image creation and analysis

Generative Fill with Adobe Photoshop

jbaum02s — Wed, 31 Jul 2024 12:37:27 +0000

Generative Fill with Adobe Photoshop jbaum02s Wed, 07/31/2024 - 07:37

Adobe Photoshop features Generative Fill capability, that allows users to prompt Photoshop to create or update an image with GenAI. Photoshop uses the Adobe Firefly Image Model to create richer, more lifelike images, that help to bring life into your photographs.

Enhance Images with Generative Fill

Demonstration of using the Generative Fill tool in Photoshop.

With a single prompt, users can create lifelike images.

To get started, open an image you would like to edit.

Select the rectangle tool and select the area of the image that you want to add an element. Your selection size and shape will influence the results of the generative fill.
In the Contextual Task Bar under the selection, select Generative Fill.
Write a prompt to add or update elements in your image.
Choose your favorite variation in the Properties panel.

Visit the Next-Level Generative Fill for more information and sample files.

Remove Image Elements with Generative Fill

Demonstration using Generative Fill tool to remove image elements.

The Generative Fill tool also allows users to remove image elements.

Select the rectangle tool and select the area of the image that you want to remove an element. Your selection size and shape will influence the results of the generative fill.
In the Contextual Task Bar under the selection, select Generative Fill.
Write a prompt to remove elements in your image.
Choose your favorite variation in the Properties panel.

Other AI Powered Photoshop Tools

Remove Tool: Replace unwanted areas in your image with similar content by simply brushing over what you want to remove.
Curvature Pen Tool: Draw straight lines and smooth curves intuitively, and then easily modify them.
Match Font: Find matches for fonts you love in your images or photos.
Preserve Details 2.0: Resize images without introducing distortions to important details and textures.
Sky Replacement: Change the sky in your image in a few clicks and Photoshop will automatically mask and blend the replacement.
Object Selection tool and Refine Edge: Create precise selections and masks with ease.
Neural Filters: Reduce larger workflows to just a few clicks with a library of filters powered by machine learning.

Find more information on AI Photo editing with Photoshop.

Resource type

Prompt Category

image creation and analysis

Bias in GenAI

jbaum02s — Fri, 26 Jul 2024 14:58:25 +0000

Bias in GenAI jbaum02s Fri, 07/26/2024 - 09:58

When using different GenAI models, you may encounter biases stemming from unequal representation in the training data. This can manifest as gender biases and a lack of diversity in the subjects depicted in the generated images.

Some examples of bias found when using GenAI image models:

Prompt: engineer

All the images generated show white, male subjects.

Prompt: nurse

All the images generated show white, female subjects.

Prompt: doctor

All the images generated show white, male subjects.

Prompt: firefighter

All the images generated show white, male subjects.

Prompt: police officer

All the images generated show white subjects, with one female and three males.

Prompt: scientist

All the images generated show white subjects, with one female and three males.

Prompt: parent at home taking care of kids

Three of the images generated show white subjects, one image shows a hispanic mother. There is one male and three female subjects.

Mitigating Bias

Users can mitigate biases in GenAI models by employing diverse and inclusive language in their prompts to encourage a broader range of outputs. For example:

Prompt: photo of diverse group of people working together at a university

Prompt: black firefighter

Prompt: female doctor

Prompt: Hispanic male nurse

Post-processing generated content to ensure it meets diversity standards and providing feedback to developers can also help improve future versions. Custom training with more diverse datasets and using multiple models to cross-check results can reduce bias. Stay informed about the limitations of GenAI models and educating yourself and others about these issues to foster mindful usage and recognition of biases.

Resource type

Prompt Category

image creation and analysis

Midjourney Image Prompts

jbaum02s — Wed, 24 Jul 2024 15:52:26 +0000

Midjourney Image Prompts jbaum02s Wed, 07/24/2024 - 10:52

Midjourney generates images from natural language descriptions, called prompts, similar to OpenAI's DALL-E and Stability AI's Stable Diffusion.

"A prompt is a short text phrase that the Midjourney Bot interprets to produce an image. The Midjourney Bot breaks down the words and phrases in a prompt into smaller pieces, called tokens, that are compared to its training data and then used to generate an image. A well-crafted prompt can help make unique and exciting images." - Source: Midjourney website

Basic Image Prompts

A basic prompt can be as simple, and describes the important elements you would like Midjourney to include.

Prompt: a technology background

Advanced Image Prompts

More advanced prompts can include one of more image URLs, text, and various parameters.

Source: https://docs.midjourney.com/docs/prompts-2

Basic parameters

Aspect ratio: Change the aspect ratio of the image output by using one of the following:
--aspect or --ar followed by the desired settings e.g. --ar 16:9
No: negative prompting to remove certain elements from the image. e.g. --no people
Chaos: Change how varied the results will be; the higher the value the more unusual and unexpected. --chaos <number 0-100> e.g. --chaos 50
Quality: --q: how much rendering quality time you want spent e.g., --q 2 for high quality
Stylization: --s: how strongly Midjourney's default aesthetic style is applied; e.g., --s 750 for high stylization.

Check out the Midjourney Prompt Cheat Sheet for more information on parameters and options for prompting.

Prompt:
abandoned hospital at night, with dark, ominous clouds filling the sky above it. The hospital should appear derelict and decaying, with boarded-up windows and walls covered in graffiti. In the foreground, a group of teenagers could be shown breaking into the hospital, with fear and apprehension evident on their faces. The illustration could also include ghostly figures, draped in white sheets, pursuing the teenagers through the darkened hallways of the hospital, conveying a sense of terror and danger. The overall tone of the illustration should be eerie and foreboding, capturing the haunting atmosphere of the story. --ar 16:9

Prompt:
Fantasy art, a really beautiful aries girl representation, ultra realistic image of the aries constellation view , very detailled, very coloured, epic scene , 8k , ultra low angle shot, panorama, impactful picture, dynamic expressive force, cinematic shock, realistic lighting, depth of field, huge v 5

Prompt:
male black model is drowned in thousands doughnuts::2, very slender, mod 70's haircut, wearing oversize fututristic 70's sunglasses, wearing striped and color block slim fit roll neck sweater, with pleated and tapered pant, with chelsea boots, colors of garments should be in neutrals with bright accents, model should be sitting in on a futiristic couch aboard a space ship in the first class lounge. The interior of the lounge should be in colors of pink with camel and gold accents. Very futuristic. r, High Quality + hyperrealistic + photography + volumetric lightning + octane render + Clear texture + arnold render 3D + Super detailed + Megapixel Cinematic Lightning+ sunlight + same volume and mass of building in attached picture +Anti-Aliasing FXAA, TXAA, RTX, SSAO, Post Processing, Post-Production, Tone Mapping, CGI, VFX, SFX, Full color, Volumetric lightning, HDR, Realistic, 8k --ar 3:2 --s 750

Prompt:
a hyper realistic photo black and white close up, side view of a 50 year old husband and wife done in high contrast detailed photography on white background, wearing sunglasses and smiling looking confident with a bit of messy hair in a ponytail but beautiful

Prompt:
amazing epic chinese ancient theme water & ink style, Fantasy style martial arts style, A woman figure, Meet a killer, rainy day, fight, bamboo forest, lightly detailed, dynamic expressive, clean lines cinematic stunning realistic lighting and shading, vivid, vibrant 8k, octane render, unreal engine very detailed concept art, realistic, Cryengine, --ar 21:9 --q 2 --s 750

For more information visit Midjourney's Quick Start Guide.

Resource type

Prompt Category

image creation and analysis