INOLENS® Prompt Academy: AI property visualization by the rules
A free course in five short modules for brokers and designers. The lessons are public. After you sign in, you mark modules as read, take the final test and receive a PDF certificate with public verification.
- / 01 What AI may and may not do
- / 02 How to write a good brief
- / 03 Recognizing AI errors
- / 04 The photo as input
- / 05 Labeling and transparency
Module 1
What AI may and may not do
The line between an acceptable visualization and a misleading edit.
Allowed as a labeled visualization
Tidying up temporary clutter, furnishing a space with movable furniture, completing missing surfaces in a new build, proposing renewal of worn surfaces, and a full renovation of a visibly damaged building.
Every edit is a proposal of a future state. It must be labeled as an AI visualization and the original must remain available.
Protected elements
Walls, dimensions, ceiling height, windows, doors, staircases, the view and surroundings, the roof and facade of a finished building, and visible defects. Changing them would give the buyer an expectation that the viewing will not meet.
Remember
- A visualization shows a possibility, not the actual condition.
- Protected elements are not changed in any mode.
- A visualization does not repair defects in reality.
Module 2
How to write a good brief
The structure of a prompt that the model understands and that does not mislead.
Four parts of a brief
Name the room and the goal, for example a living room for sale. Describe the desired style or function. State what must stay unchanged. Leave out claims about the actual condition, such as “newly renovated”.
Common mistakes
A prompt that is too general (“make it nicer”), a request to change a protected element (“enlarge the window”), and hiding defects (“cover the damp”).
Remember
- Room, goal, style, what to keep.
- A short, specific prompt works better.
- No claims about the actual condition.
Module 3
Recognizing AI errors
What to look at when checking a result before publishing.
Check after the edit
Compare the result with the original using the slider. Check the number and position of windows and doors, wall lines, the view from the windows, the floor along the walls, and built-in fixtures.
A generative model can make an unwanted change even with a good prompt. The rules limit it but do not replace the check.
Remember
- Always compare with the original.
- Windows, the view and proportions change most often.
- Do not publish a doubtful result.
Module 4
The photo as input
The quality of the input determines the quality and credibility of the result.
Good input
A sharp photo with a level horizon, enough light and the whole space in the frame. Blurry, dark or heavily cropped photos lead to worse and less predictable results.
People and privacy
Do not use photos with recognizable people, including portraits on walls, for AI edits. INOLENS® rejects them before processing.
Remember
- A sharp, bright and level photo.
- No recognizable people.
- The whole space in the frame.
Module 5
Labeling and transparency
How to publish a visualization so that the buyer knows what they are looking at.
Label and original
Clearly label the visualization in the listing as an AI visualization and keep the original available. INOLENS® offers watermark profiles with different levels of transparency.
INOLENS® Standard and INOLENS® Verified
INOLENS® Standard means a result created under the declared rules without human review. INOLENS® Verified means that a designer reviewed and approved the specific result.
Remember
- Always label an AI visualization.
- The original should be one click away.
- INOLENS® Verified is a human review of a specific result.
Final test and certificate
- The test has 10 questions, two from each module, and takes at most 60 minutes.
- To pass, you need at least 80% correct answers; you get three attempts per day.
- The certificate is free and contains a unique ID and a QR code for public verification.
- You decide whether your name appears on the verification page.