Your current browser has compatibility issues. Please switch toChrome for uninterrupted access.

[Exclusive on the whole internet] - Japanese restaurant interior LORA

0
11
0
0
RealisticInterior DesignSpace & SceneObject Enhance
Recently Updated: 23/11/09First Published: 23/11/09
Realistic,Interior Design,Space & Scene,LoRA,SD 1.5Image info
Realistic,Interior Design,Space & Scene,LoRA,SD 1.5Image info
Realistic,Interior Design,Space & Scene,LoRA,SD 1.5Image info
Realistic,Interior Design,Space & Scene,LoRA,SD 1.5Image info
Realistic,Interior Design,Space & Scene,LoRA,SD 1.5Image info
Realistic,Interior Design,Space & Scene,LoRA,SD 1.5Image info
Realistic,Interior Design,Space & Scene,LoRA,SD 1.5Image info
Realistic,Interior Design,Space & Scene,LoRA,SD 1.5Image info
Realistic,Interior Design,Space & Scene,LoRA,SD 1.5Image info

The website is full of many beautiful pictures of LORA. I downloaded some. And then...

Teaching everyone how to practice this kind of Dan, REPEAT from 50 to below 100, depending on your mood.

Find a set of beautiful materials online, for the same project, about 10 pictures are enough, label directly, CLIP for reverse push, no need to modify, optimizer select Shen Tong, if no Shen Tong, select ADAN, if no ADAN, select ADAW, aiming for overfitting. Train for about 10 EPOCH. All other parameters are default.

When the LOSS value is below 0.05, you can stop directly, hang the final LORA, weight from 0.6-0.8, use the prompts for the reverse push, then you can start spitting out the original images.

Then run a few high-resolution photo repairs of this Dan, share them on the website for people to see the effect, this Dan is awesome.

Planting LORA is awesome, except it's not very useful, everything else is an advantage.


Ah, it's really hard to find a useful LORA. The website that was not already rich is even more miserable.

Expand All
Translate into English

Discussion

Most popular
|
Newest
Send
No more comments
laowang
Artist
114
112
66.9k
784
Favorite
Download
(144.11MB)
Verified:2023/11/09
safetensors
Version Details
Type
LoRA
Generation Count
0
Downloads
11
Base Model
SD 1.5
Recommended Parameters
Recommended Checkpoint
None
Recommended weight
CFG
7
VAE
None
License
Can be generated online
Can be merged
Commercial LicenseCommercial Use Rules
Generated images cannot be used for commercial purposes
Models cannot be resold or sold as merged models
*Licensing scope is set by the creator, users must use as per requirement and standards

Gallery1

Most popular
|
Newest
avatar
laowang
Image info