What is it?
A generative AI bot trained to engage in open-ended conversations. It uses deep learning to generate human-like text and was trained on a massive amount of text from the internet, forming its broad knowledge base.

Unlike most previous language models, ChatGPT can engage on almost any topic. It’s capable of converting typed text requests into an understandable response. Think of it as a search engine you can have a basic conversation with and instruct.

What is it good for?
Producing and editing copy, generating ideas, rewriting text for understanding, clarity, or tone of voice, rapid search of information, writing and editing code.

How not to use it
While highly capable, ChatGPT has limitations. It’s only trained on data until 2021, so its knowledge is not fully current. It also does not have access to the live internet. If you’re doing something important, fact-check what it’s produced.

The software doesn’t fully understand what it’s generating, so it often won’t appreciate why information is wrong. For example, if it’s writing code, it will often repeat the same mistake, even after prompting with the specific error. Don’t use it to create anything final, only to get initial ideas and help make more considered decisions.

Check it out here.

What is it?
Adobe has been integrating AI tools into its programmes and now offers a suite of different AI-powered functions. Generative fill applies AI that can intelligently fill missing or unwanted areas in an image, based on the surrounding content.

What is it good for?
It can remove objects, people, or defects from images seamlessly. If it doesn’t work the first time, try to rephrase or think of a different way to say the action you want it to take.

How not to use it
Generative Fill isn’t great at creating something from nothing, so don’t expect it to come up with an image to use as a final asset (without you doing a fair bit of fixing). It’s also not great at working at a large scale either, so keep an eye out for pixelation or odd AI inconsistencies when you’re generating.

What is it?
Illustrator’s Icon Generator is another helpful tool which takes text input and outputs a vectorised image.

What is it good for?
For small icons that aren’t the main focus of concept work. Input text describing what you want to generate, then select one of the generated vectors.

How not to use it
It’s best at creating very simplistic illustrations which you will be required to clean up after generation. Usually, you get a more versatile result making the lines yourself from scratch. Don’t use it to make key visual assets.

What is it?
Adobe Stock Library Image Search takes pictures as input and finds similar options within the library.

What is it good for?
This function is not limited to just Adobe Stock, but is very useful in searching the library based on a reference image used in your artwork. Simply drag and drop the reference image in and it’ll search for visually similar options.

Check them out here.

What is it?
It generates photorealistic images from written prompts. Users submit text prompts through a Discord bot interface and Midjourney generates images matching the description (based on its knowledge). It is one of the leading publicly available text-to-image generators, along with DALL-E 2 and Stable Diffusion.

What is it good for?
It’s good at creating images with different artistic styles, from photorealism to abstract art (if those styles are included in the original text prompt). The quality and coherence of the generated images is often remarkably high compared to earlier text-to-image models. There’s a free plan, it's useful to note that all images generated via this will be publicly available.

How to use it
There are plenty of online tutorials on how to use Midjourney that are worth following. However, simply put, enter text describing what you want to achieve and Midjourney will generate four images at a time. Learn how to use the parameters to get the most out of it – articles like this go into more depth.

How not to use it
Midjourney shouldn’t be used to replace human skill and creativity. Don’t generate with a final design in mind – it should be used as a reference point/springboard only. Although it has undergone a lot of development since its launch, it is still prone to making fundamental mistakes that you might not notice at first glance.

Check it out here.

What is it?
RunwayML is an online platform that provides tools and services for building and deploying machine learning models, with a focus on generative AI applications like text production, image creation, and video editing.

Their goal is to make cutting-edge AI technology more accessible to developers, designers and creators who want to leverage generative AI’s capabilities.

What is it good for?

Some key features of RunwayML include:

• Model Training: Users can train custom machine learning models on their own data using RunwayML’s distributed training infrastructure.

• AI Tools: RunwayML provides tools and interfaces that allow users to create text, images, videos etc. using state of the art Generative AI models without the need for coding expertise.

• Model Deployment: Trained models can be easily deployed and served through RunwayML’s cloud infrastructure – or locally.

• Collaboration Tools: Teams can collaborate on AI projects, share models, and work together using RunwayML’s collaboration features.

• Integration Support: RunwayML models and tools can be integrated into existing apps and workflows.

• Educational Resources: RunwayML offers tutorials, documentation and educational resources to help users learn about AI and generative models.

How to use it
RunwayML requires a low level of skill to use. The settings are limited but you can get a rapid result, depending on the model and what it’s being used for. We’ve seen some success with the image-to-video model, allowing us to give movement to assets used as a part of initial concepts, and also when using the image-to-image model to apply different textures to a base image concept.

How not to use it
Because it’s limited there’s only so much you can do, and there’s not much finessing that can be done. Images and videos generated tend to be poor quality or resolution – so keep your expectations low.

Check it out here.

What is it?
Topaz Video AI is a video upscaling and enhancement software that uses machine learning algorithms to improve the quality of video footage. It can upscale low-resolution videos to higher resolution, reduce noise and artefacts, sharpen details, and enhance overall video clarity. Topaz Photo AI is a powerful image-processing software that uses AI to enhance and edit photos.

What is it good for?
Both Topaz Video AI and Topaz Photo AI are designed to enhance and improve the quality of digital media in ways traditional image and video processing methods cannot. Both can be accessed through installable apps. Unfortunately, there’s no free option, but they do get updated fairly regularly.

How to use it
There are no text input options, just settings to test and play with – most of the options are adjustable and have good annotations describing their purpose. Topaz Video AI is relatively effective at upscaling footage and removing noise in 4K videos and Topaz Photo AI is useful for finishing low res 3D renders that are time-critical.

How not to use it
Don‘t go overboard and set all the options to 100% – less is usually more with these applications. If you go too far with something like noise removal, the outcome can appear unnatural. Remember Topaz can only do so much, if the source image/video is too low quality it won‘t be able to fix all the issues.

Check it out here.

...

2LK is a Creative Lives in Progress brand partner. Every year, we partner with like-minded brands and agencies to support our initiative and keep Creative Lives a free resource for emerging creatives. To find out more about how you can work with us, email [email protected]