I made a model of an aquarium, the walls of former boxes 0.1 Ni1.5. I am sure that for thin-shaped products, the Ni coefficient is not important. It turns out to be just transparent glass, if you assign a COLOR. I noticed that for such transparent glass models it is important not to glue the surfaces together. Otherwise, unpleasant grains and sparkling ripples appear (sometimes, not everywhere). How to fix it? You need to turn on the magnet, glue the four walls, and align the Model with a tool in the column. Then select all four walls and reduce everything by 1 mm. Group the result again so as not to accidentally shift it. For the viewer, such a change in the size of the aquarium walls is imperceptible, but a tiny gap appears between the shapes, now there is no grain anywhere. Inside the aquarium, I placed the "water" from the former box with the parameters MTLd 0.1 and Ni 1.1. I precisely inserted it into the center, aligned it and again slightly reduced the size by 2 mm to get a microscopic gap between the water and the walls. I inserted an image of a fish on a transparent background inside the water, placed it on a thin box, and made the other sides invisible. The fish should have overhead lighting in the photo so that it looks truthful and impressive. I put one hemisphere inside the water, and several weak hemispheres outside the aquarium to illuminate the entire scene. I must say right away that in the 3D view, the fish is in the center of the aquarium, and the offsets begin on the render. Because the important parameter of optical refraction Ni comes into play. Next, I will talk about him.
The scene looks like this, a fish in the center of the aquarium, the image on the box is perpendicular to the viewer's line of sight.
