Downloads · 30 days
0
bedalton/C1-V11
C1-V11 is a machine learning model from bedalton. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as mit.
This model is meant to generate scenery like that in the game creatures 1. It is not meant to make objects or really do anything else.
Downloads · 30 days
0
Access
Public
Updated Aug 21, 2023
Repo size
12.9 GB
Likes
0
Public
Click a slice to open those files.
.ckpt12.9 GB · 100%
From the Hugging Face model README
This model is meant to generate scenery like that in the game creatures 1. It is not meant to make objects or really do anything else.
Trained with Everydream2 on top of a Stable Diffusion 1.5
This model is definitely over trained, some prompts give really bad results - like anything involving a tree house, anything related to water, and lots of things to do with caves
Included are the checkpoints at Epoch 70, 80 and 85.
Epoch 70 is the most general, but also tends to look powdery or like it was colored with pencils. It works well with negative prompts like "illustration" and "pencil"
Epoch 80 the objects are not powdery, but also more yellow or saturated (i.e. Overtrained), and sometimes look like cartoons. Also, this checkpoint tends to add branches or blobs of brown at the top of the images it generates.
Neither checkpoint is really recommended. It is best to use the extracted LORAs
Recently added 2 LORAs extracted from checkpoints for epochs 70 and 80. These checkpoints have no activation words
If scenery image lacks the C1 straight on view, try adding cutaway to the prompt. Sometimes that helps, sometimes though it makes things much worse.
Images, like the actual Creatures 1 world, will mostly be tinted yellow
The LORAs work well when paired with Dreamlike-Diffusion, at around 75% strength.
Dreamlike diffusion with this LORA, tends to add a house or building in the middle of the image, even when you do not expect one.