Updated: Mar 27, 2026
base model
I cannot guarantee this model is good for what you want so try it out. If you want inside the site generator, feel free to bid on it by click the gavel/hammer icon next to the notification button on what version of this model you want to bid on. Also, all images outside of thumbnails with 7 CFG Euler Simple is just me sharing the prompts of some raffle images I got, I tend to run Euler Normal, 20 Steps, 4 CFG on that stuff.
I'm working on a small AI Discord Community you can join here: https://discord.gg/ehhMrD4PQT
Valkyr Core
Valkyr Core has a more basic look to it depending on the prompts, V1.0 and beyond will be refined for a more anime base style.
The Base Version may be rough depending on what you prompt or do with it, so use at your own transgression.
For best results use local generation, I've noticed. Unless I was prompting it wrong which could be the case.
Settings I personally use:
Steps: 20 or Higher
CFG: 3 or Higher until Overfitted/Overbaked
Sampler: Euler / Euler A (Ancestral)
Scheduler: Normal
Positive Prompt that I personally use:
masterpiece, best quality, highest quality, absurdres, highres, extremely detailedNegative Prompt that I personally use:
worst quality, low quality, normal quality, lowres, blurry, out of focus, jpeg artifacts, compression artifacts, ugly, bad anatomy, bad proportions, bad hands, deformed, mutated, loli, shota, aged down, childADetailers are also recommended to be used with this however, I generate with base images to showcase off what the checkpoints does with the base generation for the thumbnails.
Feel free to test, experiment, generate with it, or even merge with it—I don't mind! Just make sure to give credit to the model existing, if you merge with it so I can eventually find your merge, and just maybe I'll give those merges a like if I like the results!~ c;
For those that it may concern:
Main Thumbnail Notes: Generated through ComfyUI then dumped into Gemini for the title, then removed the watermark, then ecaj used seedance or whatever for me as a placeholder (probably permanent) which I'm grateful. The title has a planet in the thumbnail as a nod to it's previous (and placeholder) name "Atlas Core"
Note: Unreleased is just a LoRA trained on that version of the checkpoint and will be never released as it is there to refine the checkpoint towards what I personally want. All versions were made on ComfyUI. Also "See Metadata for information." means drag and drop the model into ComfyUI to see what the flow was that made each version.
What was merged in - V1.0 Merge Info:
1) Atlas Core - Checkpoint, Version: 0.9
2) Arctenox's Refiner [Illustrious] - LoRA, Strength: 56%
3) Atlas Core - Checkpoint, Version: 0.01 - Merge Method: Simple Merge. Strength 33%
Saved as step 1 then continued the below with step 1:
4) Hex Furry and More [Illustrious & NoobAI] - Checkpoint, Version: 0.9 - Merge Method: Competitive Model Selector, Tensor Prism Node Pack
Inbred with step 1 at % strength
5) Symphtia Quality/Style [Illustrious] - LoRA, Strength: 15%
Saved as step 2 then continued the below with step 2:
6) Amanatsu (Illustrious) - Checkpoint, Version: V2.0 - Merge Method: Analyze Model Weights / Apply Merge Recipe, Tensor Prism Node Pack
Saved as step 3 then I inbred it with Atlas Core - Checkpoint, Version: 0.9 then continued with the below with step 4:
7) Arctenox's Styles - Detailed Anime | Series 1 [Illustrious] - LoRA, Strength: %
8) Ionsyx - Checkpoint, Version: V5.2 - Merge Method: Block Merge
9) One obsession - Checkpoint, Version: V20_bold - Merge Method: Competitive Model Selector, Tensor Prism Node Pack
10) Atlas Core - Checkpoint, Version: 0.85 - Merge Method: Simple Merge, Strength %
11) Unreleased - LoRA, Strength %
See Metadata for information. Note: In this version I have renamed the model from "Atlas Core" to "Valkyr Core" thank you RedactedPaws/WarWolf for the name change.
