DAINAPP
patreon
SD GUI 0.62
🕑 Added 2022-12-14 13:58:15 +0000 UTCWhile my search continue for a better host, here is a link for drive and a mirror for it.
Also made an extra build that is exactly like the regular build, but without the "torch" folder, so you just need to copy the torch folder from another build to this ne. It made the .rar a lot smaller.
SD GUI 0.62:
https://drive.google.com/file/d/1NwlsU81-ryPlArgw_ceXmS358OwW_PxX/view?usp=share_link
Mirror:
https://drive.google.com/file/d/14_IqqLf4vJaE0-qXf1OtyKCVi7pimmnv/view?usp=share_link
SD GUI 0.62 [Without the torch folder]:
https://drive.google.com/file/d/1ekMzk-6jeIVc4CP6_U-Vbt3GF31m5GjH/view?usp=share_link
https://drive.google.com/file/d/1rreS3q2mkhi524Qqw6LO5X1w0mCfi9Uw/view?usp=share_link
Link for the models [They are the same for the previous post]:
https://mega.nz/folder/2x1CBRJL#MzMfyijznMs-2t-VqBXYPQ
Driver Mirror:
https://drive.google.com/drive/folders/1MqjzOisTaV0Y1fBfYRVRS9fG6UrFOkK7?usp=share_link
This build have most bug fixes and small improvements.
- Some scredulers have some extra informations. Like the DDPM scheduler require ~1000 steps to work, so it show that information on the list.
- V2 models can't be trained with any scheduler, so schedulers for V2 that can be used on Dreambooth have an extra [dreambooth] on the name. After training, you can use any scheduler on your new dreambooth model.
- Made some chances on how memory is handled, to stop black images from rendering using V2.1 + [xformer disabled]. This may affect vram usage, so let me know if the GUI is using more Vram.
- Small changes on dreambooth, may train a little better.
- Fix Dreambooth model loading.
- There is some new schedulers that use steps*2 to render correctly. I still didn't put this information on the name, but keep in mind that some schedulers will do double the steps.
- Made some changes to the selection of models folders, should work a little better.
Will start reading and answering all the messages here in just a few minutes, but hopeful this version will fix most of the problems.
Right now i'm starting to implement the upscaler models for SD, if the results are good, will replace the current upscalers for those.
I'm also starting a GUI for OpenAi Whispers, that is used to transcribe audio files to text and also do translations.
Comments
When using ANY of the Upscale options- it works but see new message/error 66it [00:40, 1.61it/s] Pipelines loaded with `torch_dtype=torch.float16` cannot run with `cpu` device. It is not recommended to move them to `cpu` as running them will fail. Please make sure to use an accelerator to run the pipeline in inference, due to the lack of support for`float16` operations on this device in PyTorch. Please, remove the `torch_dtype=torch.float16` argument, or use another device for inference. Loading Facefix Done - if select float32 checkbox this error goes away.
Using anything v3 VAE-it renders but see this error a render start - Loading VAE making attention of type 'vanilla' with 512 in_channels Working with z of shape (1, 4, 32, 32) = 4096 dimensions. making attention of type 'vanilla' with 512 in_channels loaded pretrained LPIPS loss from taming/modules/autoencoder/lpips\vgg.pth Loading Hyper Loading Scheduler Using Half Header Prompts in Queue: 1 Rendering Text2Img:
![カネコツ[kanesys] from fanbox](https://patrefans.com/kistorage/20926.jpg)
