Home
Artists
Search Recent Random
Posts
Search DMs Popular Hash Lookup Tags Random
Importer
Import FAQ
Register
Partychan Telegram Find&Fuck Waifus ThePornDude
Home Artists Posts Import Register

SD 0.68 (Patreon)

Published:
2023-02-22 01:11:45
Imported:
2023-02

Content

This is not a big update.
Mostly I was trying to redo the code that handle the Vram Memory, to see if I could fix it to work on 8Vram again.

If you had problem with the latest problem, please let me know if this fix it.

As a bonus, there is a new option. 

Prompt2Img+(Upscale)Img2Img

This will do a regular prompt2image, upscale the image and run img2img on it.

You can select the upscale % and img2img strength in this mode. 50% seen to work good for me.


Link:

https://drive.google.com/drive/folders/1a-wrwfPw5gb_ohHzeki6MHiisZ80b4ho?usp=share_link 

Comments

cool1

In the dreambooth option of the GUI could we please have an option to stop/pause the training where we could then exit the app and then at some later time reload the app and select "resume last training" or something similar (maybe you could select which one to resume) and then it would carry on from where it left off. The reason is sometimes it can affect other things you need to do at the same time, like it might stop another app from loading (because of the resources being used) or it might make a particular app run too slowly or you need to reboot your pc. With a resume option that would solve that by freeing up the resources then continuing later. Since it can save a .pt file every x amount of training steps maybe something similar could be achieved by copying the .pt file to the right place for source models and then loading that for the source, decreasing the amount of training steps to do and then starting that (maybe taking into account that the actual number of training steps needs to be a multiple of the number of images I think). But a resume option could do something similar to above - where pausing could save a .pt file and another file if necessary (that stores any additional info needed to later resume).

cool1 (edited)

Comment edits

2023-03-24 11:13:56 Also in the configuration tab there's an option for "Change Models Folder". If possible could that please be expanded so it's a bit more specific / so you can set different models in different drives/folders. eg. if could allow you to have the main model files (eg. SD 1.5, SD 2.1 etc) on the C drive and when running dreambooth where it saves a file of around 2.5 GB every x training steps (or around 4 GB if you can train with 32 bits) then that uses a lot of space so it would be good to be able to use a different, bigger hard drive for that, even if it's a bit slower. eg. Configuration Settings: Standard models drive & folder: C:\Stable-Diffusion-GUI....\models Dreambooth output drive & folder: F:\Stable-Diff-Dreambooth-Output That would allow fast loading of standard models (eg. using a fast SSD) and seperating the dreambooth output folder would allow much more disk space for models output at enough frequency while training without worrying if you might run out of space on the drive. I don't know much about Lora training but if that was an option that could also mean disk space would be less of an issue too (ie. generating output models at enough frequency of training steps).
2023-03-08 14:16:22 Also in the configuration tab there's an option for "Change Models Folder". If possible could that please be expanded so it's a bit more specific / so you can set different models in different drives/folders. eg. if could allow you to have the main model files (eg. SD 1.5, SD 2.1 etc) on the C drive and when running dreambooth where it saves a file of around 2.5 GB every x training steps (or around 4 GB if you can train with 32 bits) then that uses a lot of space so it would be good to be able to use a different, bigger hard drive for that, even if it's a bit slower. eg. Configuration Settings: Standard models drive & folder: C:\Stable-Diffusion-GUI....\models Dreambooth output drive & folder: F:\Stable-Diff-Dreambooth-Output That would allow fast loading of standard models (eg. using a fast SSD) and seperating the dreambooth output folder would allow much more disk space for models output at enough frequency while training without worrying if you might run out of space on the drive. I don't know much about Lora training but if that was an option that could also mean disk space would be less of an issue too (ie. generating output models at enough frequency of training steps).

Also in the configuration tab there's an option for "Change Models Folder". If possible could that please be expanded so it's a bit more specific / so you can set different models in different drives/folders. eg. if could allow you to have the main model files (eg. SD 1.5, SD 2.1 etc) on the C drive and when running dreambooth where it saves a file of around 2.5 GB every x training steps (or around 4 GB if you can train with 32 bits) then that uses a lot of space so it would be good to be able to use a different, bigger hard drive for that, even if it's a bit slower. eg. Configuration Settings: Standard models drive & folder: C:\Stable-Diffusion-GUI....\models Dreambooth output drive & folder: F:\Stable-Diff-Dreambooth-Output That would allow fast loading of standard models (eg. using a fast SSD) and seperating the dreambooth output folder would allow much more disk space for models output at enough frequency while training without worrying if you might run out of space on the drive. I don't know much about Lora training but if that was an option that could also mean disk space would be less of an issue too (ie. generating output models at enough frequency of training steps).

Anonymous

seamless options not available for img2img ?