The pumpkin head photoshoot made from your full-length photo: a carved pumpkin for a head, your own clothes, on an autumn road or all in black. Workflow visible here.
Real workflow: a couple standing on a park path in autumn, she in a cream sweater and light jeans, he in a rust overshirt and dark jeans, then the same couple holding hands on a road lined with orange trees, a carved pumpkin on each head, and sitting in black in a pumpkin patch at dusk, pumpkins lit up. Blonde strands show under the woman's pumpkin, and the couple share one hay bale instead of two.
At a glance
The pumpkin head photoshoot has come back every autumn since around 2020: a real hollowed-out pumpkin on your head, full-length shots, the whole set posted as a TikTok slideshow. This year it picked up again at the end of September, often as couples dressed in black. According to pumpkinheadphotoshoot.com, one of those posted on 30 September 2026 had reportedly passed 12 million views by 20 October. This workflow starts from your full-length photo and makes two images: the couple holding hands on an autumn road, then sitting in black in a pumpkin patch.
Doing it for real means finding an 8 kg pumpkin wide enough to fit over your head, scooping it out, cutting holes you can actually see through, then holding the pose without tripping. Here the pumpkin is generated and your clothes stay yours. It also works as a dry run before a real shoot, with Halloween falling on Saturday 31 October 2026. In the demo, bodies and outfits carry over well. Two misses: a few blonde strands under the woman's pumpkin, and one hay bale for two.
For a full costume rather than a pumpkin, see the Halloween costume. The white sheet and sunglasses look is the ghost photoshoot, and for something scarier, the werewolf transformation. Both renders cost 14 credits.
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
Couples and friends who want to do the pumpkin head trend for Halloween without carving a pumpkin, people planning a real shoot who want to test the pose and outfits first, content creators putting together an autumn slideshow.
A full-length photo, head to shoes, of you or of people who agreed, and your clothes described in the prompt. The photo is optional.
Use a full-length photo
Head to shoes, standing, nothing in front of your legs. The model copies the bodies and the clothes: whatever the photo cuts off, it makes up.
Describe your outfits in the prompt
The road prompt names every piece the demo couple wears. Swap in your own, or the model may slip a cream sweater in where you had a coat.
Pick the road or the field
The road keeps your everyday clothes. The field dresses you in black and lights up the pumpkins. Run both, or just one node for 7 credits.
Check where the pumpkin meets the shoulders
Zoom in on that seam: hair can show. If it bothers you, add "hair completely hidden inside the pumpkin" and rerun that node.
The same canvas, tuned for different needs: each variation is two or three lines of prompt away.
The road keeps the photo as you took it, clothes included, with warm golden backlight. The field is the version doing the rounds this year: the couple in black, sitting, pumpkins lit by a candle. On the road people recognise you by your clothes; in the field, by your shape.
Planning to do it for real before 31 October? Run the template first with the outfits you have in mind. You'll see whether black works against a field, whether the seated pose holds up, and roughly what size pumpkin to buy.
Replace "the two people" with "the person from the photo" and drop the held hands. A solo shot works well on the road: one figure in the middle, trees on both sides. Describe your own outfit instead of the couple's.
Three or four friends side by side, each with a pumpkin: write the exact number of people and one line of clothes per person. If your photos are separate, put them together first with the group photo. Not tested here, and everyone has to agree.
A dog sitting between you with a small pumpkin on its head, or at your feet in the field. Give its breed, its colour and a pumpkin to scale, or the model hands it one that's far too big. This version wasn't tested in the demo.
The trend is all about the full figure: pumpkin at the top, shoes at the bottom. With a cropped photo, the model makes up the lower half, and the jeans or boots no longer look like yours.
The road prompt describes a cream sweater, a rust overshirt and brown boots. If you don't change those lines, the model mixes those clothes with yours. Describe what you're actually wearing.
Even with the face hidden, a body shape and an outfit are enough to recognise someone. Only use people who said yes, and never a famous person.
No, it's optional. Without one, the model invents the people from the prompt. With your photo, it keeps your shapes, your heights and your clothes, which is the whole point of the trend.
Yes. The demo is a couple, but you only need to rewrite the prompt: "the person from the photo" for a solo shot, the exact number of people for a group, and for a dog, say how big its pumpkin should be. Those versions weren't tested here.
Yes, it happened in the demo: blonde strands come out under the woman's pumpkin even though the prompt said no visible hair. Rerun the node, or ask for a bigger pumpkin that comes down to the shoulders.
14 credits, 7 per render on Nano Banana 2.1 in 1K. Rerunning a single image costs 7 more credits, or 9 if you switch to 2K.
Because it keeps bodies, poses and clothes intact while the whole setting changes. The face is hidden under the pumpkin, so a pricier model picked for copying facial features would add nothing here.
In 1K the image is 896 x 1200 pixels, vertical 3:4: fine for a slideshow or a story, too small for a print. To print, switch the node to 2K, 9 credits per render.
The model went for one wide bale that fits the couple side by side. To get two, write "two separate hay bales with a gap between them, one per person" and rerun the field node.
A glitch in the field render, at the top of the frame. Crop slightly, erase it in a photo editor, or rerun the node for 7 credits.
The pumpkin head photoshoot with AI and your own photo
Use this workflowFree account, no card required. The workflow opens pre-filled in your canvas.
Nano Banana 2.1
Flux Kontext
Seedream 5 Pro