Make AI generated Christmas cards in seconds

With the holiday season just around the corner, it's that wonderful time to adorn your surroundings with festive decorations and extend warm wishes to your loved ones through Christmas cards. If the idea of venturing outside into the chilly weather in search of the perfect cards doesn't appeal to you, fear not! You can craft personalised and artistic Christmas cards from the comfort of your cozy home, all with the help of your very own AI assistant!

If you're unfamiliar with the realm of AI-generated Christmas miracles, I recommend reading my earlier blog post to gain insight into the workings of AI artists. However, significant progress has transpired in the AI landscape since last year. Developers and artists alike have diligently pushed the boundaries, elevating models to novel horizons. These advancements involve extending capabilities to diverse domains, such as text-to-video or text-to-3D, and enhancing user-friendliness across various applications like text-to-image, image-to-image, inpainting, outpainting, and even custom finetuning.

But if you're not interested in a technical deep dive and just want to start creating, I'll also give an extensive list of online services to try out, some tips and tricks and some examples to get you excited!

Online Christmas miracles

For those who prefer to steer clear of coding tasks, several online user interfaces are available to kickstart your journey. I've categorized them into two groups based on whether you're open to spending some money or not:

Free services:

• Experience the cutting-edge capabilities of StabilityAI's latest model, SDXL-Turbo, directly on their website. This lightning-fast model surpasses the speed of your typing, boasting impressive efficiency. While its parameters have limitations, you can enhance control by importing the open-source model into other software, as detailed in the last section.
• Another noteworthy option is Dall-E, now in its third version. Accessible within ChatGPT, this service is user-friendly and available for free in Bing chat.
• If you're familiar with Canva, explore their free image generator, seamlessly integrated with their suite of services. This platform facilitates the entire Christmas creation process in one centralized location.
• For those seeking advanced features, venture into Adobe Firefly. Unleash your creativity with stunning images and unique functionalities, including generative fill. This is especially useful for when you want to do some in- and outpainting, like creating a Christmas card from your boss ;)

• The next tool, Microsoft Designer, is a versatile platform offering an array of tools. Beyond image creation, delve into designing intricate visuals and comprehensive brand kits. What sets it apart is the seamless editing capability that accompanies your creative process.
• You can also immerse yourself in the innovative offerings of Runway, a captivating company celebrated for its diverse models that skillfully transform text into high-quality images. Upon signing up, you receive free credits, granting you the opportunity to fully engage in crafting captivating Christmas images, videos and more!

For premium services:

• Midjourney, a popular Discord-based platform, stands out with its active and supportive community and a user-friendly interface (particularly for Discord enthusiasts). You can also explore their free online service, offering a limited number of creations.
• If you're already acquainted with ChatGPT, consider using its dedicated Christmas card creator. Please note that a ChatGPT Plus account is required for access.
• For those seeking a specialized card creation service, Silly Robot Cards is worth trying. This service goes the extra mile by providing support for personalized cards featuring your face, family, friends, or even your beloved pet.

However, this list is by no means exhaustive, as new image generator sites emerge regularly, empowering individuals to harness the capabilities of AI artists. Feel free to conduct your own search and discover potentially intriguing tools that align with your specific requirements.

And for artists accustomed to traditional tools like Photoshop, Unreal Engine, Blender, Figma, or even Microsoft paint; you can seamlessly integrate AI into your workflow by downloading one of the available AI extensions.

How to speak to the elves

If you tried out some of the models above but it's not really producing what you want, you may need to learn how to speak to AI. Below is a short list of guides that help you understand how you can convert your ideas to text that the AI elves understand.

• The Stable Diffusion prompt book — a visual guide explaining the basics of prompt building with many examples.
• Lexica — an extensive image library that includes the used prompts and multiple parameter settings
• Arthub — is similar to Lexica with the addition of an upvoting system.
• The parrot zone — A community-created library that contains a non-exhaustive list of artists Stable Diffusion might recognise, as well as general descriptions of their artistic style. There is a ranking system to describe how well Stable Diffusion responds to the artist’s name as a part of a prompt. The site also has a list of modifiers that can be used with Stable Diffusion, similar to the artist studies. Furthermore
• The AI Art Modifiers List — A photo gallery showcasing some of the strongest modifiers you can use in your prompts and what they do. They’re sorted by modifier type.
• Top 500 Artists Represented in Stable Diffusion — By analyzing the dataset of Stable Diffusion all artists and their prominence in the dataset or listed here.
• The Stable Diffusion Subreddit — The Stable Diffusion subreddit has a constant flow of new prompts and fun discoveries. You can't go wrong if you’re looking for inspiration or insight.

Your own Santa factory

Instead of using an online service that doesn't let you download your new Christmas cards without paying or restricts you from further tinkering with the results, you can also put the little elves in your own pc at work.

There are multiple programs to do this, but one of the most popular ones is probably Automatic1111. It is entirely community-based, open-source, extendible, actively maintained and up-to-date with the newest models! It can be a bit tricky to set up if you are not used to Github but there are numerous guides and tutorials online to figure it out. This is the perfect software if you want complete creative freedom to f.e. create images of yourself (by using Dreambooth for training), create artistic QR-codes for your party (with Controlnet) or create short videos (with Deforum).

Another popular contender is InvokeAI, which you can either run as a paid service online or locally on your own pc. Similar to Automatic1111, it is open-source and well-maintained. This service is primarily loved for its simple interface and infinite canvas, which you can use for easy in- and outpainting.

Apart from this, there also exist multiple apps to create images on your phone or laptop. These applications are optimised for the hardware of the platform they're running on. For example, DiffusionBee is an app made specifically for Mac users so they don't have to struggle because of their M1 or M2 chip.

An early Christmas present

A few days ago, StabilityAI gave the world an early Christmas present by open-sourcing their newest model called SDXL Turbo. The new model is able to create images in seconds by just using 1 to 4 sampling steps! This advancement in speed and quality is thanks to a new technique called Adversarial Diffusion Distillation, which cleverly combines the speed of Generative adversarial networks (GANs), the quality of distillation techniques and the iterative improvements of diffusion models. For a detailed explanation of the architecture they use, I advise you to read the accompanying paper.

You can try out the newest model on their website. Try it out to see how it generates images faster than you can type!

For more detailed control, you can also import the newest model into Automatic1111 webui. Below are some Christmas card examples made with the newest model in a few seconds (4 sampling steps)!

Conclusion

Another year of revolutionary AI advancement has passed. A bunch of new models have come out to create better and faster images and a lot of improvement has been made to make these models accessible to all. Whatever technology you end up using, I wish you the best of fun with it. I'm going into 2024 excited to see what the new year will bring for AI and art!
Happy holidays to you!

(written with the help of ChatGPT)