Exact order of what was merged in - V0.9 Merge Info:
1) Atlas Core - Checkpoint, Version: 0.85
2) Unreleased - LoRA, Strength: 28%
3) Arctenox's Styles - Verynsty | Series 1 [SDXL] - LoRA, Strength 38%
4) Arctenox's Styles - My Style | Series 1 [Illustrious] - LoRA, Strength 26%
5) dlckrpwjd111 Style - LoRA, Strength 41%
6) nao takemo art style - LoRA, Strength 73%
Then inbred with Atlas Core - Checkpoint, Version: 0.85
See Metadata for information.
Exact order of what was merged in - V0.85 Merge Info:
1) Atlas Core - Checkpoint, Version: 0.8
2) Hex Core - Checkpoint, Version: 2.0 - Merge Method: Block Merge
3) Atlas Core - Checkpoint, Version: 0.6 - Merge Method: Simple Merge
See Metadata for information.
Exact order of what was merged in - V0.8 Merge Info:
1) Atlas Core - Checkpoint, Version: 0.7
2) Unreleased - LoRA, Strength: 73%
3) Epic Seven - Official Art Style | Series 2 [Illustrious] - LoRA, Strength: 72%
4) Cinematography_Lighting_Base - Lora, Strength 32%
5) Vete ┃ Veltrix Style - LoRA, Strength 2%
6) Atlas Core - Checkpoint, Version: 0.7 - Merge Method: Simple Merge, Strength: 50%
7) Ecliptica-ArcaneX - Checkpoint, Version 2.0 - Merge Method: First used "Epsilon/V-Pred Converter Tensor Prism Node Pack" then I used "Competitive Model Selector, Tensor Prism Node Pack" with the above stuff.
See Metadata for information for what I did with the Ecliptica-ArcaneX Checkpoint.
Exact order of what was merged in - V0.7 Merge Info:
1) Atlas Core - Checkpoint, Version: 0.6
2) Unreleased LoRA
3) Moonvine - [Illustrious] - Checkpoint, Version: Crystal IV - Merge Method: Block Merge - Twice
4) Atlas Core - Checkpoint, Version: 0.6 - Merge Method: Simple Merge, Strength: 28%
5) Atlas Core - Checkpoint, Version: 0.4 - Merge Method: Simple Merge, Strength: 22%
Exact order of what was merged in - V0.6 Merge Info:
1) Atlas Core - Checkpoint, Version: 0.5
2) Unreleased - LoRA, Strength: 100%
3) Hidden Gem Style - LoRA, Strength: 94%
4) Yotte Style - LoRA, Strength: 22%
5) Hasanima - Checkpoint, Version: 1.0 - Merge Method: SDXL Advanced Block Merge, Tensor Prism Node Pack
6) 二次元特效强化（Anime special effect enhancement）- LoRA, Strength: 8%
7) Atlas Core - Checkpoint, Version: 0.5 - Merge Method: Simple Merge, Strength: 16%
Used the save version of the stuff above with the below listed models and then merged it back in at 16%
8) Banana Splitz XL - Checkpoint, Version: Vee 9 Point Oh! - Merge Method: Competitive Model Selector. Tensor Prism Node Pack
9) Atlas Core - Checkpoint, Version: Base - Merge Method: Simple Merge, Strength: 9%
See Metadata for information.
Exact order of what was merged in - V0.5 Merge Info:
1) Atlas Core - Checkpoint, Version: 0.4
2) Hyphoria - Checkpoint, Version: 0.1 - Merge Method: Analyze Model Weights / Apply Merge Recipe, Tensor Prism Node Pack
See Metadata for information.
Exact order of what was merged in - V0.4 Merge Info:
1) Atlas Core - Checkpoint, Version: 0.3
2) Tenkart style - Illustrious - LoRA, Strength 42%
3) anon(@betanonbeet) style - LoRA, Strength 34%
4) aes_Stylemix - LoRA, Strength 24%
5) Atlas - Checkpoint, Version: 0.3
6) Unreleased - LoRA, Strength 63%
Used in a simple merge with itself with the unreleased LoRA at a 30% then saved and uploaded.
