Hugging Face's logo

SG161222
/
RealVisXL_V4.0

Text-to-Image
Diffusers
Safetensors
StableDiffusionXLPipeline
Model card Files
xet
 Community
17

Next version will be for Stable Diffusion 3.0?

#2
by makaveli-313 - opened
Discussion
makaveli-313

Hello,

Firstly I am a very big fan of your model, the best photorealistic one I have come across thusfar!
I am more of a hobbyist category use and render a few hours every other day. I wanted to know if there will be a RealVisXL v5.0 on existing sdxl 2.1 or will you just release the next version for SD 3.0?

Thank you for your time and efforts : )
SG161222
 Owner

Hi! Thank you so much! I think RealVis on SD 3.0 base is my next goal (if I can train the model locally).
makaveli-313

Hi! Thank you so much! I think RealVis on SD 3.0 base is my next goal (if I can train the model locally).

Thank you for the response. I hope it will be possible.
Looking forward to the next iteration : )
makaveli-313 changed discussion status to closed
makaveli-313 changed discussion status to open
makaveli-313

Hi! Thank you so much! I think RealVis on SD 3.0 base is my next goal (if I can train the model locally).

Sorry I forgot to ask earlier,
How long after Stable Diffusion 3.0 comes out can one expect RealVis on 3.0?

Thanks again.
SG161222
 Owner

Hi! Thank you so much! I think RealVis on SD 3.0 base is my next goal (if I can train the model locally).

Sorry I forgot to ask earlier,
How long after Stable Diffusion 3.0 comes out can one expect RealVis on 3.0?

Thanks again.

It all depends on the training time based on SD3.0.
For example to train 20 images on SD1.5 it takes me about 16 minutes. To train 20 images on SDXL - 70 minutes.
As soon as I can start training on SD3.0 I can make some prediction.
makaveli-313

Hi! Thank you so much! I think RealVis on SD 3.0 base is my next goal (if I can train the model locally).

Sorry I forgot to ask earlier,
How long after Stable Diffusion 3.0 comes out can one expect RealVis on 3.0?

Thanks again.

It all depends on the training time based on SD3.0.
For example to train 20 images on SD1.5 it takes me about 16 minutes. To train 20 images on SDXL - 70 minutes.
As soon as I can start training on SD3.0 I can make some prediction.

I see. Thanks for the explanation.
It looks like it will be many months regardless. Looking forward to it :)
deleted

i also wonder if us lowly serfs will even be able to use it. 2.x was fine. 3.x is getting close to my systems limits. ( well short of going back to CPU and waiting hours.. ) Not everyone has an extra 1.5k laying around for toys.

Sign up or log in to comment