Should you use ChatGPT to make a funeral photo?
It is fine for a background, a slideshow or a post online, but think twice before using it on the face of the person you are printing. ChatGPT does not edit your photograph; it draws a new picture based on it, so the face can come back subtly changed, and the result is usually too small for an easel print.
What ChatGPT does to a photograph
When you give ChatGPT a photograph and ask for a memorial portrait, it does not cut the person out and move them. It generates a whole new picture, guided by yours. Every pixel of the face is drawn again.
The newest models are good at keeping a likeness. Good is not the same as identical. The smile can sit a little differently, the teeth can be too even, the eyes a shade off, a scar or a crooked tooth quietly tidied away.
Most people would not notice on a phone screen. The family will notice on an 8x10 beside the guest book, because that is the face they spend the longest looking at all day, and they knew it best.
Where a generated picture is genuinely useful
Scenery is the easy case. A sunset, a lake, a candle, soft clouds: nobody knows exactly what that lake looked like, so nothing can be wrong about it. The scenic backgrounds in our own tool were made this way, with ChatGPT’s image generator.
Screens are forgiving too. A picture for a social media post, a slide in the service slideshow or a message to relatives needs far fewer pixels than a print, and a small change to a face matters less at the size of a phone.
Where it struggles is the printed portrait. That is the one photograph where the face has to be exactly theirs, and it is also the one printed largest.
The photograph leaves your computer
To make the picture, ChatGPT needs the photograph, so it is uploaded to OpenAI. For many families that is fine. For a photograph of someone who has just died, some would rather it was not.
OpenAI’s own settings decide whether what you upload can be used to improve its models. Check them before uploading anything you would rather keep private.
Keeping the real face, and changing only the background
The usual reason for asking ChatGPT is the background: the kitchen, the party, the hospital room behind the person. That part can be replaced without touching the face at all.
Our tool does it that way. It lifts the person away from what is behind them, puts a plain color or a quiet scene there instead, and leaves the face as the pixels the camera recorded. It then checks what the photograph can print at and cuts every frame size from one framing. All of it runs in your browser, so the photograph is never uploaded.
If the photograph is small, an enlargement is offered as a separate step, shown beside the original so you can judge it. It reconstructs edges and texture. It will not rebuild a face.
If you have already made one
Nothing is wasted. Use it where it works well: online, in the slideshow, on a small print. If it looks right to the family, that is the test that counts.
For the large printed portrait, check the size first. A ChatGPT picture is usually smaller than the photograph it came from, so the original is often the better file to print, with the background changed and the face left as it was.
Related
Last updated 2026-09-18.