Recently, Nano Banana has exploded in popularity, and its character consistency ability completely exceeds that of models such as GPT and Flux Kontext, etc. Today, we will share with you a couple of Nano Banana'sPrompt wordTips.
1.Portrait IP
With Nano Banana's superb character consistency, we can easily create multi-angle images of characters, which is a great feature for making portraits of Lora.
Tips:
Based on the original image provided, create a set of 9-panel IP images of different angles for this. Requirements include: front, side, back, two 45-degree angles looking down and up and two additional dynamic poses. The overall style should be consistent, with attention to detail to ensure that character features such as clothing color and hairstyle are accurately reproduced at every angle.
2. Homemade handmade models
nano bananaYou can easily make the picture generate hand puppet model, even if you give a half body portrait, it will automatically help you to make up the body and generate a full body model hand puppet.
Cue word: Generate a realistic hand puppet model with the image of the character provided by the user. The base of the hand puppet is a round transparent acrylic, and there should not be any text on the base. Behind the artifact is an artifact box, on which is printed the original drawing of the character provided by the user. The figure and box are placed on a computer desktop, and the computer monitor is showing the modeling process of the figure in 3ds max software. The artifact should occupy the main body of the screen. Click on the body paragraph in the editorial area, and then click on one of the contents of the clipboard, which will have the effect of formatting brush.
However, there are times when the bust auto-completes the body, creating a disproportionate situation, like in the two pictures below where the legs are noticeably shorter. You can draw cards multiple times to fix this.
You can also fix this by changing the prompt word.
Tips:
Generate a realistic hand puppet model with the character image provided by the user, if the user provides a half-body image use the half-body image to create, do not expand or add to the character image, the base of the hand puppet is a round transparent acrylic, do not have any text on the base. The base of the figure is a round transparent acrylic with no text on the base. At the back of the figure is a figure box with the original drawing of the character provided by the user. The figure and box are placed on a computer desktop, and the computer monitor is showing the modeling process of the figure in 3ds max software. The artifact will be the main part of the image.
Now nano banana no longer complements the image when generating the model, so the generated model avoids the problem of disproportionate scale.
3. Dressing up
With nano banana we can easily extract all the clothes of the character.
Cue word: laying out a woman's clothes, headdress, shoes, etc., all flat within a single floor plan
Here's another woman wearing the clothes she just extracted.
Tips:
Keeping the look and pose of the woman in Figure 1 unchanged, completely remove the clothes of the woman in Figure 1 and replace them all with the clothes, shoes and accessories of Figure 2
Problems came up though, while the woman's clothes were replaced, obviously the previous clothes were still there and I didn't get any better results after drawing the card a few times
So, I think it might be that women have too many clothes, and wouldn't it be better if I found one with fewer clothes.
Then I got something like the result below, which is kind of wearing, but not as well as I would have liked, I wanted to see a full-bodied figure, to be able to see the dress and the shoes.
So the cue has to be modified to add the cue to show the full body
Cue word: Keep the woman's appearance in Figure 1 unchanged, replace all of the woman's clothes, shoes, and accessories in Figure 1 with those in Figure 2, show the woman's full-body figure, and strike a natural pose
4. Characters directly change clothes
The Nano banana also allows you to dress up your character directly.
Cue word: Have the woman on the left cosplay the woman on the right, so that the woman on the left has the same costume, makeup, and props as the woman on the right.
Here's the generated result, the woman's costume on the right didn't quite come over, but the overall style is close enough that probably a few more card draws would have yielded better results.
5. Re-lighting
Using cue words you can directly lighten your photos to improve the artistic effect of your pictures.
Cue word: Keep the characters consistent and generate studio portraits for the people in the picture. Black background, Rembrandt lighting. Side light, bust.
6. Generate 9-gang comic strip
Utilizing superb consistency and inference, we provide an image that can easily generate a set of comic strips.
Tips:
Based on the original images provided, please generate a set of 9 frames to tell a coherent and engaging story. Each frame should be centered around an original character and advance the storyline through different perspectives and scene changes.
Maintaining a consistent art style and making sure to keep the colors and emotions consistent throughout the series allows the viewer to feel the characters grow and the story develop.
If no theme is specified, Nano banana will automatically generate a comic strip with a random theme, if we want to control it precisely, we need to specify the theme by ourselves.
7. Generate coherent movements
Some coherent action pictures are often needed when making first and last frame videos, and now they can be easily realized.
Tips:
Based on the original images provided, a set of 4 images is generated that are a coherent set of actions.
8. Changing the age and size of human objects
Changing looks and age these also fall into the category of being base fucks.
Cue word: Please adjust the people in the image to look 60 years old, keeping the original poses, expressions and facial features intact while naturally adding a sense of age, including wrinkles, sagging skin and appropriate graying of the hair.
Make a woman fat.
Tips:
Please adjust the body type of the characters in the picture to the obese big fat model, the face is rounded and bulging, the neck becomes thick, the shoulders and back are wide and thick, the abdomen is greatly bulging, the limbs are thick and strong, and the overall obvious fat, but must strictly maintain the original posture, movement, expression, facial features, clothing details and picture composition unchanged, only natural body expansion, avoid distortion, distortion, or gesture change.
9. Beautify your photos
One-click beauty function is online
Tips:
AI beauty treatment for adding exquisite daily makeup to characters: create a flawless but not fake base makeup with a soft-focus filter feeling; use low-saturated earth colors for eye makeup, slightly lengthen the end of the eyes to enhance eye charm; add natural false eyelashes and clear inner eyeliner; use oblique sweeping contouring for blush to enhance the sense of three-dimensionality; use mirror-textured milk-tea or coral-colored lipsticks for lips; use well-defined eyebrows with an elegant curvature; the overall style is The overall style tends to be urban light mature or Korean idol sense, the picture is bright and transparent, highlighting the beauty of the character's advanced makeup, while maintaining a sense of reality.
10. Motion Migration
Even given a very abstract action picture, Nano banana makes it easy to migrate the character's movements.
Tips:
Please generate a new image by reconstructing the image of the character in the image on the left according to the gestures and movements of the character in the image on the right: i.e., keep the external features of the character on the left (e.g., the face, clothing style, body type, etc.), but make him/her perform the movements of the character on the right. Please make sure that the movements are natural and proportional.
The results you get are still very good and can be said to be perfectly restored.
