Forum Replies Created
-
AuthorPosts
-
April 2, 2024 at 12:42 am in reply to: Best Realistic Portrait Generation? (Dreambooth, Lora, InstantID etc) #13226
Hi, it seems that your trained model is not stable. You can:
- Train from SD 1.5 if possible. The model is in a better state compared to realistic vision.
- Try reducing the number of training steps and/or learning rate. Use the least you can get away with.
March 31, 2024 at 2:50 pm in reply to: Best Realistic Portrait Generation? (Dreambooth, Lora, InstantID etc) #13221Great!
Dreambooth on SDXL consumes too much memory. I can’t find a way around it on Colab. You can try training a LoRA on SDXL.
Realistic Vision is a good starting point for what you want to achieve.
March 29, 2024 at 7:34 am in reply to: Best Realistic Portrait Generation? (Dreambooth, Lora, InstantID etc) #13201Welcome!
Dreambooth is the most powerful method, so it should work.
You can try
- Get better input pictures with quality similar to my training example images.
-
If your face doesn’t show, train more steps.
Thanks for the compliment! Great idea. I will add a feedback section when I got a chance.
Welcome! It’s great to have an expert like you here!
Hi! I updated the notebook today (3/21). Give it a try.
I’m not aware of such controlnet model.
The SDXL canny controlnet works pretty well.
You use SD1.5 segmentation controlnet to generate something close and use multiple rounds of img2img with controlnet to transform the image to an SDXL style.
Hi, this sounds like an issue with SD Forge. Please file an issue in their repository.
Yes, that’s all it takes.
The E2E workflow should be accessible now.
Thanks for reporting the issue!
The issue is fixed. Thanks!
Thanks for reporting issue! Can you send me the link to the quiz?
Hi! I don’t recall seeing this artifact. You can try the new soft inpainting feature in A1111. Increase mask blur to 20 or more. You can use high denoising strength.
(tutorial coming soon)
-
This reply was modified 1 year, 1 month ago by
Andrew.
The first image can be from an input image. But animatediff will change it big time.
It is not possible to insert an image during the video generation.
I will add this to my list.
Hi! This list is beyond my resources to collect and maintain. Perhaps one approach is to only use a few of them so that it is easier to understand and monitor.
-
AuthorPosts