Start with a visible pose
Choose a pose from the six references above, or upload a reference of your own. Each built-in pose is a PhotoGenerAI illustration that becomes an actual image input when you select it. The AI pose generator reads that reference together with your instructions. You can begin with a standing figure, seated figure or a more active movement without arranging a digital mannequin first.
Look at the body direction and camera angle before choosing. A side-view running pose suggests a different image from a front-facing standing pose, even when the character and setting stay the same. Use a reference that already shows the framing you need. Include the whole figure when your finished image needs complete hands and feet, and leave some room around the body for a new scene.
The upload field accepts a single pose reference. If you have a contact sheet with several figures, prepare one pose as its own image before uploading. A single clear figure gives the generator a more specific instruction than a sheet of alternatives. Keep any objects that establish the action, such as a chair for a seated pose, visible in the reference.
Create a new character or bring your own
A character upload is optional. To create a new character, choose a pose and describe an adult fictional person, their clothing and the style of the finished image. You might want a painted explorer, an anime swordswoman or a fashion illustration with a particular silhouette. Add the setting and light once the character brief is clear. The pose reference supplies the starting body arrangement.
For an existing character, upload the character separately. The generator receives the character first and the pose second. Explain which facial features, hairstyle, clothing and proportions should remain. Give the pose image the narrower job of guiding body position and camera angle. This helps communicate that you want the character to perform the action rather than adopt the reference figure's appearance.
A clear character image makes your intent easier to express. Choose one where the face and important outfit details are visible. When a new pose turns the face away, compare the features you can still see rather than assuming that similar clothing proves the same identity. A front or three-quarter view is a useful starting point when keeping the face recognizable matters most.
Give the pose generator a concrete image brief
The editable description above starts with instructions that separate the character and pose references. Keep that image-role information, then replace the explorer and courtyard with your own idea. Describe the medium, clothing, background and lighting in ordinary language. A short visible detail, such as a red scarf or an evening street, gives the generator a clearer target than a long series of quality adjectives.
Match the description to the selected reference. If the reference is seated, an instruction to sprint will introduce competing directions. When you want a small change, name it directly: turn the head toward the viewer, open the left hand or keep both feet on the floor. For a large change in body position, choose another reference so the image and written brief agree.
Choose an image shape that leaves room for the action. Portrait framing works well for an upright full-body pose; a wider frame can help with an extended arm or running stride. Model and quality settings show their credit cost before generation. The default GPT Image 2 Low setting supports the shared free-credit allowance. Other settings use the same purchased-credit rules as the rest of PhotoGenerAI.
Compare the pose, then the character
After generation, compare the main action first. Check the torso direction, bent knees, raised hands and relationship to any supporting object. Then look at the camera view and body proportions. A convincing picture can still miss the pose you selected, so judge the result against your reference as well as its overall appearance. The AI pose generator produces an image guided by a reference, and individual limb positions can change.
Next, inspect the character. Compare the face, hair, outfit and distinctive accessories with the character reference. Check that both hands and feet are present when the brief calls for a full-body image. If a hand is meant to rest on a knee or hold an object, inspect that contact closely. These details often matter more than whether the background matches every word of the prompt.
To refine a result, change the missed instruction or replace the pose reference, then generate again. Continue editing makes the current result your character reference so you can select another pose for the next image. The ordinary download and private history controls remain available. Download versions you want to keep; the history follows PhotoGenerAI's existing retention rules.
