A few weeks ago, we received a very passionate request to make use of NovelAI’s Image Generation in Japan’s first AI Art contest.
The passionate request was heard, and we decided to attend and cover both planned events in person!
The AI contest team consists of three passionate students with a mission to bring AI Art to everyone and help them experience it firsthand — by teaching them how to prompt by doing! Visitors of their website and exhibitions are encouraged to do their own prompting!
All exhibit pieces were narrowed down through their own contest, which had over 2000 entries and many different AI entries! You can view the winners / exhibited pieces here on their Winner Announcement Post. Please take a look at all the amazing works yourself.
The photos taken at the event wouldn’t do them justice!
1st place Winner: “Japanese Kawaii” by Citrus
For the exhibits, their winners are presented on beautifully printed canvas along with a QR code that lets viewers quickly scan and run the prompt — or adjust it themselves! It’s a fantastic and fun concept for an immersive experience with like-minded AI artists and first-time AI Image Generation connoisseurs!
We decided to bring out Aini to Japan (that’s me writing!), and I didn’t have much time to get settled in before the pre-exhibit event since I arrived in Japan just two days before the Sunday of the event. The Saturday was reserved for a lengthy NHK interview and interacting with the AI Contest Team while being filmed for intro footage. I was stuck in an exciting haze of preparations, AI talk, and jet lag, but everyone was a pleasure to work with — even with some language barriers. I cannot thank everyone enough for the excellent interpretation work (and send blessings to both Google Translate & DeepL!)
Sunday’s pre-exhibit rolled around faster than I could have expected!
From my hotel in Yokohama, it only took a short train ride out to the Izumi Chuo Station location of the pre-exhibit.
Behold, the steps up to the Izumi Ward Cultural Center in Yokohama City, a 1-minute walk from Izumi Chuo Station. It was a very warm and sunny morning!
I arrived an hour before opening and quickly set up my Macbook and Anlas QR codes. The AI contest team had already set the gallery up, and at the time, the NHK crew was filming some general exhibit footage.
Since this was the Pre-Exhibit, only the winners and a fraction of the full exhibit were displayed.
It’s quite intimidating when a gigantic camera is just chilling in the corner of your vision!
We got word that some people lined up outside right before opening, and it was really neat to see how excited everyone on the team got!
Once the doors opened the first few attendees filed into the venue to be greeted by a wonderful array of colorful AI art pieces. They quickly began scanning the exhibit image QR codes to generate their own versions of the exhibited pieces.
It was really fun to watch people experience their own AI image journey picture by picture — at the end, they had the option to purchase a keychain of their personal favorite generation! Most seemed to be veterans, but throughout the day, we had some complete AI Image generation newbies stop in as well!
There were some technical difficulties getting those cute little keychains printed and cut in a timely manner, but that’s something we’ll have straightened out by the main event!
All generated works were shown via projector throughout the day as well! It was amazing to see everyone’s different types of prompting and their unique style of generations!
I was up late into the night fighting with my QR code printer trying to create as many NovelAI keys as I could — it took the devs a few days to implement the option to provide Subscription + Anlas keys — only to realize at the event that already-subscribed users have no way of inputting the QR code or the gift key… Good thing we have this pre-exhibit event for these kinds of realizations! But not to worry! Flexibility is my second name!
I quickly whipped up an alternative to provide Anlas by getting the Account IDs of the already subscribed attendees and voila, everyone in attendance got their promised amount of Anlas!
Now I just need to make sure by the time the main exhibit rolls around that the developers have implemented the Anlas Gift Key field option…
We also documented the changes/instructions on the go with handwritten signs. It was funny seeing those posted on the Japanese Discord (as they evolved).
It was really amazing to meet NovelAI users in person for the first time, actually. It felt a lot like Twitch streaming!
Everyone gathered and brought their questions to the table in a very kind, open, and understanding manner. It was a pleasure to be face-to-face with the people that make use of what we’ve created over the past year and a half!
We had a few visitors show off their works — one of them even had his own AI art printed on his shirt! It was neat to be able to snap a picture and shoot it to the dev team on the spot, and everyone was just blown away!
I honestly teared up a little when one of our visitors came up to our table and read out his machine-translated note detailing how image generation has enriched his life. I managed to get my hands on the actual note and will definitely treasure it forever!
We actually received a lot of in-person questions. In the beginning, many people asked what our general reaction to Nijijourney coming out is, which is exciting. We’re hyped to see more new AI art styles and variety popping up, and hopefully, there are aspects that our team can learn from as well! Also, we’ve seen some of their Beta testers combine NovelAI with Nijijourney, creating amazing results!
Most other questions were more technical, surprisingly, we’ve mainly gotten requests for expanded prompt character allowance and general discussion on how to make the best use of Undesired Content.
Despite appearing a little on the upset side, one visitor was very open to hearing my explanations of our hardware, computing, and infrastructure costs after asking how we could charge for our service. Providing an ongoing service, customer support, continued development, and, most of all, the actual training of models is also a big monetary investment on our side and not just a huge time investment from the whole team. I like to refer to our previous blog post that mentioned some of the non-consumer hardware we had running for three months straight to train our NovelAI Diffusion models. It was great to be able to have civil discussions in a face-to-face and very down-to-earth manner.
We have a good amount of additional Q&A content ready and lined up for the main exhibit at the end of the month!
During the event, we had a good number of visitors, and many even stayed around for most of the afternoon. I really enjoyed watching people chat, learn prompts, and generally exchange experiences and mingle. The atmosphere was super cozy.
All in all, this day was a magnificent learning experience and a generally wholesome time meeting everyone. I was honored to have had the opportunity to partake in the event.
Many thanks to the AI Contest team for allowing NovelAI to attend in this capacity!
We’re currently preparing something exciting with the NovelAI Winter Contest in collaboration with the AI Contest Team for the main exhibit taking place December 28th and from January 4th through 8th!
So please stay tuned to https://novelai.net/contest for the NovelAI Winter Contest in collaboration with the @AI_contest team!
