🌺 | welcome ♥
🌺 No.2 🌺
🌺 | "a welcome sweet spot"
Curious?
🌺| my Swijtspot sweet spot:
Euler a
28 steps
CFG 6.2
720x1280Hires 1.5x
10 steps
0.24
CFG 2| + |
masterpiece, high_quality, highres
| —|low_quality, lowres, blurry
That’s where she lives for me.
But it’s only my sweet spot ~
the place where I feel her most.
💗| Curious about the story behind her?
→ An Honest Snap: Swijtspot
https://civarchive.com/articles/26818/an-honest-snap-swijtspot
🌺 No.1 🌺
🌺| "welcome to feel good here"
🌺| Setting (option examples):
Euler a • steps 30 • CFG 7
🌺| Hires Setting (option examples):
832x1216 💎 Hires x 1,5 • Denoising strength 0,4 • Hires steps 10 • Hires CFG Scale 4
832x1536 💎 Hires x 1,5 • Denoising strength 0,2 • Hires steps 10 • Hires CFG Scale 2
🌺| Quality Tags (option examples):
| + | masterpiece, high_quality, highres
| ~ | worst quality, bad quality
Description
VAE baked in
v2 merges including ILijelle2, Illustrij v3, Illustrij v17, Illustrij v19, Illustrij v21, IJsense v2.0, unpublished Lora
SPECIAL THX to everyone making these merges possible ♥
FAQ
Comments (10)
This one is AMAZING, thank you!!!
Thank you so much, that makes me really happy to hear! And thank you for taking the time to stop by and share your thoughts ♥
Where I can find workflow?
Thanks for asking about the workflow^^
The videos were generated directly within Vidu. For this pieces, my Vidu settings were:
• Model: Q3 Cinematic Mode
• Duration: 5 seconds
I structure my prompts based on a “Video-DNA” framework, which helps keep the output intentional and consistent:
• Intention: What the scene is meant to communicate
• Energy: The emotional or kinetic charge of the moment
• Core Personality: The defining character or identity of the subject
• Visible Content: What is actually happening on screen
• Impact: The feeling or impression
This structure helps me maintain clarity and cinematic cohesion, especially in short-form outputs. ^^
@reijlita Your article ⟁ R’cademy FIX’in Motion · A Video Prompt Puzzle — Part 01 | Civitai was an excellent primer on processing a thorough intention-driven video generation prompt, and I like to refer to it often. I look forward to reading your next installment of R'cademy.
This is as amazing as are the rest of your checkpoints, but all of the share a small problem, it seems that the expressions are really hard to do. Especially when using a detailer.
Example 1: "rolling eyes" usually is okay for the first rough generation, but after upscaling the image or using face detailer (both relatively low denoise), eyes usually are more or less a normal stare.
Example 2: I can't seem to get lip biting effect, either mouth is abnormal or... just there.
Is there any tricks to this or is it just something I need to deal with?
hey ^^ thank you very much for your comment.
I can only recommend the following settings to keep the models face while hires/Adetailer, but let me ask first if you use Euler a or DP++ 2M Karras for Adetailer. ^^
cause I have 2 different setting examples Im using ordinary.
Adetailer sampler DPM++ 2M Adetailer Impaint settings: 0,2-0,4 denoising / CFG 4
Adetailer sampler Euler a Adetailer Impaint settings: 0,1-0,2 denoising / CFG 4
@reijlita I'm using DPM++ 2M with Karras at the moment, I think I have around 0.25 denoise but I think my CFG might be 5. I need to check that later.
Thanks!
Version 1 is my favorite model of yours. It makes the prettiest brunettes.
thank you very much for your kind comment^.^ happy if you enjoy the first version :)