See Metadata for information.
Exact order of what was merged in - V0.3 Merge Info:
1) Atlas Core - Checkpoint, Version: 0.2
2) Unreleased - LoRA, Strength 100%
3) VariVi’s Abstract Nightmare Style - LoRA, Strength 24%
4) Horux - LoRA, Strength 3%
5) Hex Ionsyx [Illustrious] - Checkpoint, Version 2.0 - Merge Method: Merge Method: Competitive Model Selector, Tensor Prism Node Pack
6) ✨ JANKU Trained + NoobAI + RouWei Illustrious XL ✨ - Checkpoint, Version: 6.9 - Merge Method: Analyze Model Weights / Apply Merge Recipe, Tensor Prism Node Pack
7) Atlas Core - Checkpoint, Version: V0.01 - Merge Method: Simple Merge
See Metadata for information.
Exact order of what was merged in - V0.2 Merge Info:
1) Atlas Core - Checkpoint, Version: 0.01
2) Arctenox's Styles - Style Mix 4 | Series 2 [Illustrious] - LoRA, Strength: 87%
3) Arctenox's Styles - Style Mix 5 | Series 2 [Illustrious] - LoRA, Strength: 57%
4) Unreleased - LoRA, Strength: 40%
5) RAMTHRUST'S-NSFW-PINK-ALCHEMY-MIX✨ - Checkpoint, Version: Alchemy_Mix_1.76 - Merge Method: Block Merge
6) Atlas Core - Checkpoint, Version: 0.1 - Merge Method: Competitive Model Selector, Tensor Prism Node Pack
See Metadata for information.
Exact order of what was merged in - V0.1 Merge Info:
1) Atlas Core - Checkpoint, Version: 0.01
2) Arctenox's Refiner [Illustrious] - LoRA, Strength: 53%
3) Stabilizer IL/NAI/CK - LoRA, Version: illus01 V1.198, Strength: 43%
4) Symphtia Quality/Style [Illustrious] - LoRA, Strength: 42%
5) Fine Anime Screencap XL | Anime Screencap Style LoRa Illustrious and PonyXL - LoRA, Version: V3.0, Strength 37%
6) Arctenox's Styles - My Style | Series 1 [Illustrious] - LoRA, Strength: 33%
7) Arctenox's Styles - Noxfall | Series 1 [Illustrious] - LoRA, Version: V3.0, Strength: 22%
8) Esmaysia - [ILLU V2] - Checkpoint, Version: V0.5 - Merge Method: Block Merge
9) Arctenox's Styles - Detailed Anime | Series 1 [Illustrious] - LoRA, Strength 97%
10) Epic Seven - Official Art Style | Series 2 [Illustrious] - LoRA, Strength: 37%
11) Arctenox's Styles - Celfade | Series 1 [Illustrious] - LoRA, Strength: 21%
12) Lyn's Sketch Style - lindeza - LoRA, Strength: 17%
13) Contrast Controller [IL/NAI] - LoRA, Strength: 17%
14) Zackary911 (Artist Style) [PonyXL/Illustrious] - LoRA, Strength: 14%
15) (IL Slider | Tweaker) Color Temperature, Saturation, Brightness | 色温 亮度 饱和度 (IL/Noob调节滑块) - LoRA, Version: saturation, Strength: 16%
Saved as AtlasCoreV0_1_00001 and used in a simple merge with Atlas Core Base at 32% then saved again as Atlas_CoreV0_01_00001 and uploaded
See Metadata for information.
Exact order of what was merged in - V0.01 info:
1) Chenkin XL V0.2 - (NoobAI Epsilon 1.1 Finetune) - ComfyUI SDXL Block Merge with Illunorou V1.0
2) Hex Core V1.0 - Used Full CLIP (Base Knowledge) Only
3) Illunorou V1.0 - (Rouwei V0.8.0.0 / Illustrious 2.0 Stable / NoobAI Epsilon 1.1 Merge Model) - ComfyUI SDXL Block Merge with Chenkin XL V0.2