Forum Replies Created
-
Posts
-
That’s correct.
If you want to generate furniture with a similar style, you can use an IP-adapter controlnet and supply your images.
You can first start with prompting to see if you are about to get you want. Sdxl model is a good starting point.
you can train a Lora if you want to incorporate your own furniture.
Hi Iris, you can use the Adetailer extension to fix faces in each generation automatically.
You will need to upload it to a public site like imgur.com and post the link.
Or send me an email if you prefer. [email protected]
Thanks for reporting the issue. Civitai has added a login requirement to downloading models, so the “download from URL” function no longer works for them.
Please download the model to your Google Drive folder: AI_PICS/models
Or find a Huggingface link. It should still work.
Hi, this is possible with inpainting with a manual mask. You will need to create a mask with photo editing software. The mask should be white outside the product (where you want to add background), and black on the product (where you don’t wish to change). You also need to enable controlnet depth or lineart.
Hi Iris, I cannot see your pictures. Currently, Stable Diffusion can generate images from a sketch but not the other way around.
It could be an error in the workflow. Do you have an reference image? Sharing the workflow would help troubleshoot.
Negative prompts like “deformed, disfigured” can beautify images. We only need to test them in the positive prompt:
SD uses a frozen CLIP language model from OpenAI. Although no one trains images using these keywords, it still knows what it means semantically.
However, I don’t think they can fix fingers.