors handle it, is stated in the privacy policy.

### Can I use what SDXL makes commercially?

Every paid plan includes a commercial licence for what SDXL makes; the free tier is for personal use.

### How does SDXL compare, and what alternatives fit a different workflow?

SDXL: Use SDXL with text input, optional negative prompt, seven ratios, 512-2048 resolution settings, steps, guidance, seed, and output count. Related options are Stable Diffusion Models Hub: Compare the Stable Diffusion models and the other image and video models here: the inputs each takes, what a run costs, and the plan it needs, Seedance 2.0: Use Seedance 2.0 with text, first/end frames, or up to 9 image, 3 video and 3 audio references; six ratios, 480p/720p, 4-15 seconds, seed, and audio switch, Runway Gen-4: Use Runway Gen-4 Turbo for 5- or 10-second video from text or one reference image, with 720p/1080p controls and paid access, Seedance 2 Mini: Seedance 2 Mini is the lowest-cost Seedance 2 variant - the same multimodal ByteDance model for 1080p video with native audio, text-to-video, Nano Banana Pro: Use Nano Banana Pro for prompt-based generation or reference-guided edits with 1K, 2K and 4K output, four-image input and paid access, and Seedream 4.5: Use Seedream 4.5 with text or up to ten JPEG, PNG or WebP references, 2K or 4K output, eight ratios, seed, output count, and paid-model access. Choose by the required input, delivered output, controls, limits, current availability, and credit cost; the related-tools section links each canonical page.

## Test a current SDXL workflow

Open the generator, enter text and optional exclusions, set the displayed controls, review the credit estimate, and inspect every returned image.

## Related tools

### [Stable Diffusion Models Hub](https://stable-diffusion-web.com/models)

### [Seedance 2.0](https://stable-diffusion-web.com/seedance)

### [Runway Gen-4](https://stable-diffusion-web.com/runway)

### [Seedance 2 Mini](https://stable-diffusion-web.com/seedance-2-mini)

### [Nano Banana Pro](https://stable-diffusion-web.com/nano-banana-pro)

### [Seedream 4.5](https://stable-diffusion-web.com/seedream-4-5)

_Last updated July 14, 2026_