Gathered Scenes Zine is a set of image-generation Skills written for Codex, created by Zeejay0. Its positioning is not to fit photos into a fixed template, but to first read the scene: identify the subject, space, color, action, and the emotions left unsaid, then decide whether to preserve the real scene or distill the scene into a new work on paper. According to the materials provided by the repository: photos provide facts, and creation decides how to keep them.
[Purpose and Two Creative Paths] The same photo can be handled in two ways, corresponding to the two complementary Skills included in the repository.
scenes-gathered-zine-v1-3: Preserves the irreplaceable scene relationships in the photo, using real photography as an anchor, allowing abstract shapes derived from the original image, a single high-purity color, and visible hand-torn fiber edges to extend onto the paper. Suitable for situations where you want the finished work to still retain the factual basis of the original photo.
scene-distillation-zine-v1-3: Does not retain the original photo in the finished work; instead, it extracts the semantic core, emotional tension, and visual metaphor from the photo, then uses paper, illustration, color, and free text to create a new work. Suitable for situations where you want a completely reconstructed new image.
[Visual Principles] The materials state that this visual language is built on five principles: real scenes as anchor, where the real scene is not decorative material but the factual basis of the work; illustration creates the scene, where illustration expands the space the photo has not yet finished speaking, rather than copying the photo; color becomes structure, where high-purity colors handle balance, direction, and visual weight; the paper surface breathes, where negative space participates in the narrative, and complex information is compressed into a small number of clear forms; boundaries have texture, where torn paper, fibers, dry ink, and printing errors form the language of materials. The materials also emphasize that each creation goes through the same observation process, but does not apply a fixed template, and prioritizes preserving the relationships that truly matter in the scene, such as people and distance, window and light, road and direction, plants and blank space, and the lingering emotional aftertaste left by a moment.
[How to Use] The steps given by the materials are: clone the repository, then copy the needed Skill into the Codex Skills directory; if the Skill does not appear immediately, restart Codex; choose "real-scene collage" or "image distillation"; invoke the corresponding Skill, and you may also add the relationships you want to preserve, the text language, or the emotional direction. In addition to images, both Skills also return a brief creation note; the specific output rules are subject to their respective SKILL.md.
[Work Archive] The materials state that the repository saves representative cases as "original photo -> observation record -> final work," rather than only displayingResult image wall; each file explains what the work retains, what it discards, and how color and composition grow out of the scene. Examples include: Real-scene Collage 01 · Tbilisi Distant View, which retains the church tower as a real anchor and compresses urban density into a blue printed field; another example retains the crowd on the bridge and the water reflections, translating the snowy forest into a scene on blue-gray paper; Image Distillation 01 · Time Waves Back, which discards the photograph itself and retains only the waving gesture, the distant stone statue, and the response that has not yet arrived between the two, using a yellow gesture trail as a visual metaphor for time; Image Distillation 02 · Snow Falls Lightly, which distills a seated pose in the snow and the colors of winter clothing into loose paper-scrap figures, preserves the lightness of snow with large areas of paper white, and establishes a visual response with a single warm red small block.
[Privacy and Sharing] Data statement: The Skill uses only the photos provided by the user as reference for the current generation task; unless the user explicitly requests it, it should not browse, share, separately upload, or save the original photos, and the specific handling rules are subject to the corresponding Skill documentation. After the first two generations in the same conversation are completed, each Skill will give a lightweight reminder: if publicly sharing, you are welcome to credit Visual Skill by @Zeejay0; from the third time onward it will not be repeated.
[License and Restrictions] The data makes clear: personal, non-commercial use only. Selling, paid generation, subscription services, done-for-you work, consulting, training, SaaS/API, company or client projects, and other commercial uses are not allowed. Any commercial use requires prior explicit written permission from Zeejay0; see the personal non-commercial license for details.
[Other Information] The data mentions that the author's website is now online https://zeejayzine.com/ , and states that it connects to the top-tier GPT5.6+image2 model and is usable in the domestic environment; it also mentions the zine AI recharge card website https://zeejayai.store/ which provides GPT and Claude top-up and ready-made account services, involving Plus top-up, Pro 5x, 20x, etc. The unified username on Douyin and other content platforms is Zeejay0. Regarding platform support scope, installation directory details, specific output rules, and the full license text, the data points to the respective SKILL.md and license files, which are not expanded item by item here.
