Hello everyone, I am an architect and museum exhibition designer，you can call me 'Moogone', I am very frustrated that it is difficult to find AI models for public space interior design online. Therefore, I have decided to take matters into my own hands. I will be publishing various types of interior design models for public spaces on my Civitai account. These may include museums, planning halls, corporate management spaces, temporary exhibitions, game exhibitions, and other categories. Please follow me to avoid missing out.
大家好，我是一名建筑及博物馆展陈设计师。我很苦恼为什么网络上很难找到公共空间室内设计的AI模型，所以我打算亲自动手了。以后我会在我的Civitai账号上发布各种各样类型的室内公共空间设计模型。可能会涵盖博物馆、规划馆、企业管理、临展、游戏展览等各种各样的类别。大家请关注我以防失联。
这是一个关于图书馆的dreambooth模型，因为我最近在做图书馆室内设计，所以整合了这个模型用于项目前期生成图书馆的一些效果图，我用了30多张选定的图书馆意向图，这些意向图精挑细选，涵盖的方面比较多，所以我为此对每一张图片编辑了详细的.txt描述文档，以此控制模型不过度发散，但是我同时也在每一个文本中放了一个token：ohwx，所以当你要调用这些数据的时候最好先在开头输入ohwx.
Description
Please use keyword: "ohwx" to activate this lora
FAQ
Comments (18)
You didn't have to make a whole checkpoint for a single subject. A Lora would more than suffice.
You are theoretically correct, but in my actual testing, the impact of Lora on the results is too small to generate expected and good results. Directly training the model in a larger model will yield better results. You can also try training Lora and checkpoint on the same set of images to see the difference between the two. If you're interested, I can also share another Lora model trained on exactly the same data.
Sure, I'd like to check out the Lora whenever you have the time to upload.
I would be interested in seeing both too actually.
The advantage of a Lora is that you can combine it with other models.
@Moogone This is an aspect of SD that every other model maker has seen, and yet LORA is still preferred for this type of model.... it just takes a little more user skill to utilize effectively. LORA is not the same as a checkpoint, but so as long as you expect that, it shouldn't prohibit you from making LORAs.
@Nexustar Thank you for your reply. I agree with your assessment, as I have limited knowledge of SD and am not yet proficient in the necessary techniques. While exploring the Dreambooth training options, I was curious and selected the "LORA" option, but the resulting model was much larger than expected. Moving forward, I plan to focus on training large models in the correct manner, integrating more materials as museums cover a vast array of subjects with specific directions. Ultimately, my goal is to create high-quality commercial models, and it seems that LORA may not be the best fit for this purpose. Thank you again for your response and for providing me with this valuable information. ^^
@Polygon checkpoint can do it now
@Moogone If you're set on making large models, a good compromise could be to incorporate multiple interior design subjects in each one. Something like that could be quite useful.
@kawaiibully Yes I know that, and this is the LoRA version trained by same pictures: https://civitai.com/models/61206/loralibrary-interior-design-v001
It feels more like a bookstore to me, but either way, it's a large interior space full of shelves filled with books. Which means I like it! I've actually been hoping to find something like this as well, so I'll try it out.
Ha ha, I was a bit casual when naming it. Please don't mind. But you are right, 60% of the samples used for training are bookstores. I'm glad you like this model. I will also release more indoor design models in the future, but they will be more focused on exhibition halls
@Moogone My local library branch is at the mall and looks a bit like a bookstore. I guess designs have changed a lot from when I was a kid.
When to release a model of a storefront ?
Within 3 days
Although I am not an architect, I still want to thank you
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.