Upload one clear photo of your own face, name an era, and get back a portrait of you in a jazz age ballroom, a Renaissance studio or a Roman forum. Wardrobe, light and film grain do the heavy lifting, and the prompt words on this page tell you which ones to pull.
A single selfie can come back as you in a 1920s ballroom, posed in a Renaissance painter's studio or standing in a Roman forum, with the clothes, the light and even the film grain of the period. It is not a time machine and the history will not always be right, but as portrait play it works surprisingly well. This page covers the models, the era vocabulary that steers them, and the places where the illusion cracks.
You could type a woman in a 1920s flapper dress into any image model and get a lovely stranger. That is not time travel, that is stock imagery. The whole appeal here is that the person in the picture is you, and for that the model has to see your face. Image editing models take a reference photo plus a plain sentence and rebuild everything around the face: new clothes, new room, new light, same person.
Two models on Picasso IA are a natural fit. nano-banana-pro accepts up to 14 reference images next to the prompt, which helps when you want to feed it a selfie and a second picture of a costume or a setting. picassoia-image-editor-pro accepts up to three reference images and lets you address each one in the prompt as image 1, image 2 and so on, which makes instructions like keep the face from image 1 easy to write. Both sit in the catalog of 502 models alongside the other models in the 212-strong text-to-image category, so you can run the same selfie through several engines and keep whichever one held your likeness best.
New accounts get free credits, enough to try a few eras before you decide whether it is worth spending more. The cost of each run depends on the model and the settings, and the current plans are on the pricing page.
An era is a bundle of cues, and the model reads the cues you name. The more concrete the nouns, the better it behaves. Saying 1920s alone gives you a vague sepia mood. Saying a dropped-waist beaded dress, a cloche hat, finger waves and an art deco ballroom gives it something to build.
| Era | Wardrobe and setting words | Look to ask for |
|---|---|---|
| 1920s jazz age | beaded dropped-waist dress or three-piece suit, cloche hat, finger waves, art deco ballroom | soft studio portrait, slightly faded silver tones |
| Renaissance | velvet doublet or brocade gown, ruff collar, painter's studio, heavy drapery | oil painting, warm window light, fine surface cracks |
| Ancient Rome | draped tunic or toga, laurel wreath, marble columns, forum steps | sunlit stone, warm bronze tones, wide portrait |
| Victorian | high collar, bustle or frock coat, parlor with patterned wallpaper | stiff studio pose, albumen print, soft focus edges |
| 1970s | wide lapels, patterned shirts, feathered hair, wood paneling | warm film stock, faded colors, mild grain |
Pick the medium as deliberately as the clothes. A Renaissance request that ends up as a glossy photograph feels wrong, so ask for an oil portrait. A Victorian one wants a monochrome print. The portrait styles page goes deeper on how medium words change a face, and the effects page has preset looks if you want a starting point rather than a blank prompt.
This is the flow that ends with a portrait you would actually share, rather than a folder of almost-you.
The strongest prompts read like a short photo brief. Start with the instruction, then the era, then the garments, then the room, then the camera. For example: place the person from image 1 in a Roman villa courtyard, wearing a cream wool tunic with a red border and a thin leather belt, standing beside a marble column, late afternoon sun, shot like a medium format portrait, keep the face identical to image 1.
Light carries the era more than most people expect. Candle and window light say Renaissance, hard flash and plain backdrops say early photography, flat overcast daylight says Victorian studio. Film vocabulary helps in the modern direction: faded warm color for the 1970s, silver tones and soft focus for the 1920s. These are style words, not facts the model checks, so treat them as steering and not as guarantees.
Tip: decide what happens to your hair before you hit generate. The model tends to copy the modern haircut from your reference into the past, and a sharp modern fade under a Roman wreath reads as a costume party. If you want a period style, say so in the prompt.
If you want to go further than one portrait, the same selfie can anchor a whole set: you in a Roman forum, you in a Victorian parlor, you in a jazz club. Keeping the reference identical across runs is the same trick behind consistent AI characters.
