Burupen Releases New Models "true_pencil-XL" and "mellow_pencil-XL"! Available on Shared ComfyUI!
Today, we have a topic about AI generation for illustrations after a long time.
Mr. Burupen X@blue.pen5805, who releases original AI models for illustrations using Stable Diffusion, has released two new SDXL models, "true_pencil-XL" and "mellow_pencil-XL." They are now available on AICU Lab+!
true_pencil-XL
I created a mysterious model called true_pencil-XL! I wanted to create something a little different from usual, so I merged it from scratch! Most of the recent ones are this (or ones I was trying to create in the process).
I'll probably upload it to Civitai later https://t.co/H4oqIhtftQ pic.twitter.com/CHgCXjXI1W
— ぶるぺん/blue.pen5805 (@blue_pen5805)
Loading tweet component...
?ref_src=twsrc%5Etfw">July 1, 2025According to the description on Civitai: Unstable, but interesting
Sketch style model based on Illustrious XL 2.0 Unstable but interesting
https://civitai.com/models/1737210
mellow_pencil-XL
As the name mellow suggests, it seems to be a model suitable for broken illustrations(?).
Second in the weird model series, mellow_pencil-XL! It's a model that feels like it has (realistic:-6) in the negative prompt by default!
Of course, it gets extremely rough! If you are confident in your image generation skills, you should use the chaos version, it's probably more interesting! https://t.co/7ViNiaLi80 pic.twitter.com/m5a4jRpTTS
— ぶるぺん/blue.pen5805 (@blue_pen5805)
Loading tweet component...
?ref_src=twsrc%5Etfw">November 5, 2025
What is mellow? The base is just a merged one, and the one adjusted with Concept Sliders is the one without base. So please download the one without base. The one with chaos is more lively but difficult to handle. But even the one without it is usually difficult to handle. The one with chaos is more interesting, but the world will collapse unless you lengthen the prompt or raise the CFG value (even to levels like 10 or 15).
Experimented with AiCuty's official prompt
Elena Bloom (Elena Bloom) is originally Checkpoint: WAI-NSFW-illustrious-SDXL LoRA: Niji anime illustrious, EnchantingEyesillustrious , Gradient Hair Steps: 28 Sampler: DPM++ 2M SDE Karras CFG Scale: 5 Clip Skip: 2 Seed: 798458095628920, and the official prompt is based on the one published here.
You can use it freely if you check this AiCuty copyright usage guideline.
AiCuty Copyright Usage Guidelines | AICU Japan __ www.aicu.jp
# Composition and Quality masterpiece, best quality, solo, full body, sweet and gentle anime idol girl in center position, from front, soft and affectionate personality, slightly shy warm smile, # Hairstyle and Decoration fluffy twin tails tied high with big pastel pink ribbons and rose flower hair clips, soft curled ends with subtle rose pink gradient, choppy bangs, loose side bangs softly framing forehead, # Facial Features and Expression silky rose-pink hair, big bright sparkling pink eyes with soft highlights, large expressive eyes, looking directly at camera, tidy silhouette, # Outfit (Gorgeous and sacred idol stage outfit) wearing a goddess-inspired idol stage outfit mainly in pastel pink and white with fine gold accents and subtle black trims: off-shoulder flowing mini dress made of lightweight futuristic tech-fabric with soft iridescent glow and subtle sheer overlays along skirt edges, layered skirt with outer bright pink tech-fabric and inner layer of airy semi-transparent white chiffon, slim gold belt with tiny pink gemstone and subtle black edge lines, cropped white bolero jacket with semi-sheer mesh sleeves and hem with faint see-through effect and pink piping, subtle glowing circuit pattern on cuffs outlined with fine black piping, fingerless lace gloves with tiny pink bows, white ankle boots with subtle gold trims and small black edging at soles with small pink ribbon flares at back, # Pose standing in a cute idol pose with both hands forming a heart near chest, looking directly at camera, # Background white background, simple background, clean studio light, # Lighting (Unified Instructions) soft directional white light from upper left, consistent shadowing across all elements
NegativePrompt: low quality, worst quality, bad anatomy, poorly drawn face, deformed eyes, asymmetrical eyes, multiple faces, extra limbs, cropped, blurry, out of frame, watermark, text, nsfw, loli, mature woman, heavy makeup, inconsistent outfit, duplicate costume
It's certainly interesting with pencil-drawn lines, but as instructed in the model, I'll try to interact with the model by removing the prompt. I'll remove the composition and quality, background, and negative prompt, and only leave "low quality, worst quality, bad anatomy."
