deep-imager AI Image Generator: One Source Image, 5 MB

Image to image on the deep-imager AI image generator starts from a single source file — not a stack of references.

One image. 5 MB max. Then generate.

How many source images can you attach?

One. If you select more, the workspace keeps the first and tells you the limit is one image.

That cap applies to image to image and to Watermark Remove, which also needs a source. Make Same loads one reference when the public still was itself an image-to-image job.

Crop or composite references offline, then upload the single file you want the AI image generator to follow.

What file rules apply before Generate?

The file must be an image, and it must be 5 MB or smaller — larger files are rejected before a credit is spent.

A non-image pick shows an error and does not become the source. Compress or export a JPEG or PNG under 5 MB, then upload again.

Watermark Remove still uses Nano Banana only. Paint a mask or close the editor for full-frame cleanup — see Watermark Remove.

• One file, image type, 5 MB or less.
• Preview the thumbnail before you click Generate.
• Replace the file if the subject is the wrong SKU.

When should you skip the upload and use text to image?

When you have no reference file — text to image does not ask for a source image.

Open text to image, write subject and ratio, and generate. If the still is close, download it and come back to image to image with that one file.

Listing swaps that must keep a bottle are image to image. The product photo page covers that path.

Does a rejected upload spend credits?

No — size and type errors stop before submit. A successful generate still spends the model cost on the button.

New accounts have 5 free credits. Pricing from $1/month: Basic 10 credits, Advanced $9.99 for 120, Professional $19.99 for 300. Failed jobs that the product marks failed return the credit.

Typical still time is 30 seconds–2 minutes. Read how credits work before a remix loop.

Open image to image.