How to create a toy with your own image via chatgpt: prompt and suggestions

For a moment let’s leave aside i duties And the doubts about the future that we are drawing with digital. And let’s play a game with Chatgpt. In recent days the race – on social media – was to those who posted your Miyazaki style image. An indispensable trend, we wrotewhich brought Sam Altman to the « pockets » -, CEO of Openai, a dry million new users in an hour. It is not known if there is an agreement between the American company and the Ghibli studio, but given the opinion of Hayao Miyazaki on the theme it seems difficult/impossible: «I would never want to incorporate this technology into my work. I firmly convince that it is an insult to life itself». It seems more a matter of laws on copyrights still very vague. However, it remains evident that the AI, however well programmed and trained, without the human spark remains sterile code. On LinkedIn, in recent days, a nice prompt suggestion has appeared from several users, precisely to play with the creation of chatgpt images without « stealing » anyone’s work. So here is the « formula » to create a barbie -style toy with your image.

The English prompt works well

Go on Chatgptalso very well the free version, and in the dialog box upload yours full -figure photo, possibly on a neutral background. Following the prompt, what is in fact the real secret to obtain good results from genai. Initially we tried it in English, with this text:

Create Image. Create to toy of the person in the photo. Let it be an action figures. Around the Figure, There Should Be The Toy’s Equipment, Each in Its Individual Blisters. 1) to newspaper Called « Corriere della Sera ». 2) A picture in frame with two girls, one teenascent and the other Younger. 3) Two Cat, One With Short Air and the Other With White And Black Long Hair, with Tag That Says Red One « Noah » and Black and White « Nami ». 3) to Coloured Console Gamepad. 4) A Comic Called « Spider -Man ». 5) A Book Called « Sein und Zeit ». Don’t Repeat the Equipment Under Any Circumstance. The Card Holding the blister Should Be Strong Orange. Also, on Top of the Box, Write « Federico Cella » and Underneath It « Toy Suitable for all ages ». The Figure and Equipment Must All Be Inside Blisters. Visualize this in A Lifelike Way

The final result can be found in the image above. How does it seem to you? The chatbot performed the task to perfection, Moreover with « personal » touches: the faces of the girls, who obviously are not the real ones; the cover of the spider man in the style of the original of the first Spiderman; The Xbox -style controller. The least adherent to reality is the image of the toy which – despite the excellent representation of the clothes – it is not very satisfactory regarding the face. But you can try with different photos, reminding you that for a task of this kind – in which chatgpt « reason » for a few minutes – with the free version You will not be able to do more than 2-3 attempts below.

The Italian prompt is rejected

Interesting to note that the same, identical, prompt proposed in Italian has given us a « refusal » as a result: «I cannot generate this image because The request violates our content policies». With the suggestion to try again, indicating a style that was not « realistic »the translation of English « Lifelike ». We tried with « manga », and here is the result. Much less satisfactory (with the addition of a cat-mix and a neologism, « Giogatto »).

To the question about where the violation in the Italian prompt took place, the answer leaves no room for replicas: the request « could have violated our policies because the combination of a real photo of a person with a stylized or realistic-commercial visual representation (like a toy or action figures) can raise problems related to identity, implicit consent and the use of the image». Clear, and also understandable. But why can you in English instead? Chatgpt’s response is manual for the use of genai. We summarize it like this: it is not a question of language but of interpretation by the model, which on English goes in confidence -the system « has presumed that it was not creating a direct and recognizable representation of the real person in the photo, but rather an inspired figure » -, while with Italian he is very close on the terms. Hence the need for well -calibrated prompt, even with several attempts. The one proposed by Chatgpt has something ridiculous. But it works: «Create a realistic style image. Represents an Action Figure Inspired by an imaginary person, called Federico Cella».