Hello everyone,
This Checkpoint Merge aims for a Semi-Realistic style.
I have used these settings almost exclusively. They are a good starting point and can be adjusted depending on what you want to generate.
Sampling method:
DPM++ 2M Karras
DPM++ 2M SDE Karras
Sampling Steps:
25-45
CFG Scale: 7
clip skip 2
Recommended Upscaler Settings:
Hires. fix Upscaler:
R-ESRGAN 4x+
Hires steps: 10-20
Hires upscale: 1.5-2
Denoising strength: ~ 0.3-0.5
For better results use Hires.fix for better Results!
I recommend don't use restore faces. Use ADetailer for a better face!
I would greatly appreciate any feedback!
If you like this model, I would greatly appreciate a positive review along with constructive criticism, suggestions for improvement, or anything similar!
Description
Hello dear people! I've created and tested 4-5 different versions. However, they all seemed too realistic to me or leaned too much towards realspice. Now, with this version, it's more like the other semi-realistic/3D style ones. Additionally, I've finally started using multiple negative prompts and adjusting them when I'm not satisfied with something in the output. I am very pleased with the composition in this update.
I hope you have fun with it!
FAQ
Comments (24)
I first read the title as "AnalReal". Blush
One of the greatest models ever.
Looks awesome. Any plans for an SDXL 1.0 version?
Thank you! No, I don't have any plans for it at the moment. Currently, I don't have a good enough pc to do this. Maybe sometime in the future.
@Fusch I understand. Will cross my fingers for a possible release in the future ;) Take good care and keep up the good work!
@Chonsu Thank you for your kind words! :)
Great model!
Looks great!
Thank you very much! I'm happy to hear that. Especially from you! ❤️
Looks very good; downloading now to try via AegisFlow 1.5
Coming back to comment---this is a great semi-real anime model, but can do other stuff well too. I particularly like its interaction with painterly LoRAs, which can inject some feeling into what can otherwise be beautiful, but also a bit common, Waifus.
Thank you~! Your model gives me another way to create images with different styles. The outcome is great and I highly appreciated it !
I'm happy that you like it! Thank you for your kind words! ❤️
this will be my new favorite..well done amazing..
Why after doing "generate", the color of the image becomes more faded.
try using a VAE
same question, wtk how to solve
This is most likely a VAE issue
v10 is much better WTF
whats the diff between this one and anrealspicemix?
Generates images, but at the final stage turns them into squares, filled with 1 colour. Just one BG colour without anything in it. Yesterday changing hirex to smaller number helped, but today nothing helps at all, even repeating the settings after the author. I have no idea, why it does not work for me.
This is the only model i use that don't have VAE baked on it, feel like it really should have
Details
Files
Available On (2 platforms)
Same model published on other platforms. May have additional downloads or version variants.