# AI Pose Changer > To change a pose with AI, upload a clear photo to an AI image editor and describe the new pose — for example "standing with arms crossed; keep the face, outfit and background the same" — or add a second picture and say "copy the pose from image 2". Kenerate's AI pose changer opens the editor with that instruction filled in on Qwen Image 3.0, a free-plan model, so you add your photo and press Create. For pose references to draw from, describe the pose with no photo at all. Hands and parts of the body that were hidden in the original are drawn by the AI, so check them before you use the picture. URL: https://kenerateai.com/ai-pose-changer Publisher: Kenerate AI (https://kenerateai.com) Last updated: 24 October 2026 ## Key facts - Works on: Your own photo, one you may edit, or your own character art - Ways to set the pose: Describe it, or add a pose reference image - What stays: Face, outfit, background and light — by instruction - Best input: Full-body or three-quarter shot, plain background - Pose references: Text only, no photo needed (9 credits on Qwen Image 3.0) - Default model: Qwen Image 3.0 · free plan · 9.9 credits per edit - Watermark: None ## Steps 1. Open the pose changer — Press “Add your photo”. Image Studio opens with an arms-crossed pose instruction in the prompt box. Sign in free for starter credits. 2. Upload your photo — Add a clear photo of yourself, or one you have permission to edit — full body or three-quarter works best. 3. Set the pose — Swap in prompt words from the pose table, or add a second image as a pose reference and say “copy the pose from image 2”. 4. Create and check — Press Create, zoom in on the hands and the outfit, run it again if anything drifted, and download without a watermark. ## In depth ### Which poses can AI do? Twelve poses and the words that get them Copy the prompt words into the editor in place of the default pose. The last column is the crop the pose needs: a model can't stand up a person whose legs aren't in the picture without inventing them. - Arms crossed: Confident standing pose, arms folded across the chest. Prompt words: "standing facing the camera with arms crossed and weight on one leg". Needs at least: Waist-up. - Hands in pockets: Relaxed, casual standing pose. Prompt words: "standing relaxed with both hands in the trouser pockets, shoulders down". Needs at least: Three-quarter (to the knees). - Hands on hips: Wide, assertive stance. Prompt words: "standing with hands on hips, feet shoulder-width apart, chin slightly up". Needs at least: Full body. - Walking: Mid-stride toward the camera. Prompt words: "walking toward the camera mid-stride, arms swinging naturally". Needs at least: Full body. - Sitting: Seated on a stool or chair. Prompt words: "sitting relaxed on a simple wooden stool, one foot on the rung, hands on the knees". Needs at least: Full body. - Leaning: Shoulder against a wall. Prompt words: "leaning one shoulder against the wall, arms loosely crossed, one ankle over the other". Needs at least: Full body. - Over the shoulder: Body turned away, face looking back. Prompt words: "body turned away from the camera, looking back over the right shoulder". Needs at least: Waist-up. - Hand on chin: Thoughtful portrait pose. Prompt words: "resting the chin on one hand, elbow on the table, looking at the camera". Needs at least: Chest-up. - Jumping: Both feet off the ground. Prompt words: "jumping with both feet off the ground, arms raised, joyful". Needs at least: Full body. - Kneeling: One knee on the floor. Prompt words: "kneeling on one knee, the other foot flat, one forearm resting on the raised knee". Needs at least: Full body. - Contrapposto: Classic figure-drawing stance: weight on one leg, hips and shoulders tilted opposite ways. Prompt words: "contrapposto stance, weight on the left leg, right knee bent, hips tilted, shoulders tilted the other way". Needs at least: Full body. - Action reach: Dynamic, stretched pose for comics and games. Prompt words: "dynamic action pose, lunging forward and reaching out with the right arm, strong line of action". Needs at least: Full body. ### How does an AI pose changer work? It's an instruction edit. Your photo and one sentence go to an image model's edit endpoint; the model redraws the person in the new position and repaints what the move reveals — the wall behind a lowered arm, the side of a jacket — while the instruction tells it to keep the face, clothes and scene. Pose control has a longer history in image models: ControlNet, published in 2023, showed that a model can be steered by a human-pose skeleton as well as by edges, depth or segmentation. Newer edit models take the pose straight from words or from a second picture. ### Pose, camera angle or motion: which tool do I need? Three different jobs that people often mix up: - Pose changer (this page): the body moves, the camera stays. Image Studio, one photo plus a sentence or a pose reference. - Camera Angle: the camera moves, the pose stays — front, side, back, high or low. 7.5 credits per view, up to 4 views in a run. - Motion Transfer: a still image copies the movement of a driving video. Video tools are priced per second. - Fashion Studio's “Pose & angle pack”: one garment on a model in Front, ¾ turn, Side, Back, Walking, Sitting and Detail shots — one image per pose. ### What can't an AI pose change do well? Know these before you rely on the picture: - Hands and fingers: the most common failure. Check them every time. - Hidden parts: anything not in the original — legs below the crop, the back of an