Observe what the AI is generating in your Scene and add props explicitly. Here is what we mean by that:
Imagine you want to shoot a image of a couch/sofa. So we might start with a single product image of the couch. The AI already knows that couches are usually in living rooms, but we add a simple prompt preset like “modern living room” which already gives the AI more hints on what to create:
We observe that the AI starts generating a living room and filling the space. The images are still a bit trist, boring and maybe a bit unrealistic. We see some shy attempts to add plants and most of the images have a carpet. So now, we add plants and carpet props in the scene where the AI generated them in the step before:
This worked well and we already get better results than in the step before! Notice that one image has a shelf on the left and also one image contains a coffee table (or, something like that at least?) Lets add both explicitly (the coffee table prop does not have legs, but don’t worry, the AI will take care of that):
Fantastic! We are already at a pretty high level of quality, and have to take a closer look. We see one image on the bottom right that there was a second couch generated, so we might add a couch there. However for this tutorial we might follow a different idea, and this is just a faint hint: there are frequent rectangles and sunlight in the background. Maybe we should try adding a window and add a prompt like “sunshine though window” to improve the scene?
We are pretty close now. One last thing that we can observe is that in two images there are items on the table. Lets put some books and a bowl on the table to add some life!
This looks great. Now, we can generate a few batches of images and pick one that we like. Great job!