Commit
•
ba4e69e
1
Parent(s):
25f7223
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,60 @@
|
|
| 1 |
-
|
| 2 |
-
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Penelope Palette: Portrait Generation Model
|
| 2 |
+
|
| 3 |
+
Important note : Provisory Model card
|
| 4 |
+
|
| 5 |
+
## Model Description
|
| 6 |
+
Penelope Palette is an advanced AI model designed for creating lifelike portraits. It leverages the same architecture as Stable Diffusion 3, ensuring high-quality image generation with remarkable detail and style. Most of the description was copied from the stable diffusion 3 since the informations remains generally the same.
|
| 7 |
+
The model is weaker than Stable Diffusion 3 medium , having trouble generating realistic content ; nudity and anatomy but it performs really good in portraits , having a unique style .
|
| 8 |
+
## Model Description
|
| 9 |
+
Developed by: Penelope Systems
|
| 10 |
+
|
| 11 |
+
Model type: MMDiT text-to-image generative model
|
| 12 |
+
|
| 13 |
+
Model Description: This is a model that can be used to generate images based on text prompts. It is a Multimodal Diffusion Transformer (https://arxiv.org/abs/2403.03206) that uses three fixed, pretrained text encoders (OpenCLIP-ViT/G, CLIP-ViT/L and T5-xxl)
|
| 14 |
+
|
| 15 |
+
# License
|
| 16 |
+
Apache llicense 2.0
|
| 17 |
+
|
| 18 |
+
# Model Sources
|
| 19 |
+
|
| 20 |
+
For local or self-hosted use, we recommend ComfyUI for inference.
|
| 21 |
+
It has built-in clip so it shoul be plug & play .
|
| 22 |
+
|
| 23 |
+
ComfyUI: https://github.com/comfyanonymous/ComfyUI
|
| 24 |
+
|
| 25 |
+
|
| 26 |
+
# Training Dataset
|
| 27 |
+
We used synthetic data and filtered publicly available data to train our models. The model was pre-trained on 1 billion images. The fine-tuning data includes 30M high-quality aesthetic images focused on specific visual content and style, as well as 3M preference data images.
|
| 28 |
+
|
| 29 |
+
|
| 30 |
+
# Uses
|
| 31 |
+
Intended Uses
|
| 32 |
+
Intended uses include the following:
|
| 33 |
+
|
| 34 |
+
Generation of artworks and use in design and other artistic processes.
|
| 35 |
+
Applications in educational or creative tools.
|
| 36 |
+
Research on generative models, including understanding the limitations of generative models.
|
| 37 |
+
|
| 38 |
+
|
| 39 |
+
# Out-of-Scope Uses
|
| 40 |
+
The model was not trained to be factual or true representations of people or events. As such, using the model to generate such content is out-of-scope of the abilities of this model.
|
| 41 |
+
|
| 42 |
+
#Safety
|
| 43 |
+
Same safety measures used by Stable Diffusion 3 were deployed .
|
| 44 |
+
|
| 45 |
+
# Use recommendations :
|
| 46 |
+
For best use we recommand :
|
| 47 |
+
- steps : 32
|
| 48 |
+
- cfg : between 4.0 and 7.0
|
| 49 |
+
- sampler_name : dpmpp_2m
|
| 50 |
+
- scheduler : sgm_uniform
|
| 51 |
+
|
| 52 |
+
|
| 53 |
+
|
| 54 |
+
|
| 55 |
+
|
| 56 |
+
|
| 57 |
+
|
| 58 |
+
|
| 59 |
+
|
| 60 |
+
|