top of page
  • X
  • Writer's pictureMagik Chance

The Welcoming AI Art Cafe: NightCafe Studio (Part II)

If you missed the first part of this article from last week, check it out here. There I covered getting started on NightCafe Studio and earning credits. In this article, we will take a look at creating AI art in NightCafe and participating in their chats and challenges.

These articles are part of a Creator Collaboration I am doing with @NightCafeStudio. As part of this collaboration, I will be in an Ask Me Anything space later today, and tomorrow, I will be creating a chat and challenge on NightCafe. I hope you join me for both of these events!

Creating Your Art AI

NightCafe has a Stable Diffusion base. If you’ve ever used Leonardo AI or Stable Diffusion itself, you should be at home here.

There are two big differences between a Stable Diffusion based platform like NightCafe and platforms like Midjourney, Dall-E, or Ideogram: 1) you can select different models and LoRAs and 2) you can use a negative prompt (or more than one positive prompt) instead of just a single prompt.

Models & LoRAs

NightCafe has a range of models, including big favorites such as Dreamshaper, Juggernaut, Crystal Clear, Real Cartoon, and Starlight, along with many smaller models.

You may also train your own LoRAs on NightCafe. You get two free fine-tunes with additional ones costing credits or offered as part of a monthly subscription. Pro members additionally can use community LoRAs.

Your model / LoRA will be your first decision in the creative process. Your model can make a large difference in your outcome so I highly suggest experimenting!

(You might have noticed Dall-E 3 and Creative Upscaler under models. These are Pro only options and work differently than regular models.)

AI Art. Pinkish purple tulips with fresh morning dew.
Some models offer great photo-realism while other specialize in stylized images. Juggernaut XL v8 was used to create this image.

Text Prompt

You have your standard prompt, but if you open the “Advanced Prompt Editor,” you will find a toggle for a negative prompt (there is a default, but you can tweak it) and the option to add another prompt.

You will also find a section for “Presets,” “Modifiers,” and “My Favorites.”

Presets are base prompts for various styles while modifiers are tokens you can add to your prompt. I highly recommend going through these as they can offer some great ideas.

My Favorites is where you can find your own saved prompts. To save a prompt, look for the “Save” button on the upper right corner of the “Your text prompt” box within the Advanced Prompt Editor.

If you’re new to Stable Diffusion prompting, or you just want more ideas, I suggest heading to the “Explore” tab and checking out (#)open-prompt-gang. People who use this hashtag leave their prompts and settings open for others to see and use. To see these settings, including the prompt and model, select the “Settings” tab located next to “Comments.”

Starting Image

Next is your (optional!) starting image. I often experiment with mixing models by using a starting image from one within another.

If you choose to use a starting image, you will find ControlNet options. These are very powerful, but do note that they take slightly more credits to use.

You will also see an “Inpainting” option. This is still experimental, but if you choose to use it, hit the button to create a mask. You will notice there’s no place to put a prompt here – you will want to edit the regular prompt to include the change you want made.

Lastly there’s a bar for “Noise.” You can almost think of noise as image weight – the more noise, the more different the final image will be.


Next are the settings. Here you may select how many images you want to generate and how long you want the runtime to be. Keep in mind that both these options will up your credit cost. Personally, I stick to 1 image at a time and to “short.” I have not found much of a difference with a longer run time personally, but it’s something to experiment with.

Aspect ratio will not affect credit cost but keep in mind that if you use a starting image, you will be locked into that image’s ratio. Overall Prompt Weight also does not affect credits, so experiment away.

You will want to pay attention to the Refiner Weight. Some models work well with a refiner and some do not. Luckily NightCafe lets you know if your model does better with low or no refiner so you don’t have to memorize it!

The last options are Seed (toggle off to input a specific seed) and sampling method (which I know nothing about and personally keep on default).

AI art. A woman with messy, short blue hair. Strong bokeh effect in the background.
One of NightCafe's newest models is Dreamshaper XL Lightning, which was used to create this image.


Once you’ve made some images, you can choose to “publish” your favorites. Publishing means the image appears under your profile and in the explore section of the website where people can like or comment on it.

You can add a title, description, tags, and "buy link." The buy link is for if you want to sell the image as a print or an NFT -- you can link the product page here.

If you’d like to keep your gallery organized, you can additionally make collections and add your images to them. Collections are also a great way to find your images later if you want to upscale or evolve them.


If you’d like to share AI art from elsewhere with the NightCafe community, you may also upload images. To do this, hit the “Create” button then on the left hand menu, click to the “Upload” tab.

Uploaded images may be used as starting images or published as is. You cannot use uploaded images, or images created with them as starting images, in Daily Challenges; however, some community challenges do allow uploads.

Daily Challenges

There are two types of challenges on NightCafe: Daily and Community.

Daily Challenges are run by the NightCafe team and can be found under the “Challenges” tab. Each day you may submit an image for that day’s challenge as well as vote in the previous day’s challenge.

Placing within the top 10% or above in a Daily Challenge will earn you free credits. You can also vote on at least 40 images for the day to earn 2 credits.

Community Chats & Challenges

When in the Create menu, you will see a speech bubble icon on the side of your screen. This is the chat menu. Click the “Start or Join” tab to find chats to join or create your own room.

Chats allow you to actively talk with community members as well as share your art. They are a great place to ask questions and meet your fellow artists.

Most chats will have a community challenge attached to them. Challenges will show their status, whether uploads are allowed, their theme, and how many players they currently have. If you join a challenge, you can see this information under the dice icon that appears in the top right corner of your screen. The dice icon will display how many active challenges you are a part of as well.

If you do join a challenge, be sure to vote when it is time! It’s common courtesy to vote in challenges you are participating in as it ensures a fair competition.

AI Art. An adorable baby owl.
Starlight XL, which was used to create this image, is great for concept art and illustrations.


I enjoy NightCafe because it offers much more than other AI art platforms, especially in terms of community. I greatly enjoy participating in challenges and encourage you to give them a try!

Tomorrow I will be opening my own community chat and challenge on the platform. I hope you’ll join me there and submit your art! (Uploads will be allowed.) Please keep an eye out for the link on X.

14 views0 comments


bottom of page