Skip to content
eWiki The practical AI reference

Tutorials

AIAI Image to Image Tutorial: Reference, Prompt and Settings

The tool in this tutorial · AIAI.com

Give your reference a new direction

Open AIAI’s Image to Image workspace to prepare a reference-based edit. Keep your subject clear and describe the changes you want.

This operation guide shows how to prepare AIAI’s Image to Image generator with an existing reference image, a manually written prompt and the settings visible in our session. Follow the real recording, then use the illustrated steps to prepare your own form.

Setup only — no generation submitted. We stop before Generate. No edited image was generated or downloaded, and the displayed balance stayed at 6 trial credits. The preview is our uploaded input; the Inspiration gallery contains platform examples.
  • Beginner
  • English video + illustrated steps
  • Observed September 18, 2026
  • 0 generation jobs

Disclosure: eWiki has a business relationship with AIAI. See our disclosure. Model names and controls below reproduce the interface labels; this lesson does not independently verify model specifications.

Watch the setup walkthrough

Real continuous window recording with waiting intervals accelerated 8×. English synthetic narration uses the original Jenny voice, with captions. Zooms and pointer guides are editorial annotations. No output or processing-time test was performed.

1. Open Image to Image and inspect the model

Sign in to AIAI and choose Image → Image to Image. Open the model selector to inspect the choices. We kept the preselected GPT Image 2.5 Sunburst. Changing the model can change the available upload fields, controls and credit quote, so review the resulting form before following later steps.

The AIAI Image to Image model menu with GPT Image 2.5 Sunburst selected
The model menu captured during this preparation session. Select the image to inspect it at full size.

2. Add your reference image

Under Image Upload, click Add and choose your file. The visible counter allowed up to 16 images for this selected model; our example uses one. The form listed PNG, WEBP, JPG and JPEG with a displayed limit of 20 MB. These are the labels observed during recording, not promises for every model or future version.

Use an image you own or have permission to upload. Our reference is a pre-existing synthetic product scene containing a perfume bottle, compact and brush: a 1024 × 1024 WebP file of about 31 KB. We did not generate a new reference image for this guide. Wait until its thumbnail appears before writing the prompt.

Our existing product reference added to AIAI Image to Image
The thumbnail is the uploaded reference, not a newly generated result. Select the image to inspect it at full size.

3. Describe the edit clearly

Separate what should stay from what should change. In our example, the three products should remain recognizable while the background, lighting and shadows change. This gives a later result review a specific target.

Enter this text in the main Prompt field:

Use the uploaded product photo as the reference. Keep the perfume bottle, compact, and brush recognizable. Restyle the background as a warm beige studio set with soft side lighting and gentle shadows. Keep packaging blank. Add no text or logo.

We entered English manually, left Translate to English off and did not use Polish your Prompt, Enhance Prompt or the separate Prompt Assistant chat. Avoid Create Similar One while following this setup-only exercise.

The manually written AIAI Image to Image prompt
The complete prompt was entered without an AI helper or translation request. Select the image to inspect it at full size.

4. Review image settings

Keep 1:1 for the square example, select 1K and leave quality on Medium. We scrolled inside the input panel to make the quality control visible.

Control Our setting Observed detail
Aspect Ratio 1:1 Square frame retained.
Resolution 1K Changed from the initial 2K setting.
Quality Medium Kept at the visible default.
Credit quote 3 Initial quote 6; final quote 3 after setting 1K. No charge was submitted.
Prepared AIAI Image to Image settings and the untouched Generate button
1:1, 1K and Medium are the settings used here. The button shows a quote of 3 credits. Select the image to inspect it at full size.

5. Check visibility and stop before Generate

Public Visibility was on; we observed it without changing it. Before any future generation, review the current visibility options and whether they fit your material and account.

The final Generate 3 label is a displayed quote, not a charge from this tutorial. Our displayed balance stayed at 6 trial credits. Quotes can change with model, settings, plan or product updates.

Do not click Generate when following this exercise. The prepared input panel is the end state of this lesson. Images or videos in Inspiration belong to the platform gallery and do not demonstrate that your prompt has run. Save your prompt separately if needed; persistence of an unsubmitted form was not tested.

Final AIAI Image to Image setup with uploaded input, prompt and settings
The completed setup, unchanged credit balance and Inspiration gallery. No generation has been submitted. Select the image to inspect it at full size.
Prepare your next image edit

Open Image to Image on AIAI and follow the steps with your own reference. You can complete this exercise without submitting a generation.

Before you continue

Question What to check
My form looks different Check the selected model and follow its current field labels and file limits.
The reference did not appear Check the selected file, supported type, size and visible upload message. Do not submit generation just to test the upload.
The quote exceeds my balance Stop before Generate. Prompt preparation and reviewing the available settings are the scope of this exercise.
Does the output preserve the product correctly? No output was generated. We did not verify fidelity, motion, processing time, watermark, dimensions or export availability.

Continue with the AIAI Text to Image guide, or browse all video and illustrated tutorials.

Session notes

Recorded in signed-in desktop Chrome on September 18, 2026. Actions: model-menu review, upload of one existing reference image, manual prompt entry and the settings described above. Generation count: 0. Credits consumed by these actions: 0; displayed balance before and after: 6 → 6. No assistant request, translation request, payment, output generation or result download was submitted. The video and screenshots come from the same recorded setup. Uploading this input was observed separately from the untouched Generate action; this does not establish upload behavior for other tools.

Improve this guide

Report a problem or suggest the next question

Submissions enter a private editorial queue. They are reviewed and deduplicated before any public update or new page is created.