The robot in my imagined workshop is not a polished machine built for combat. It is a small maintenance unit with a chipped yellow shell, one narrow work light and a tool drawer that never quite closes. For a fictional game environment, I would first develop a still character concept showing how this robot belongs in a repair bay. The immediate goal would be an image that explains its shape and role, before anyone builds a model or plans animation.
I would give the robot a practical job: inspecting loose fittings beneath a row of old transport vehicles. That suggests a low body, broad wheels and a light aimed toward the floor. It also gives the wear marks a reason to exist. Scratches would collect around the lower shell and drawer edge, rather than appearing as random texture across every surface.
A compact machine with one clearly defined task makes a useful brief for Nano Banana 3. Its still-image generation and editing capabilities can help explore the shell, materials and workshop setting from written descriptions and visual references. I would use the resulting artwork as a concept discussion aid; it would not be a rigged model, an animation or a technically validated machine design.
Find the silhouette before the rust
My first sketch brief would describe a body roughly shaped like a rounded toolbox, carried on two large side wheels with a small stabilising wheel at the rear. The work light would sit on a short stalk rather than a human-like neck. I would keep the front simple, with a narrow inspection slot and a recessed panel that suggests access for maintenance.
The silhouette should remain readable when reduced to a small dark shape. A cluster of antennas, cables and tiny tools might make the concept look complex, but it could also hide the robot’s basic form. I would begin with only the shell, wheels, light and drawer. Additional details would be considered later if they explained something the first four elements did not.
For the first Nano Banana 3 image, I would request a three-quarter view on a plain workshop floor, with the robot occupying most of the frame. A neutral background would make the wheel placement and shell proportions easier to compare. I would avoid an elaborate environment until the character itself could hold attention without dramatic scenery.
Let the materials tell a small history
The yellow paint would be worn but not abandoned-looking. I imagine a machine that is still useful and regularly repaired, so bare metal could show along exposed corners while the central panels remain relatively clean. A replacement drawer front in dull grey might suggest that one component has been changed. That single mismatch would tell more than covering the whole robot in rust.
I would gather reference images of industrial paint, rubber wheels and a simple metal drawer, using material I own or have permission to use. Each reference would have a specific role. The paint image would guide the finish, the wheel reference would establish a practical texture, and the drawer would help describe the small imperfection that gives the robot character.
A material revision in Nano Banana 3 would preserve the main shape while adjusting the wear around contact points. I would check that the tool drawer still appears attached to the body and that the wheel treads remain plausible. If the generator added a decorative grille or an extra arm, I would remove it unless it served the maintenance job already described.
The work light should illuminate work
A glowing eye aimed directly at the viewer would make the robot feel like a character posing for a portrait. I would prefer the lamp to point down toward a loose bolt on the floor. That creates a relationship between the machine and its task. The robot can feel purposeful without a face, a smile or an exaggerated human gesture.
My lighting prompt in Nano Banana 3 would describe a small, warm work lamp against a cooler ambient workshop light. The bright area should stay close to the bolt rather than flood the entire floor. I would inspect the direction of the shadow under the robot and the reflection on its shell. If the light appeared to come from several contradictory sources, I would simplify the scene.
The bolt itself would remain secondary. It needs to be large enough to read as something being inspected, but not so prominent that it becomes the subject. I would avoid adding a labelled warning screen or a string of technical numbers. At this stage, a clear light beam and a believable point of attention are enough to communicate the action implied by the still image.
Move into the bay without losing the robot
Once the plain-background concept felt coherent, I would develop a second still image in the repair bay. A vehicle chassis could occupy the upper edge, with a bench and a few indistinct tools farther back. The robot would remain low in the frame, its lamp forming the warmest small accent. This setting would make its compact scale useful rather than merely cute.
The Nano Banana 3 environment study would be guided by the approved robot image, but I would not assume perfect consistency. The shape of the shell, drawer colour and wheel arrangement would need comparison with the earlier concept. If the machine changed too much, I would revise the environment study or keep the original concept as the clearer reference.
I would also resist filling every empty patch with pipes or cables. A quieter area of floor around the robot gives the eye a place to settle. The bay should suggest a working environment, not overwhelm the character with a catalogue of science-fiction decoration. One vehicle, one bench and a few purposeful shapes would support the scene more effectively.
Make the handoff honest and useful
My final presentation would pair the isolated character concept with the environment study and a short written description of the robot’s job. I would identify unresolved details, such as how the drawer opens or where the rear wheel sits. A single rendered view cannot answer every modelling question, and an attractive joint should not be treated as proof of a workable mechanism.
These Nano Banana 3 images would be starting points for further design work. A modeller could reinterpret the proportions, resolve hidden surfaces and build the actual asset in appropriate software. I would consider the concept successful if the chipped shell, imperfect drawer and downward lamp all support the same idea: a small machine that has spent years doing useful, unglamorous work. That is a stronger foundation than an impressive robot with no reason to inhabit its scene.
About the Author
