AI Questions & Processors
Photo questions with AI prompts built from earlier answers, reference photos, In-line and Post-AI processors, and background removal.
Survey AI questions explained: Photo Capture with AI dynamic prompts, reference photos, In-line and Post-AI processors, and automatic background removal.
AI Dynamic Prompt questions
The three photo types also come in an AI Dynamic Prompt version. The photo or image the guest provides is transformed by an AI prompt that can include their earlier answers, so each guest gets a personalized result.
Photo Capture (AI Dynamic Prompt)
Take a selfie and transform it right away. Add several named prompts, each with its own aspect ratio, and optionally let guests edit the prompt text.
Upload Photo (AI Dynamic Prompt)
Transform an uploaded photo, including one sent from a phone by QR code, using a prompt built from earlier answers.
Image Select (AI Dynamic Prompt)
Guests choose from predefined images, and each option can carry an AI prompt that uses their earlier answers. With more than one option, guests see them as a card grid; style it in AI Prompt / Image Option Card Grid.
Prompts can only refer to earlier questions of these types: Short Text, Long Text, Single Choice, Multiple Choice, Dropdown, Number, and Yes/No. Insert a placeholder such as {Q1-Your favorite animal} from the buttons above the prompt box. Put the questions you want to reference first.
Reference photos for AI prompts
Every question type that has an AI prompt can use a reference photo that the AI looks at when it generates the result. This covers Photo Capture and Upload Photo (and their AI Dynamic Prompt versions), Image Select and Image Select (AI Dynamic Prompt) option prompts, the In-line AI Processor, and both Post-AI Processor prompts. Each Reference Photo field offers two sources: use a previous question's image answer, or upload your own reference photo, which is how it always worked.
Use previous question as reference
Pick an earlier Photo Capture, Photo Capture (AI Dynamic Prompt), Upload Photo, Upload Photo (AI Dynamic Prompt), Image Select, or Image Select (AI Dynamic Prompt) question. The guest's answer is sent to the AI as the reference photo, so every guest gets a different one.
Upload a reference photo
Upload one fixed image yourself, for example a style guide or a character, that every guest's result is based on. Existing prompts with uploaded reference photos keep working unchanged.
Image Select choice
When the previous question is an Image Select, a second dropdown appears. Choose Guest's selected image (the default) to use whichever image the guest tapped, or pick one specific image option by name.
Original image, not the AI result
The reference is always the question's original image, never its AI-transformed result. This applies to the main submit-time AI, the Post-AI Processor, and the In-line AI step.
To set one up:
- 1
Order the questions
Place the question that provides the image (for example an Image Select of themes) before the question whose prompt will use it.
- 2
Open the prompt
Open the question with the AI prompt and find the Reference Photo field.
- 3
Choose the source
Select Use previous question as reference, then choose the earlier question. For an Image Select, also choose Guest's selected image or a specific option.
- 4
Preview as a guest
Preview the survey, answer the earlier question, and check that the generated image follows the reference.
If there is no earlier question of a valid type, the previous question option is disabled with a short hint and uploading is the only option. AI options set on the selected question are ignored at this step; they still process when the survey is submitted.
A Photo Capture or Upload Photo question with no AI of its own is normally kept on the device and never uploaded. If another prompt uses it as a reference, it is uploaded so the AI can read it.
Use an Image Select of styles as the reference and a selfie as the source to get a different look for every theme. See the full recipe in Advanced AI: trading card.
AI processors
Processors are not questions guests answer. They take an image from an earlier question and run more AI on it. They are covered in depth, with a full recipe, in Advanced Use Cases.
In-line AI Processor
Pick a source question (an Image Select, Photo Capture, or Upload Photo question that comes before it), add one or more AI prompts, and choose how many images to generate at once, from 1 to 6. The guest sees the results straight away, taps the one they like. Several prompts appear as a card grid of styles, and a prompt can be editable by the guest. It ignores any AI prompt set on the source question. It works on its own; you do not need a Post-AI Processor.
Post-AI Processor
Runs silently after the guest submits the survey, so the guest never sees it. For each source question you choose one action: use the original input image as is, process the original image with a prompt, process the AI output image with a prompt, or remove the background. Use it to restyle or clean up a result that was already generated, or to remove the background (see Remove background). It is always placed after the other questions. It works on its own or together with an In-line AI Processor.
Each generated image uses AI credits, so keep the number of outputs modest at busy events. See Billing & Credits for how credits work.
Remove background
The Post-AI Processor can also remove the background from an image and output a transparent PNG. This opens up more creative AI results: Nano Banana (Google Gemini image models) cannot produce transparency, so generate the image first, then cut out the subject. Turn on Remove background for a source question and choose whether to remove it from the Original image or the AI output image. It is one action per source question, so it replaces the other three options for that question.
Original image
Cut out the subject of the photo the guest took or uploaded, or of the image they picked in an Image Select.
AI output image
Cut out the subject of an AI-generated image, such as a stylized portrait or a character. This is the default.
Transparent PNG
The result is a PNG with a transparent background that can be mapped onto an image placeholder, or used on top of other artwork.
Credits
One background removal credit is deducted from the event owner for each successful removal. Top them up on the account settings page; see Billing & Credits. Removals are listed in the Background Removal Log.
Guests can also remove the background themselves in the editor, see Design Editor. A great combination: generate an AI image of the guest, remove the background, and use it for stickers or cards. See Advanced AI: trading card.
Related guides
Link a survey to an event and map answers in Surveys, style the guest screens in Branding and colors, and review every answer in Manage Responses. AI settings for the event are in AI Features. For a complete example combining reference photos and background removal, see Advanced AI: trading card.