Try a focused photo edit in iMini
A photo-editing request can fail before there is an image to judge. The upload may not finish, the conversation may return instructions instead of a picture, or the request may remain pending. Adding a longer creative prompt does not tell you which stage failed.
This guide helps you identify the stage, run one small diagnostic edit and decide whether to retry, change the input or move the task to another editor. If an image was produced but its face, label or edges are wrong, use the separate follow-up correction guide.
Start by identifying what is missing
Confirm that the source image is attached, the chosen experience offers image editing and no visible error or limit needs attention. Then request one simple change to one image. Do not buy a plan or keep rewriting a prompt before checking what the interface actually reports. Availability and controls can differ between products, accounts and workspaces.
| What you see | Check first | Smallest useful next action |
|---|---|---|
| No image preview after attaching | Whether the attachment finished and the file opens locally | Save a normal PNG or JPEG copy if the interface accepts it, then attach once |
| A written explanation with no edited image | Whether image creation is available in the selected experience | Explicitly request an edited image from the attached source |
| A pending request or an error | The visible message, account limits and service availability | Record the error; avoid submitting a stack of identical jobs |
| The wrong photo is used | Which attachment belongs to the current request | Start a clean request with one clearly identified source |
| An image arrives, but the face or product changes | The output against the original | Switch to visual correction or restore the original layer |
1. Make sure the upload is actually usable
Open the source on your device. Check that it is an image rather than a cloud shortcut or a file that was renamed without conversion. If the editor displays accepted formats or a size limit, follow those current rules; there is no single upload limit that applies to every image tool.
Keep the original file. For diagnosis, export a separate, ordinary PNG or JPEG when supported, use a simple filename and attach only that copy. Wait for the attachment preview before sending the edit. Re-exporting is a compatibility check, not a way to recover detail or prove that a service problem is fixed.
2. Separate a text answer from an image-editing result
A response describing how to edit a photo is not an edited photo. Confirm that the selected product or mode can create an image, then state the desired deliverable explicitly. A text-only experience cannot gain image-editing capability through a special phrase.
Minimal diagnostic request
Edit the single photo attached to this request and return an edited image, not a written editing tutorial. Change only the plain background to light gray. Keep the subject, framing and visible details unchanged. Add no text or extra objects.Use a simple, non-sensitive test image if you only need to check the workflow. If the interface reports that image editing is unavailable, follow that message or choose an available editor; do not interpret another text reply as proof that the wording was almost correct.
3. Handle pending requests and errors before changing the creative brief
Save the source and prompt before refreshing or leaving the page. Read the actual error: upload failure, connectivity, a usage limit and an unavailable feature require different responses. Use the service's published status information when available, but remember that a normal status page does not rule out an account-specific issue.
Avoid repeatedly clicking submit while a job may still be running. Check whether a result already exists in the current conversation or task history, if that interface provides one. Retry after resolving the reported condition; there is no reliable universal rule that every image request must finish within a particular number of seconds.
If the service declines a request, read the reason and use an allowed task. Rewording to evade a restriction is not a troubleshooting step.
4. Remove ambiguity about the source photo
A long conversation can contain originals, rejected outputs and style references. For a clean test, attach only the original to a new request and name its role. Reintroduce additional references one at a time if the selected editor supports them.
Restart with a clear source
The attached image is the original source photo. Edit this image only; do not recreate a different scene from its description. Make one change: [specific change]. Preserve [critical visible details]. If you cannot access the attachment or return an edited image, state that limitation.Once you receive an image, compare it with the source. Keeping a detail in the prompt does not lock its pixels. For exact product packaging or a face that must remain untouched, preserve the original area with a suitable non-generative editing or compositing method.
5. Use a controlled handoff when you need another editor
If the current experience cannot complete the task, move the original file and one clear brief to an available image editor. In iMini, open the image workspace and check the selected model's current input controls before uploading. Keep the same source and first edit so that you can compare results fairly; switching tools does not guarantee a better output.
A brief you can reuse in iMini
Use this original product photo as the reference. Create a clean studio background for a store listing. Preserve the product shape, packaging, visible label wording, logo, colors and camera angle. Add no prices or headlines. I will compare the output with the original before approval.| Task | Useful route | Acceptance check |
|---|---|---|
| One background change | An image editor that accepts your photo | Original subject, edges and contact shadow remain credible |
| Exact logo, label or price | Original layers plus editable type or compositing | Characters and shapes match approved files |
| Several campaign assets | An approved master followed by a documented brief | Every variant passes the same product and layout checks |
Common mistakes that waste another attempt
Changing everything at once: a new file, new model and longer prompt make the cause harder to isolate. Treating access as a wording problem: prompts cannot enable an unavailable feature. Using a damaged output as the only reference: return to the clean source. Assuming payment fixes a failure: check the actual plan and error before spending.
For support, keep the error text, approximate time and timezone, product or model selected, file type and whether the minimal request worked. Share only the material needed to diagnose the issue.
Frequently asked questions
Why does ChatGPT explain the edit instead of returning an image?
The request may be interpreted as a tutorial, or image editing may not be available in the selected experience. Ask explicitly for an edited image and check the current controls or message.
Why can I not upload a photo?
Check that the file opens, the attachment finishes and the file meets the interface's current requirements. A failed upload is not fixed by a more detailed art prompt.
Should I keep resending a stuck request?
First check the visible status and any existing result. Repeated submissions can create duplicate work without resolving the original issue.
Does a longer prompt solve image-editing failures?
Only when it removes relevant ambiguity. It cannot repair a failed upload, enable an unavailable tool or remove an account limit.
What if an image is generated but my face changes?
That is a visual accuracy problem. Compare with the original, repair only the affected area and use original pixels when exact identity preservation is required.
Can I continue the same task in iMini?
You can bring the original image and brief to an available iMini image workflow. Check its current controls and review the output; capabilities vary by selected model.
Continue with the right editing guide
Use the upload-to-review workflow for a first edit, follow-up commands for a visible defect, and background-change prompts for a focused scene replacement.
The change-versus-preserve and one-change-at-a-time principles follow OpenAI's image prompting guide. Its API controls do not establish which buttons or features your ChatGPT account or another app exposes.