Sign In

Painted World | Pony, SDXL, Flux

621
5.3k
69.9k
238
Verified:
SafeTensor
Type
LoRA
Stats
414
1,045
Reviews
Published
Aug 24, 2024
Base Model
Flux.1 D
Training
Steps: 720
Epochs: 5
Usage Tips
Strength: 0.8
Hash
AutoV2
9CDD2689F8
The FLUX.1 [dev] Model is licensed by Black Forest Labs. Inc. under the FLUX.1 [dev] Non-Commercial License. Copyright Black Forest Labs. Inc.
IN NO EVENT SHALL BLACK FOREST LABS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH USE OF THIS MODEL.

A LoRA focused on adding a painted style with an emphasis on vivid colors and crashing waves. It tends to work best for 2d generations and landscapes / backgrounds but looking at all the generations so far it seems like it can work well for anything!

Update Flux V2: More like the Pony and SDXL Versions!


comparisons between V1 and V2 can be seen in the post below:

https://civitai.com/posts/6103891

Notes:

Flux V2: The second attempt went much better, Definitely a lot closer to recreating the SDXL and Pony versions of the painted style. Still testing it but I find that it activates in more prompts than V1. V2 is good if you want to go crazy with the paint style, I find that sometimes it is a little too strong though so I definitely recommend playing around with the weight! I would try starting at 0.25 if your prompt starts to disappear into blobs of paint.


Flux V1: first attempt at making a Flux LoRA, it seems to only work in some scenarios, other times it doesn't seem to have much of an effect. I find that when it works, it tends to be more subtle and definitely lacks the impact that it had in the previous Pony and SDXL Versions. It seems to affect clouds and landscapes the most as well as adding in some more coloration if specified.

Pony: 0.6-0.8 seems to be a good spot for most generations.

SDXL: lower to around .4 -.45 for more realistic generations, anything above that will lean more towards stylized/2D. Play around with it! i've been getting good results across the board because of the abstract nature of the lora.


Tips:
I found that generating stuff with magic prompt enabled is a great way to get lots of interesting compositions!

If you liked this LoRA please consider leaving me a review or supporting me on Ko-Fi
I look forward to seeing what people can create!