outfit — is invented, not recovered. - Big moves from a tight crop: a headshot can't become a full-body jump without the model making up most of the person. - Physics: a pose that leans on a wall or sits on a chair needs that wall or chair; say “add a simple stool” if the scene has none. - Proof: an edited pose isn't a record of what happened. Don't use it as evidence or to show someone doing something they didn't. ### How do I write a pose prompt? Four parts are enough: - The body: “standing”, “sitting on a stool”, “walking toward the camera”, “kneeling on one knee”. - Arms and hands: “arms crossed”, “hands in pockets”, “right hand on the hip”. - Head and gaze: “looking back over the right shoulder”, “chin slightly up”, “looking at the camera”. - What must not change: “keep the face, identity, hair, outfit, background and lighting exactly the same; change only the pose”. ## FAQ Q: How do I change someone's pose in a photo with AI? A: Upload the photo to an AI image editor and describe the new pose, ending with what must stay: “standing with arms crossed; keep the face, outfit, background and lighting exactly the same”. Kenerate's pose changer opens Image Studio with that instruction filled in, so you add the photo and press Create. Swap the pose words to try walking, sitting, leaning or any pose from the table. Q: Is the AI pose changer free? A: Free to start. A new Kenerate account gets 25 credits on sign-up, and Qwen Image 3.0, the default model here, is on the free plan; a pose edit with one photo costs 9.9 credits, so the starter credits cover two tries. Nano Banana 2.1, Seedream 5.0 Pro and GPT Image 2.5 Sunburst use credits from one-time packs — there's no subscription. Downloads carry no watermark. Q: Can I copy a pose from another picture? A: Yes. Upload your photo first and the pose reference second, then write “the person in image 1 takes the exact pose of the figure in image 2; keep image 1's face, hair and clothes”. Qwen Image 3.0 takes up to 3 images, and two cost 10.8 credits. The reference can be another photo, a mannequin, a stick figure or a sketch — only its pose is copied. Q: What is an AI pose generator for artists? A: It makes a reference picture of a body in a pose you describe, for drawing, painting or 3D blocking. In Image Studio, describe the pose with no photo — for example “full-body figure drawing reference of a grey artist's mannequin, contrapposto stance, plain light-grey background, even studio light” — and press Create; on Qwen Image 3.0 that's 9 credits. Ask for the same pose “seen from the side” or use Camera Angle to get more views. Q: Will the face and outfit stay the same? A: They should, because the instruction asks the model to keep them, and current edit models are built for it — Alibaba's model card for Qwen-Image-Edit-2509 lists pose changes with identity kept. Clothing that was hidden in the original (the back of a jacket, a sleeve behind the body) has to be invented, though, so check it. If the face drifts, run it again or switch to Nano Banana 2.1. Q: Why do the hands look wrong? A: Hands are small, have many joints and are often half hidden, so image models get fingers wrong more than anything else in a pose change. Zoom in before you use the picture, run it again if a hand looks off, or pick a pose where the hands are in pockets, crossed or out of frame. Q: What's the difference between changing the pose and changing the camera angle? A: A pose change moves the person's body — arms, legs, head — while the camera stays where it was. Camera Angle does the opposite: the pose stays and a virtual camera moves around the subject (8 positions, 4 heights, 3 distances, 7.5 credits per view). For a character sheet, Character Turnaround AI makes front, right, back and left views in one run. Q: Can AI change the pose in a video? A: Not with a photo edit. For video, Kenerate's Motion Transfer tool takes the movement from a driving video and applies it to the character in a still image, and its LTX Control engine follows a clip's pose, depth or edges. Video tools are priced per second, so check the cost in the studio before you run one. Q: Can I change the pose of a group photo? A: One person at a time works best. Name who to change (“the woman on the left”) and say “everyone else stays exactly the same”. With several overlapping people the model has more to invent, so expect a few tries. Q: Can I change the pose in someone else's photo? A: Only in your own photo, or one the person has agreed to. A pose change still edits a real person's likeness, so share the result only with their consent and don't present it as something they really did. Never upload photos of anyone under 18, and don't edit photos of public figures; Kenerate's content rules apply to every edit. ## Related pages - Pose reference generator: https://kenerateai.com/pose-reference-generator - Change camera angle: https://kenerateai.com/change-camera-angle-ai - Character turnaround AI: https://kenerateai.com/character-turnaround-ai - AI motion transfer: https://kenerateai.com/ai-motion-transfer - AI fashion model generator: https://kenerateai.com/ai-fashion-model-generator - AI smile generator: https://kenerateai.com/ai-smile-generator - AI image editor: https://kenerateai.com/ai-image-editor ## Sources - Qwen-Image-Edit-2509 model card (Hugging Face): https://huggingface.co/Qwen/Qwen-Image-Edit-2509 - Zhang, Rao & Agrawala — Adding Conditional Control to Text-to-Image Diffusion Models (ControlNet): https://arxiv.org/abs/2302.05543 - Kenerate Image Studio: https://kenerateai.com/app/image