Hi Hugo, quick practical question. With AI projects generating so many images, prompts, variations and working files, how do you recommend organising everything and staying on top of version control? Do you have a folder/file structure you use for client or campaign work that keeps things easy to reference? Is there a specific way you name files? Without a good system things can quickly descend into chaos!!

Hi Hugo, is it ok to mix and match models when creating videos so you can take advantage of each models strengths? Or do you find it better to use just one to ensure you get more continuity? Thanks!

Hi Hugo, do you think about the emotional role of a product before creating the imagery? For example, whether it creates calm, status, energy, confidence, escape, control, intimacy, etc.? Or does this come from the brief?

Hi Hugo, I’m using your structure of Art Direction, Location, Characters, Styling and Shot Inspo for a product-led campaign concept I'm working on. Because the product itself is quite important to the idea, would you recommend adding a small Product Truth section (or something like this) to define things like photographic style for the product? Or would you keep that information inside the existing sections rather than adding another category? Just want to make sure I’m not overcomplicating the board or drifting away from your framework. Many thanks!

Hi Hugo, when using a client-supplied product image as a reference in a lifestyle or campaign scene, how do you control the product’s scale, proportion and location ...especially when it’s a new product the AI model won’t already recognise? My current workaround is: 1. Create the lifestyle scene first, without the product. 2. Generate a rough rotatable 3D version of the product using Figma Weave / Rodin 3D AI. 3. Place the 3D product into the scene at the correct size, perspective and position using a compositor node. 4. Write a detailed super prompt based on that composed image making sure there's a clear description of the product within the super prompt. 5. Re-render the final image in Nano Banana, using the comped image (with the 3D product render), the prompt and the original product image as the visual reference for product accuracy. This usually gives me accurate scale and placement, but I’d be interested to know whether there’s a cleaner workflow you’d recommend. Many thanks!