How a Creation-Obsessed Man Won the Grand Prize at a Total Production Cost One-Coin Film Festival
AICU cooperated with the video generation AI competition "Total Production Cost One Coin Film Festival", which led to the discovery of new talent. In this article, Mr. Esunishi Shinjo (X@esunishi_shinjo), the first Grand Prize winner of the "One Coin Grand Prize", has contributed a making-of article.
Nice to meet you. I'm Esunishi Shinjo, a factory worker and a man obsessed with creation. I've seen a wide variety of movies, and while I can watch any movie... I still have a hard time with romance. I like both old and new movies, but I'd like to catch up on the masterpieces I haven't seen yet.
Now, at the "Total Production Cost One Coin Film Festival", which was held on X (formerly Twitter) from November 12th to December 17th, 2025, I was fortunate enough to win the Grand Prize for " Loading tweet component... A great work came in just before the deadline!
Without deviating from the cheapness of one coin, the original story is homaged and finished into a 5-minute story.
I felt the love of horror and the creative spirit š„ I can't imitate it at all š https://t.co/p7F6ov2rSY ā Esoragoto TV / AI Kawauso (@boke_ai) Loading tweet component... I'm going to write about the behind-the-scenes of creating this work, and who is Esunishi Shinjo anyway? Thank you in advance. The creative history of Esunishi Shinjo can be summarized as follows: I believe that I am the only one who can weave stories, so I drew all the plots for this film. I used AI to get feedback. First, I think about the character I want to imagine.
For example, the following characteristics are entered into grok for the main character this time. ā«ļø Characteristics of the actress "Final Girl" this time "A non-existent actress. Black hair with a light brown tint, long hair. A cute and neat woman. White blouse, black long skirt, black leather shoes. Full body. Western face without exception. Realistic. Adult woman." Then, many women with the same characteristics will appear, so I choose one actress from among them that I think is the one! (I call it "Audition")
And, this is a technique I learned when I was trying to make animation, I create an image that summarizes the four images of the front, side, and back three views and a close-up of the face based on the image at this time. I repeat this to create the characters that appear in the work. I think that ChatGPT's image generation (Image-1) is great, from the miniature of the cardboard lodge, the cheap interior of the room, the cheaply made axe, and the exterior of the forest, all created using Image-1. This work had many expressions that were likely to be rejected due to regulations. Blood expression, semi-nude expression, etc. How did I solve this? I had SeaArt generate the blood and corpses. I specified the clothes the corpses wore, how they fell, and the expressions, etc., and had them generate corpses without any sense of incongruity. Originally, it was only supposed to be one cut, so I compromised quite a bit. Most of the cuts for all the other scenes were generated with Gemini. Many of the key shots were made with Nano Banana Pro, but only the addition of the corpses was SeaArt, and the lodge and the inside of the lodge were ChatGPT (Image-1). Only one cut, I couldn't get that angle from the "Shining" parody, and I reluctantly asked Chat GPT to make it. Now, I will turn all the first cuts I have made into video footage. This can be done in parallel with the generation of image materials, or it may depend on your personality, but it is easier to create all the cuts at once, and then modify them while making the video, and then create additional cuts... And this work is also a battle with the AI service's regulations, the search for compromises, and the generation breakdown bug. When using grok to move the images, if the original image is nude or has blood, strange parts will light up, or blood splatters will move meaninglessly. Or, the nude image may be detected by moderation. As an example of generation breakdown, when I tried to get grok to show a scene of lights flashing, it over-interpreted "flashing", and the lights started to move like the Pixar opening... I don't know how many times I wanted to punch the screen. There were also times when the corpses were breathing with KlingAI (the stomach started to dent). "Finding compromises" is adjusted by reducing the number of generated cuts compared to the initial plot and improving the tempo. I am looking for compromises between whether the AI can act as I want it to, and the differences between the AI. The "shooting" while trial and error with prompts is better done while seeing through the quirks of the cameraman (video generation AI). Veo3 for texture and natural phenomena, KlingAI for good acting, grok that moves everything even if you specify where to move it, SeaArt that has loose regulations... I have all the materials ready, but the editing is actually all done manually. I hear that there is AI that can also do the editing software, but I don't care. This is the process of turning my delusions into a "movie", and I can't stand it if it's over in a few seconds or minutes. I should grunt and suffer and do it step by step. Most of the scenes are made by connecting the first cut and the end cut to create videos of several seconds or more, but there is one cut where I twisted my head a little.
In the film, there is a scene where "a killer breaks down the door and enters". Actually, the part where the door opens is grok, and I gave the image after the killer appeared from the door and gave the prompt "create a reverse playback video" and made a video of the door going back from the cut where the door is rolling on the floor. I used the editing software to "reverse playback" it and connected it to the cut "the door is broken and the killer appears and comes in from there" in one go. My expressed image moved the heart of the organizer, Raimu Light, and I was selected for the Grand Prize. The trial and error was quite difficult, so I was overwhelmed.
I realized that AI is not magic but a tool, and that it can be both poison and medicine depending on the user. I actually made a noise when it was announced.
I didn't think I would win the Grand Prize. I thought I was going to fall out of my chair, and after I said my acceptance speech and turned off the microphone, I had no choice but to stand up and soak in the afterglow. #TotalProductionCostOneCoinFilmFestivalā¦
https://t.co/NM1fFZMmjB ā Esunishi Shinjo (@esunishi_shinjo) Loading tweet component... Without the restrictions of the cheap atmosphere, and with the ability to easily resize with Nano Banana Pro. Next up is the production of 16:9 size movies. I'm going to make a cheesy muscle hero action movie that I was going to make during the One Coin Film Festival period, inspired by the atmosphere of the 1980s! I'm working on it with sincerity. Looking back, I'm glad I came up with "HORRER FOREVER". Action movies are cut hell, and it's pretty bad. But I'll do my best. As of the writing of this article, I have already exceeded the "5-minute wall" that I created with "HORRER FOREVER". How far will it go? Will it be completed! Please look forward to it! That's all. Please look forward to my future activities! I made a medal to commemorate winning the Grand Prize at #TotalProductionCostOneCoinFilmFestival. I'm planning to use this to plan an event at the Gen'ei movie theater in #gogh, so please look forward to the sequel. https://t.co/NM1fFZMmjB
pic.twitter.com/BZF1fPxdux ā Esunishi Shinjo (@esunishi_shinjo) Loading tweet component... This article is scheduled to be included in AICU Magazine Vol.20 "AI Video to Start This Year"(ver2). Please support us by subscribing to Kindle Unlimited and sharing until the beautiful paperback version is released! AICU Magazine Vol.20 The image is Ver.1
Click here for AICU Magazine back numbers j.aicu.ai/Kindle
"Esunishi Shinjo" History of Creation
Publicly Available Production Flow!
Story
Casting (grok, Gemini)
Stage Design: ChatGPT (Image-1)
Cut Generation: Gemini (nano-banana pro), SeaArt
Video Generation: KlingAI, grok, Veo3, SeaArt
Editing and Sound Effects: Filmora, Various Free Sound Sources
One-Point Technique
Generated cutSummary
Future Prospects ~Liberated from the One Coin Chains~
Poster of the movie being made to raise the tension. 
AI unique? Fake making-of style image. 
Off-shot of the cast