Most failed time travel photos fail for the same few reasons, and all of them are fixable in the prompt.
The first is the vague era. A request for old fashioned clothes lets the model guess, and it tends to guess a mashup of three centuries. The second is modern grooming left on: visible highlights, a sharp fade, perfectly shaped eyebrows. The third is the wrong medium, a crisp digital look in a century that never had one. The fourth is an overstuffed prompt that asks for six props, three locations and a mood, so the face loses priority and drifts.
A fifth mistake is trusting the history. The model has seen countless costume photographs and film stills, so it often blends real period dress with movie versions of it. If a detail matters to you, check it against a museum collection or a costume reference before you present the picture as accurate. For fun, it hardly matters. For a school project or a family history display, it does.
This works as portrait play and fails at anything that needs documentary truth. Here are the places it stops being useful:
None of this makes the tool less fun. It just tells you where to stop zooming in.
It is a portrait where an AI image model takes a reference photo of you and rebuilds the picture around your face in another era, with period clothes, setting, lighting and sometimes film texture. You are not getting a historical reconstruction. You are getting a new, stylized image that keeps your likeness and changes everything else, closer to a costume portrait than to anything archival.
Image editing models that accept a reference photo are the natural fit. nano-banana-pro takes up to 14 reference images and picassoia-image-editor-pro takes up to three, each addressable inside the prompt. Other editing models in the catalog, such as qwen-image-edit-plus and flux-kontext-pro, are worth a run if the first pair does not hold your face. Likeness is uneven between engines, so testing two or three on the same selfie is a cheap and informative experiment.
Use your own face, or a friend or relative who has clearly agreed to it. Do not put real, identifiable people who have not consented into period scenes, and never a public figure, a classmate or a stranger from social media. A fictional character or an invented face is fine. The point of this use is playful self portraits, and consent is what keeps it that way.
Start from a sharp, front-facing, evenly lit reference and add an explicit line to the prompt telling the model to keep the face and features of image 1 unchanged. Keep the pose reasonably close to the reference, avoid extreme angles, and change one thing at a time when you refine. If the face drifts in a run, discard it instead of trying to patch it, because a fresh run is easier than rescuing a wrong face.
Name the garments, not just the decade. A dropped-waist beaded dress, a cloche hat, a doublet with a ruff collar or a draped tunic gives the model concrete things to draw, while a phrase like old fashioned clothes leaves it guessing. You can also attach a second reference image of a costume you like, which is easier than describing it. Cross-check the result against a museum or costume archive if accuracy matters.
Usually the era is under-specified, the modern hairstyle is still being copied from your reference, or the medium is wrong. Fix it by naming a specific decade, asking for a period haircut and grooming, and requesting the right medium, such as an oil portrait for the Renaissance or a monochrome studio print for the Victorian look. A weaker prompt produces a modern photo in a costume rather than a photo of the past.
Yes, as a style. Ask for faded silver tones, soft focus edges, film grain or the look of an albumen print, and the model will imitate those qualities. It is still a new image made today, so do not present it as a genuine archive photograph or use it to deceive anyone about its origin. If you want to repair real old pictures instead, the old photo restoration page covers that separate job.
It is possible but much less reliable than a solo portrait. Each person needs a clear reference, and the models tend to blend or swap features between faces, especially in the back row. Everyone in the picture must also have agreed to appear. A practical approach is to make individual period portraits of each person in the same era and styling, then place them side by side in a layout or collage afterward.
You can, and upscaling usually helps, since the first output may be smaller than a print needs. The toolkit has upscaling models for that step, and the photo upscaler page explains the options. Check the print size and file requirements of whichever print service you use, and look closely at faces and hands at full size before ordering anything.
Each generation uses credits, and the amount depends on the model and the settings you choose, so there is no single figure to quote. New accounts start with free credits, which is enough to test a few eras. Image edits sit at the lower end of what the catalog can do compared with video or 3D, and the current plans are always listed on the pricing page.
Ready to see yourself a century or two away from the present? Open the Picasso IA toolkit, attach one clear selfie and name the first era you want to visit.
Choose your language
