SXZ Luma Experimental - DEv4
Prompts négatifs recommandés
easynegative,easynegativev2
(bad quality, worst quality, normal quality:1.2), 3d, realistic, bad hands, bad anatomy, text, watermark
Paramètres recommandés
samplers
steps
cfg
clip skip
resolution
vae
Paramètres haute résolution recommandés
upscaler
upscale
steps
denoising strength
Conseils
Use short and meaningful tags in prompts for best results.
Add extra styles like anime, oil painting, flat color, or cartoon to diversify outputs.
Avoid using Face Restoration; instead, use AutoChar or AfterDetailer for better faces.
Feel free to experiment with different parameters.
Points forts de la version
Finetuned with pics and promts from Dall-E 3 via Bing Image Creator.
Mostly trained in text encoder part.
Very flexible, but sometimes may produce artifacts with Lora's on high values.
Sponsors du créateur
All the content I make is and will be completely free on CivitAI.
Recommended embeddings:
easynegative
easynegativev2
bad-hands-5
badhandv4
Recommended extensions:
AutoChar
AfterDetailer
🌕 LUMA Experimental Edition
This is Luma-based semi-realistic checkpoint, finetuned experimentally with different datasets, ways of captioning and training parameters.
Please note, that it can be incompatible with some of Loras on high weights and may produce some overburns.
Model1: Luma DEv4 - finetuned with some images and prompts from Dall-E 3.
I highly recommend to use short and meaningful tags in your prompts. You may see some examples in sample images. Also this model is very flexible, so you may add some extra styles (like anime, oil painting, flat color, cartoon, etc.) for better variations.
All the content I make is and will be completely free on CivitAI.
But if you want to support me with a coin, then you can do it here.
Usage
Recommended embeddings:
1. easynegative
2. easynegativev2
3. bad-hands-5
4. badhandv4
Recommended extensions:
1. AutoChar
2. AfterDetailer
My regular settings:
Clip Skip 2
Necessary negative prompt: easynegative or easynegativev2
VAE: vae-ft-mse-840000
Sampler: any
ENSD: 31337
Hires.fix: x1.5 and higher
Upscaler:
Option 1: 8x_NMKD-Superscale_150000_G for realism or R-ESRGAN 4x+ Anime6B for illustrations (Denoising strenght 0.3-0.5)
Option 2: Latent (any) or Lanczos (Denoising strength 0.5-0.6)
Never use Face Restoration!
If you want better faces just use AutoChar or AfterDetailer.
But you may feel free to experiment.
Contributeur
Détails du modèle
Type de modèle
Modèle de base
Version du modèle
Hash du modèle
Créateur
Discussion
Veuillez vous log in pour laisser un commentaire.