Hairstyle and Decoration
fluffy twin tails tied high with big pastel pink ribbons and rose flower hair clips, soft curled ends with subtle rose pink gradient, choppy bangs, loose side bangs softly framing forehead,
Facial Features and Expression
silky rose-pink hair, big bright sparkling pink eyes with soft highlights, large expressive eyes, looking directly at camera, tidy silhouette,
Outfit (Gorgeous and sacred idol stage outfit)
wearing a goddess-inspired idol stage outfit mainly in pastel pink and white with fine gold accents and subtle black trims: off-shoulder flowing mini dress made of lightweight futuristic tech-fabric with soft iridescent glow and subtle sheer overlays along skirt edges, layered skirt with outer bright pink tech-fabric and inner layer of airy semi-transparent white chiffon, slim gold belt with tiny pink gemstone and subtle black edge lines, cropped white bolero jacket with semi-sheer mesh sleeves and hem with faint see-through effect and pink piping, subtle glowing circuit pattern on cuffs outlined with fine black piping, fingerless lace gloves with tiny pink bows, white ankle boots with subtle gold trims and small black edging at soles with small pink ribbon flares at back,
Pose
standing in a cute idol pose with both hands forming a heart near chest, looking directly at camera,
Recommended settings were listed on the Civitai version. https://civitai.com/models/1737210
Short prompts are recommended. (Long prompts overwrite the style.) Sampling method: Eular a (Euler ancestral) CFG: 5
Since it's an SDXL model, I'll try using 1024x1024 as the standard image size. The prompt might be shortened even more.
It's a model with strong "habits" such as bangs and eyebrows!
Let's experiment with "mellow_pencil-XL" using the same prompt.
Faint roses are raging in the background!
It's quite attractive as it is, but adding 1giri, chibi makes it even more attractive!
Included
"mellow_pencil-v1.0.0-chaos"
In addition to base and unmarked, mellow_pencil has a model called "chaos."
The one with chaos is more lively but difficult to handle. But even the one without it is usually difficult to handle. The one with chaos is more interesting, but the world will collapse unless you lengthen the prompt or raise the CFG value (even to levels like 10 or 15).
https://huggingface.co/bluepen5805/mellow_pencil-XL
It's inconsistent, but it's a model that you don't get tired of looking at!
I'll experiment with other characters as well
Mina Azure
Nao Verde
With true_pencil, it feminizes the character, but mellow_pencil and mellow_pencil.chaos are wonderful. In particular, the default expressiveness of chaos is wonderful.
Saki Noire
mellow_pencil
true_pencil
mellow.chaos
Mei Soleil
true_pencil
mellow_pencil
mellow_chaos
Available for use with Shared ComfyUI!
The official download is here
bluepen5805/true_pencil-XL · Hugging Face We’re on a journey to advance and democratize artificial inte huggingface.co
bluepen5805/mellow_pencil-XL · Hugging Face We’re on a journey to advance and democratize artificial inte huggingface.co
Thank you, Mr. Burupen, for the wonderful model!
true_pencil-XL/mellow_pencil-XL will be available on [Shared ComfyUI-Unit 2]!!
Since the licenses are both Fair AI Public License 1.0-SD****, we carefully considered commercial use and concluded that it is possible to provide Fair AI Public License 1.0-SD models on Shared ComfyUI.
- Display the full text of the license or the source URL in a way that is visible to the user.
- If there are modified or learned derivative models, make the source or model available.
- Do not use for prohibited purposes (discrimination, surveillance, medical judgment, etc.).
Based on the above, it is now available on AICU.jp Shared ComfyUI (Unit 2).
Shared ComfyUI Model License | AICU Japan __ www.aicu.jp
Shared ComfyUI is an ancillary service of AICU Lab+, which builds an environment where it is easy to study by sharing GPUs for servers.
It is now available with a new interface.
Shared ComfyUI is 3,500 yen per month. It is available with the AICU.jp subscription "AICU Lab+". We provide a community that explains how to use various AI generation models to generate images/videos using ComfyUI with textbooks, and the AI generation service [Shared ComfyUI]. Online study sessions carefully explain everything from basic knowledge of image/video generation to those who are using image/video generation AI for the first time, those who are migrating from A1111 to ComfyUI environment, setting up the environment, and those who are considering equipment selection. Let's become "creators" in the full-scale image/video generation era through community, contests, events, certification programs, and video archives for those who aim higher!
AICU Lab+ Plan Policy | AICU Japan __ www.aicu.jp
Please refrain from destructive or occupancy behaviors that may bother other users (Terms of Use).
