The model is a lora model trained on the basis of the SDXL base model. It aims to solve the problem that SDXL is not very satisfactory in expressing indoor scenes, enhance the texture of light and shadow and some style elements, etc. It is recommended that you can run the map on LibLib's online SD, which does not require configuration.
If you use the SDXL model locally, it is recommended that the video memory is at least ≥8G. If it is 8G, it is recommended to follow the following settings (I have a 2070 laptop with 8G video memory and use the Autumn Leaf Launcher). If it is a 3090 or 4090 graphics card, please feel free to do so:
Written in the front: It is recommended that you use natural language to describe your picture. Remember, it is natural language! ! Natural language!
The weight recommended by the Lora model: the default "1" is OK. If the effect is not right, please adjust it between 0.6~1.0. You can also increase it to 1.2. Remember, the maximum value should not exceed 1.2, otherwise accidents will occur! ! ! !
There is a trigger word, which is recommended to be written at the beginning: mrares
Video memory optimization is selected as medium video memory, and half-precision optimization is turned off.
This lora covers a variety of styles and elements. It is recommended that you add corresponding prompts before the specific style you want. If some styles do not work after adding them, please try adding the underscore symbol "_". If you need to enhance the photography texture, you can add the tag "depth of field".
Style tags:
1. New Chinese style: New Chinese style/New Chinese_style; Chinese elements
2. Nordic style: Nordic style/Nordic_style; Nordic elements
3. Wabi-sabi style: wabi-sabi style/ wabi-sabi_style; wabi-sabi elements
4. French style: French style/ French_style; French elements
There are also industrial style, modern, modern minimalist, Bohemian, Mediterranean and other styles
Space type tags:
Home decoration: living room; bedroom; bathroom; kitchen; study/study room/children's room/ dining romm, etc., as well as cloakrooms and storage rooms
Workwear: mainly restaurants and commercial stores, everyone can try it
———————————————————————————————————————————
When using the SDXL model, it is recommended that you turn on refiner, and the sampling value can be between 20-25, and it does not need to be too high.
———————————————————————————————————————————
Then, let’s give a few examples (so that everyone will know more clearly how to describe and draw pictures):
《工业风餐厅》
Prompts:
in an industrial style restaurant,there are many dining tables and chairs arranged,industrial style decorations hang on the wall,pendant,a cool tone,wide angle,long-focus,depth of field, . 35mm photograph, film, bokeh, professional, 4k, highly detailed
Negative Prompts:
drawing, painting, crayon, sketch, graphite, impressionist, noisy, blurry, soft, deformed, ugly, lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry
《现代简约家居》
Prompts:
indoor,(((modern minimalist style))),a living room,a warm atmosphere,there is a coffee table next to the long sofa,an irregular table,there are interesting art paintings hanging on the wall, . 35mm photograph, film, bokeh, professional, 4k, highly detailed
Negative Prompts:
drawing, painting, crayon, sketch, graphite, impressionist, noisy, blurry, soft, deformed, ugly, lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry
Finally, I hope you will send me more pictures and try more!
This model is the V1.0 test version, I hope you will test it more
Please send me more pictures. If the feedback is good, I will open it for download soon. If the effect is not satisfactory, I will further adjust the relevant content and open it for download after it is improved. Thank you all~
Description
FAQ
Comments (1)
Looks like a great Lora. Gonna give it a shot. That said..
"It is recommended that you can run the map on LibLib's online SD"
I'd personally recommend people get a local setup if they plan on generating on the regular. If they're gonna generate a couple images, have a chuckle and move on, then that's not feasible.
But for those that are gonna sit there and bust out 1000's of images, get a local setup, cuz you'll pay less in the long run, and then you don't have any big corporation dictating what you can make, or spying on your personal creations.
Avoiding the easy route is the only way to keep your freedoms and your privacy. Cuz corporations will take it all otherwise, and you'll be supporting them financially in doing so.
Also, download software and store current/old versions while you can. Cuz eventually they'll hard-wire in updates, which seems convenient at first, until they start forcing updates, then eventually charging you for updates, or charging a fee to use the program.
They've done it a lot already, there's some people that can't even use printers that they bought anymore, cuz they aren't paying monthly fees for ink services.
If you think I'm being paranoid, consider this; everything you currently pay a monthly fee on, used to be available for a one time purchase, or even free at one point, or at the very least, much cheaper and more fully featured.
So, take advantage of what is free while it still is, and lock it in at that cost by sandboxing your AI PC, or at least creating backups of all the software, models dependencies, etc., that way you can create a sandboxed AI PC for when it becomes impossible to generate anything without a fee.
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.