Skip to main content

Generate Image Tools

Automatic detection

You can simply mention @Hana and describe your request. If the intent is clear, she will automatically run the relevant image tool for you.

Introduction

Hana can generate and edit images using the Generate Image tools.

Generate Image tools do not support:

  • Chat history context

Hana will create an image generation prompt herself which is optimized and considers everything she knows about your business.

Available Tools

ToolInvocation KeywordsDescriptionParameters
GENERATE_IMAGE_FROM_TEXTgenerate an imageCreate a new image from a text prompt.model_prompt (string, required)
model_negative_prompt (string, required)
file_name (string, required)
aspect_ratio (string, required)
GENERATE_IMAGE_FROM_IMAGEremix this imageGenerate a variation of an existing image.model_prompt (string, required)
model_negative_prompt (string, required)
starting_image_attachment_name (string, required)
starting_image_influence_strength (number, required)
file_name (string, required)
GENERATE_IMAGE_FROM_STRUCTUREpreserve structure of this image and generateGenerate an image while preserving the structure of a reference.model_prompt (string, required)
model_negative_prompt (string, required)
attachment_name (string, required)
control_strength (number, required)
style_preset (enum, required)
file_name (string, required)
GENERATE_IMAGE_FROM_SKETCHconvert this sketch to an imageTurn a sketch into a finished image.model_prompt (string, required)
model_negative_prompt (string, required)
attachment_name (string, required)
control_strength (number, required)
style_preset (enum, required)
file_name (string, required)
GENERATE_IMAGE_FROM_STYLEuse the style of this image and generateGenerate an image in the style of a reference image.model_prompt (string, required)
model_negative_prompt (string, required)
attachment_name (string, required)
aspect_ratio (string, required)
fidelity (number, required)
style_preset (enum, required)
file_name (string, required)
STYLE_TRANSFER_IMAGEapply the style of first image to second imageApply the style of one image to another.file_name (string, required)
init_image_attachment_name (string, required)
style_image_attachment_name (string, required)
model_prompt (string, required)
negative_prompt (string, required)
style_strength (number, required)
composition_fidelity (number, required)
change_strength (number, required)
seed (number, required)
SEARCH_AND_RECOLOR_IN_IMAGErecolor this object toSearch for an object in an image and recolor it.model_prompt (string, required)
model_search_prompt (string, required)
model_negative_prompt (string, required)
attachment_name (string, required)
file_name (string, required)
SEARCH_AND_REPLACE_IN_IMAGEreplace this object withSearch for an object in an image and replace it with another.model_prompt (string, required)
model_search_prompt (string, required)
model_negative_prompt (string, required)
attachment_name (string, required)
file_name (string, required)
OUTPAINT_IMAGEoutpaint this image to haveExtend an image in any direction.model_prompt (string, required)
left (number, required)
right (number, required)
up (number, required)
down (number, required)
creativity (number, required)
attachment_name (string, required)
file_name (string, required)
ERASE_IMAGE_BACKGROUNDerase background of this imageRemove the background from an image.attachment_name (string, required)
file_name (string, required)
UPSCALE_IMAGEupscale this imageImprove image quality or resolution.attachment_name (string, required)
file_name (string, required)
IMAGE_INPAINT_WITH_MASKinpaint this image with a mask to addEdit an image using a mask to specify the area to change.model_prompt (string, required)
model_negative_prompt (string, required)
attachment_name (string, required)
file_name (string, required)

How To Use

  1. Invoke the Tool: In your Google Chat, use an invocation keyword like generate or recolor to begin creating or editing an image. These tools leverage AI to visualize or modify what you describe.
  2. Specify Your Text Description: Clearly describe the image you want to generate or edit. Be as detailed as possible to ensure the AI can accurately produce the image you envision.

Example:

generaterecolorreplaceoutpaintupscaleeraseinpaintstyle
generate an image of a serene beach at sunset with palm trees and gentle waves.
@Hana generate an image of a serene beach at sunset with palm trees and gentle waves
  1. Receive the Generated Image: Hana will process your request and generate an image based on your description. The image will be shared in the chat for you to view and use as needed.

Sample Inputs and Outputs

generate
generate an image of a polar bear eating ice cream
@Hana generate an image of a polar bear eating ice cream

Image generation example: polar bear eating ice cream

generate a logo for a cafe called Double Shots. Keep the logo minimalistic, like Google but capture the essence of coffee.
@Hana generate a logo for a cafe called Double Shots. Keep it minimalistic and capture the essence of coffee

Image generation example: Double Shots logo

New Image Generation Flow!

NOTE

Due to a limitation in dependent systems that Hana uses, Hana will attach the generated image from your account, but this allows Hana to consider images in history like natural conversation and you can easily make edits to images or follow up on image changes or generation considering previous images

Recoloring an Image

Recolor: input image (before)

Recolor: output image variant 1 (after)

Recolor: output image variant 2 (after)

Recolor: output image variant 3 (after)

@Hana recolor this object to teal

Replacing Objects in Image

Replace object: input image (before)

Replace object: output image 1 (after)

Replace object: output image 2 (after)

@Hana replace the chair with a sofa

Replace Curtains color

From this

Recolor curtain: input image (before)

To this

Recolor curtain: output image (after)

@Hana recolor the curtains to white

Replace Wooden floor to Marble floor

From this

Recolor wood: input image (before)

To this

Recolor wood: output image (after)

@Hana replace the wooden floor with marble flooring

Image generation from text

from text

People in restaurant: input image (before)

to image

People in restaurant: output image (after)

@Hana generate an image of people talking in a restaurant

Replace Color

from this

Recolor people in restaurant: input image

to this

Recolor people in restaurant: output image

@Hana recolor the table to black

Replace Object

from this

Replace object in restaurant: input image

to this Replace object in restaurant: output image

@Hana replace the coffee cup with a teapot

Hana May Ask You to Define Areas to Edit - Using a Mask

from this

Image edit with mask: input image

to this

Image edit with mask: output image with crown

Invocation Screenshots

Mask edit invocation prompt example

@Hana make her wear a crown

Mask selection request in Hana

Hana dashboard: image mask editing page

Hana dashboard: image mask editing page (2)

Hana dashboard: image mask editing page (3)

Mask edit final output

Troubleshooting

  • Vague prompts: Add subject, style, lighting, and composition details for better results.
  • Edits not applied: Ensure you’ve provided or selected a mask for the area to change.
  • Aspect ratios: Specify square, portrait, or landscape if composition matters.