Your portrait in black and white with only the knit sweater in colour, sleeves over your mouth. Burgundy and green, 38 credits for both, flaws shown.
Real workflow: a woman with long dark brown hair in a white T-shirt against a beige wall, then the same woman in black and white in a chunky burgundy cable knit, sleeves over her mouth, and in a forest green turtleneck on a charcoal background. In both renders her hair looks much shorter than in the photo.
At a glance
The "ChatGPT Sweater Portrait Trend" fits in one picture: a black and white portrait where only the knit sweater keeps its colour, often burgundy, with the long sleeves pulled up over the mouth and a sharp gaze above them. According to the viralvideoai.app guide, which quotes MakeViral's notes, the first prompt post went up on TikTok on 9 November 2026. Here you upload your portrait and an image model remakes the photo, without going through ChatGPT.
You get two still pictures, as in the trend. The first follows the original recipe: chunky burgundy cable knit, both hands in the sleeves in front of the mouth, light grey background. The second is a variation in a forest green turtleneck, collar pulled up over the mouth, on a charcoal background. On the demo, the face and the freckles are faithful. The hair, long in the photo, comes out short in both renders, and in the green one a bare hand comes out of the sleeve to rest on the cheek.
For another portrait session, see the AI photoshoot or the professional portrait. In the same seasonal mood there is the autumn lakeside daydream, and for something more melancholic, the heartbreak photo or the photo as a painting. Both renders cost 38 credits, see pricing.
These are the files of the workflow above, exactly as the models produced them, unretouched. The starting point is shown when there is one.
Starting point
Render
People who want to try the sweater portrait trend on their own photo for a story, a post or an autumn profile picture, without going through ChatGPT.
A sharp portrait of you or of someone who agrees, facing the camera with the eyes clearly visible, framed head and shoulders, and a description of their hair in the prompt (colour, length, texture). The photo is optional.
Upload a sharp portrait with your eyes clearly visible
The whole portrait rests on the gaze, since the mouth ends up hidden. Use a front-facing, well-lit photo with no sunglasses and no hair over the eyes. The demo starts from a plain white T-shirt against a beige wall.
Describe your hair in the prompts
Both prompts mention long dark brown hair, the demo woman's. Replace it with yours. If your hair is long, add "long hair falling over the shoulders, outside the collar": on the demo, without that line, it came out short.
Pick the sweater colour
The colour is a few words in the prompt: "deep saturated burgundy" for the burgundy one, "rich forest green" for the green one. Swap them for the shade you want, and keep the sentence that says only the knit is in colour.
Check the hands and the framing before you post
On the demo, fingertips poke out of one sleeve in the burgundy render, and a bare hand touches the cheek in the green one. If one render does not work for you, rerun only that one for 19 credits.
The same canvas, tuned for different needs: each variation is two or three lines of prompt away.
Replace "deep saturated burgundy" with the shade you want, for example "deep violet", "warm cream" or "bright red". A cream sweater will not stand out much on a light grey background, so use the charcoal one instead. Not tested.
Replace "the woman" with "the man", "her" with "his", and describe his face instead of the long dark brown hair: beard or clean-shaven, glasses, haircut and hair colour. Not tested on the demo.
Upload a photo where both people appear and write "the man and the woman in this photo", each in their own colour sweater, sleeves pulled up over the mouth. Describe each face in the prompt. Not tested.
Replace "chunky cable-knit sweater" with "oversized hoodie" and keep the long sleeves over the hands. You lose the knit texture, which carries a good part of the trend's look. Not tested.
Replace "light grey seamless studio background" with a colour, for example a pale blue, and remove "the background" from the list of things in black and white. The sweater will no longer be the only colour. Not tested.
On the demo, both renders cut the hair into a bob. Add "long hair falling over the shoulders, outside the collar" right after the face description so the strands stay visible over the knit. Not tested.
Start from the burgundy render and plug in a Video node: describe the hands slowly pulling the sleeves up to the mouth, eyes fixed on the camera. The template does not include this node. Not tested, and paid separately.
The prompts describe the demo woman's hair. If yours is blonde, red, curly or short, change those words, or the model may give you someone else's hair.
With the mouth hidden by the knit, the gaze is almost all that is left to recognise you. With sunglasses, a strand of hair or a blurry starting photo, the model guesses your eyes.
On the demo, the green render does not rest the chin on the forearms, a bare hand touches the cheek and a knee shows at the bottom left. The pose in the prompt is a direction, and the model follows it loosely.
Both demo renders cut long hair down to a bob. If people know you with long hair, they will notice straight away. Look before you send, and rerun if needed.
No. The demo starts from a woman in a white T-shirt, and the model invented the cable knit, the turtleneck and the pose. What you wear in your photo does not matter, only the face and hair are used.
On the demo, yes for the eyes, eyebrows, nose and freckles, in both renders. The mouth is hidden by the knit, so you mostly judge the upper face. The hair, though, got much shorter. Compare at full size before posting.
On the demo, both renders cut long hair down to a bob, despite "the same long dark brown hair" in the prompt. We do not know why. Add "long hair falling over the shoulders, outside the collar". We have not tested it.
38 credits, 19 per render on Nano Banana Pro at 2K. If one render does not work for you, rerun just that one for 19 credits.
Yes, by switching the node to Nano Banana 2.1 at 1K, which costs less per render and gives smaller pictures. We did not test it on this template. The demo runs on Nano Banana Pro.
1792 x 2400 pixels, portrait, in a 3:4 ratio. Enough for a story, an Instagram post or a print of about 15 x 20 cm.
We do not know who started it. The "ChatGPT Sweater Portrait Trend" guide on viralvideoai.app, updated on 13 November 2026, quotes MakeViral's notes: a first prompt post on TikTok on 9 November 2026, then posts on 11 and 12 November. It is our only source. Imaginode has no link with the site.
No. The trend carries ChatGPT's name, but this template runs on Nano Banana Pro, right in the Imaginode canvas. Imaginode has no link with OpenAI.
The photo is optional in the canvas, but without it the model invents a person. For it to be you in the sweater, you need your portrait.
ChatGPT sweater portrait trend, made from your own photo
Use this workflowFree account, no card required. The workflow opens pre-filled in your canvas.
Nano Banana Pro
Flux Kontext
Seedream 5 Pro