Generate pictures with Arthemy
✦ Arthemy Comics XL - Quick Start ✦
1_____Size: Around 1280 x 1024
2_____Prompt: Start from the prompt present here and tune the results based on your needs:
Quick Prompt:
A 2D European comic book illustration with clearly recognizable line art of a [Character] [facial expression/pose description] in a close-up shot. The character is a [age and physical description], and they are portrayed in front of a [background object/scene] that is glowing with [color/brightness description]. They have [unique physical feature], giving them a distinctive look. The scene is dramatic, with complex lighting that highlights the main character, making them stand out in their [expression/pose] amid the chaotic scene. The background gradually shifts from a focused and sharp illustration around the subject to more undefined shapes, eventually blurring into a simple background that surrounds the entire artwork, with a sense of atmospheric fog depth. The overall drawing uses [color tones] with flat colors, maintaining high coherence and continuity between shapes.
I hope you like my merge and I can't wait to see what you all are going to create with it!
For future update, about the AI Editor I'm building with an italian team and my models you can check:
https://www.instagram.com/arthemy_ai/
____________________________________
Merged / subtracted / inspired by Models:
SVision - https://civarchive.com/models/428996/svision
Cheyenne - https://civarchive.com/models/198051/cheyenne
GlobbyXL - https://civarchive.com/models/454008/globbyxl
+ some of my own trained models
If you like Arthemy Comics, remember to check these other models and consider to support their creators!
Description
After a few times, going back and forth between the different versions of my models, I grew dissatisfied with this merge and I decided to bring much more consistency and precision to the model - Cheyenne is an absolutely fantastic model and it's still the main base for this model which I've brewed with some of my own trained LoRAs and several other models - you can find them in the description.
FAQ
Comments (23)
i think it doesn't work very well when applying a controlnet for depth or similar with a reference image
Oh no! I'll try to understand what's giving this issue - do you mind sharing more informations about the UI you're using and the ControlNET version? :D
Works very well with any kind of controlnets for me. Depth, Mistoline, Openpose etc
It's a mixed bag for me. It sometimes works well with depth and openpose, not so much with normalmap. It also seems slow with controlnet in relation to other models. I use auto1111. ArthemyXL is a great model, by the way. IMHO it's the best for comic art in SDXL.
@lavender69studios I mainly generate with comfyUI so I'm not really aware of possible incompatibility in A1111. I'll have to test it further. Thank you for the feedback and the kind words. ❤️
@Arthemy Thanks to you for your excellent work!
@lavender69studios You're absolutely welcome! :D
What has changed in the v2.0?
I tried generating the same seed with both checkpoints and saw no visual difference...
Since I didn't want to focus this model on a specific aeshtetic, it's mainly a re-made merge of the same models with just a few hints of small loras in order to gain a higher consistency.
If you got the same images... my opinion is that you tried to make a subject that was "outside" the slices that were modified during this process.
@Arthemy thanks for the explanation
The onnx model wont work on Amuse, sadge
I'll have to investigate that!
Unfortunately I have a bad relationship with "Pony". To me it's harder to write prompt and it feels to me a little bit uncontrollable most of the time.
If you can suggest me a Pony model good enough to replicate complex prompts I'll be glad to work on that format too!
@Arthemy Thanks for the answer) I understand, but in this case I can recommend something like Homemade Pony. No matter how much it generates in general, it responds quite well to sometimes complex hints and prompts, but she really demands VAE:
https://civitai.com/models/600594/homemade-pony
P.s. If I understood correctly what you meant))
@missfidonyo I'm going to check that out as soon as I'll have a free afternoon to experiment with SD on my own! Thank you very much for the tip!
@missfidonyo if you're still interested, I've created a Pony version too!
@Arthemy ty! c: of course interested *_*
Hello! Im so happy that i'v found your model! Whats the recommendations for Sampling, Schedule, Steps and CFG?
Well, I didn't get much deeper into the sampling - pretty much all the latest ones seems pretty good.
About the schedule, I'm a fan of "exponential" and I'm feeling comfortable with 4.5-5.5 CFG scale and 50 steps (as I've said, they seems a lot, but less steps usually brings to less developed concepts). ✨
Can't wait to see what you're going to create! 👌🏻✨
@Arthemy Thank you! Thats important
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.